E-LETTER - Multimedia Communications Technical Committeemmc.committees.comsoc.org/files/2016/04/E-Letter-September11.pdf · Multimedia streaming in wireless multimedia sensor networks

http://www.comsoc.org/~mmc/

MULTIMEDIA COMMUNICAIEEE COMMUNICATIONS SOCIETYhttp://www.comsoc.org/~mmc

E-LETTER

Vol. 6, No. 9, September CONTENTS SPECIAL ISSUE ON EMERGING COMMUNICATIONS TECHNOLOGI ES FOR MOBILE MULTIMEDIA COMMUNICATIONS

Emerging Communications Technologies for Mobile Multimedia Communications ................................

Xianbin Wang, The University of Western Ontario, London, Ontario, CanadaQoS-Driven Video Transmission Schemes in Spatial Multiplexing MIMO Systems................................................................

Qian Liu and Chang Wen Chen, State University of New York at Buffalo, NY, 14260, USA ................................

Optimizing Resource Allocation for Multimedia Cloud ComputingXiaoming Nan, Yifeng He and Ling Guan (IEEE Fellow), Ryerson University, Canada ................................

Minimising Cell Transmit Power: Towards SelfOFDMA Femtocells ................................

David López-Pérez and Xiaoli Chu, King’s College London, London, UKAthanasios V. Vasilakos, Holger Claussen, Alcatel

Cross Layer WiMAX Scheduler for Scalable Video CodingShehzad Amir and YoungYeongtong-gu, Suwon, 443Kyungtae Kim, NEC Laboratories America, Princeton, NJ 08540, USA

Multimedia Streaming in Mobile Sensor Networks with Environmental Energy Harvesting

Lei Shu, Takahiro Hara, Shojiro Nishio, Department of Multimedia Engineering, Osaka University, Japan

DSP Techniques for LowHai Lin, and Katsumi Yamashita, Graduate School of Engineering, Osaka Prefecture University ................................

Network Planning for DTV Broadcasting to Mobile and Portable DevicesJosé Antonio Arenas, TelefónicUnai Gil and Pablo Angueira, Spain ................................

TECHNOLOGY ADVANCES

.org/~mmc/ 1/60 Vol. 6, No. 9, September

MULTIMEDIA COMMUNICA TIONS TECHNICAL COMMITTEE IEEE COMMUNICATIONS SOCIETY http://www.comsoc.org/~mmc

LETTER September 2011

SPECIAL ISSUE ON EMERGING COMMUNICATIONS TECHNOLOGI ES FOR MOBILE MULTIMEDIA COMMUNICATIONS ........................................................

Emerging Communications Technologies for Mobile Multimedia ................................................................................................

Xianbin Wang, The University of Western Ontario, London, Ontario, CanadaTransmission Schemes in Spatial Multiplexing MIMO Systems................................................................................................

Qian Liu and Chang Wen Chen, State University of New York at Buffalo, ................................................................................................

Optimizing Resource Allocation for Multimedia Cloud Computing .......................Xiaoming Nan, Yifeng He and Ling Guan (IEEE Fellow), Ryerson University,

........................................................................................................................Minimising Cell Transmit Power: Towards Self-organized Resource Allocation in

................................................................................................Pérez and Xiaoli Chu, King’s College London, London, UK

Athanasios V. Vasilakos, National Technical University of Athens Athens, GreeceHolger Claussen, Alcatel-Lucent Bell-Labs Dublin, Ireland ................................

Cross Layer WiMAX Scheduler for Scalable Video Coding ................................Shehzad Amir and Young-June Choi, AJOU University, Woncheon-dong,

gu, Suwon, 443-749, Korea ................................................................Kyungtae Kim, NEC Laboratories America, Princeton, NJ 08540, USA .................

Multimedia Streaming in Mobile Sensor Networks with Duty-cycle and Environmental Energy Harvesting ................................................................

Lei Shu, Takahiro Hara, Shojiro Nishio, Department of Multimedia Engineering, a University, Japan ...........................................................................................

DSP Techniques for Low-cost Mobile Terminals ....................................................Hai Lin, and Katsumi Yamashita, Graduate School of Engineering, Osaka

................................................................................................Network Planning for DTV Broadcasting to Mobile and Portable Devices

José Antonio Arenas, Telefónica Móviles 28013, Madrid, Spain .............................Unai Gil and Pablo Angueira, Bilbao Engineering College, University of the Basque,

.........................................................................................................................TECHNOLOGY ADVANCES ......................................................................................

Vol. 6, No. 9, September 2011

TIONS TECHNICAL COMMITTEE

SPECIAL ISSUE ON EMERGING COMMUNICATIONS TECHNOLOGI ES FOR ........................ 3

........................................... 3 Xianbin Wang, The University of Western Ontario, London, Ontario, Canada ........ 3

Transmission Schemes in Spatial Multiplexing MIMO Systems......................................... 5

Qian Liu and Chang Wen Chen, State University of New York at Buffalo, Buffalo, .......................................... 5

....................... 9

Xiaoming Nan, Yifeng He and Ling Guan (IEEE Fellow), Ryerson University, ........................ 9

organized Resource Allocation in .................................... 13

Pérez and Xiaoli Chu, King’s College London, London, UK ............. 13 National Technical University of Athens Athens, Greece 13

.................................... 13 .................................. 17

dong, .................................... 17

................. 17

cycle and ........................................... 20

Lei Shu, Takahiro Hara, Shojiro Nishio, Department of Multimedia Engineering, ........................... 20

.................... 25

Hai Lin, and Katsumi Yamashita, Graduate School of Engineering, Osaka ................................ 25

Network Planning for DTV Broadcasting to Mobile and Portable Devices .......... 28 ............................. 28

Bilbao Engineering College, University of the Basque, ......................... 28

...................... 33

http://www.comsoc.org/~mmc/ 2/60 Vol. 6, No. 9, September 2011

ADVANCED VIDEO COMMUNICATIONS ......................................................... 33 Guest Editor, Honggang Wang, University of Massachusetts Dartmouth, USA ...... 33

Multi-Channel Peer-to-Peer Multimedia Streaming ............................................... 35 Miao Wang, Zhipeng Ouyang, Lisong Xu and Byrav Ramamurthy, Department of Computer Science & Engineering, University of Nebraska, Lincoln, NE, 68588, USA................................................................................................................................... 35

Mobile Multimedia Technology for Internet of Things........................................... 38

Roy Lai, Confederal Networks Inc ............................................................................ 38 Min Chen, School of Computer Science and Engineering, Seoul National University, Korea......................................................................................................................... 38

Downlink VBR Video Scheduling in Cellular Networks with Orthogonal Channels....................................................................................................................... 40

Shiwen Mao, Yingsong Huang, and Yihan Li, Auburn University, Auburn, AL, USA................................................................................................................................... 40

Motion-Based Region of Interest (ROI) Transmission with Variable Frame Rate over Wireless Multimedia Sensor Networks ............................................................ 43

Jian Fang and Alvin Lim, Auburn University, Auburn, AL 36849, USA .................. 43 Qing Yang, Montana State University, Bozeman, MT 59717, USA .......................... 43

Real-time Video Surveillance for Large Scenes ....................................................... 48 Hanyu Liu, Chong Tang, and Shaoen Wu, School of Computing, University of Southern Mississippi, Hattiesburg, MS 39401, USA ................................................ 48

Cross Layer QoS Provisioning for Cooperative Video Transmission ................... 51 Kun Hua, Department of Electrical and Computer Engineering, Lawrence Tech University, Southfield, MI ......................................................................................... 51 Yin Wang, Department of Mathematics and Computer Science, Lawrence Technological University .......................................................................................... 51

Wei Wang, South Dakota State University ............................................................... 51 Shaowen Wu, University of Southern Mississippi..................................................... 51

Cross-layer Based P2P Scheduling in Wireless Multimedia Networks ................. 55 Haiyan Luo, Song Ci, Dalei Wu, Wei An, Department of Computer and Electronics Engineering, University of Nebraska-Lincoln, Lincoln, NE 68588, USA ................ 55

NEWS ............................................................................................................................... 59 E-LELLER EDITORIAL BOARD ............................................................................... 60 MMTC OFFICERS ........................................................................................................ 60

IEEE COMSOC MMTC E-Letter

http://www.comsoc.org/~mmc/ 3/60 Vol.6, No.9, September 2011

SPECIAL ISSUE ON EMERGING COMMUNICATIONS TECHNOLOGI ES FOR MOBILE MULTIMEDIA COMMUNICATIONS

Emerging Communications Technologies for Mobile Multimedia Communications

Xianbin Wang, The University of Western Ontario, London, Ontario, Canada {xianbin.wang}@uwo.ca

Recent advances in mobile broadband communications have witnessed a dramatic technological evolution of multimedia content dissemination. The objective of the special issue of the E-Letter on Emerging Communications Technology for Mobile Multimedia is to identify and promote necessary enabling wireless technologies for efficient wireless multimedia communications and distributions, particularly emerging technologies to maintain QoS under dynamic mobile channel conditions. In the first paper, “QoS-Driven Video Transmission Schemes in Spatial Multiplexing MIMO Systems”, several QoS-driven video transmission schemes for spatial multiplexing SU-MIMO and MU-MIMO systems have been presented. It has been shown that high performance multimedia experience can only be achieved if the transmission schemes have sufficiently considered the characteristics of video sources. In the second paper, “Optimizing Resource Allocation for Multimedia Cloud Computing”, queuing models are employed to study the resource allocation in multimedia cloud. Specifically, the resource allocation problems in multimedia cloud are formulated into the service response time minimization problem or the resource cost minimization problem. The third paper, “Minimising Cell Transmit Power: Towards Self-organized Resource Allocation in OFDMA Femtocells”, considers the problem of radio resource management in distributed cellular network for multimedia distribution. In this E-Letter, a new resource allocation model taking realistic resource allocation constraints into account and the corresponding performance evaluation are presented. In the fourth paper of the special issue, “Cross Layer WiMAX Scheduler for Scalable Video Coding”, Scalable Video Coding (SVC) is studied to support video services with versatile types of devices with different QoS requirements. This

work exploits the scalability of SVC for wireless systems such as WiMAX systems by the cross-layer approach. The next paper, “Multimedia Streaming in Mobile Sensor Networks with Duty-cycle and Environmental Energy Harvesting”, focuses on Multimedia streaming in wireless multimedia sensor networks (WMSNs). In this E-letter, the main problem of how to successfully and efficiently send the multimedia data produced by the source node to the base station while the multimedia source node moves in the network is addressed. In the sixth paper of the special issue, “DSP Techniques for Low-cost Mobile Terminals,” implementation related considerations for mobile terminals supporting multimedia communications are considered. In this E-Letter, existing DSP-based compensation techniques for RF/analog impairments in the direct conversion receiver (DCR)s have been reviewed. Studies in this letter show the digital compensation is a promising solution to low-cost mobile terminals. The last paper of the special issue, “Network Planning for DTV Broadcasting to Mobile and Portable Devices”, summarizes the results obtained for different network deployment strategies for delivering digital terrestrial television to mobile and portable devices. Results gathered in this paper, lead to a quantification of different gains or aspects which should be taken into account for network planning depending on the reception scenarios.

Xianbin Wang is an Associate Professor at The University of Western Ontario and a Canada Research Chair in Wireless Communications. His primary research area is



wireless communications and related applications, including adaptive and cognitive communications systems, wireless infrastructure based position location, cross-layer design, and wireless resource allocation. Dr. Wang has over 150 publications on various communication system design issues, including around 100 IEEE Journal and Conference papers, 15 granted and pending patents, and several standard contributions. Dr. Wang is an IEEE Distinguished Lecturer and a Senior Member of IEEE. He was the recipient of

the IEEE Scott Helt Memorial Award for the Best Paper published in IEEE Transactions on Broadcasting in 2004. He currently serves as an Associate Editor for IEEE Wireless Communications Letters, IEEE Transactions on Vehicular Technology and IEEE Transactions on Broadcasting. He was also an editor for IEEE Transactions on Wireless Communications between 2007 and 2011. Dr. Wang was involved in a number of IEEE conferences including GLOBECOM, ICC, WCNC, VTC, and ICME, on different roles such as symposium chair, track chair, TPC and session chair.



QoS-Driven Video Transmission Schemes in Spatial Multiplexing MIMO Systems Qian Liu and Chang Wen Chen, State University of New York at Buffalo, Buffalo, NY,

14260, USA {qianliu, chencw}@buffalo.edu

1. Introduction As the demand for mobile multimedia consumptions is growing exponentially, offering high-quality services to multimedia consumers requires significant increase in transmission rate. Therefore, multiple-input multiple-output (MIMO) technique has become an excellent choice for modern broadband multimedia communication due to its great potential in improving the data throughput. However, there are several significant technical challenges in adopting the MIMO systems for video transmission. There are two major types of MIMO techniques: single-user MIMO (SU-MIMO) and multi-user MIMO (MU-MIMO). In both cases, spatial multiplexing is considered as an effective strategy for data throughput enhancement. In SU-MIMO, spatial multiplexing [1] has adopted channel decomposition technique to decompose MIMO channel into several sub-channels, so that multi-stream independent data can be simultaneously transmitted. In MU-MIMO, diagonal block decomposition [2] has been explored to transmit simultaneously independent data requested by each served user. However, conventional spatial multiplexing schemes are designed for regular data users rather than multimedia consumers. Therefore, without considering the characteristics of multimedia sources, conventional MIMO systems are unable to meet the quality-of-service (QoS) requirement of multimedia consumers. In this paper, we will highlight recently developed schemes for reliable video transmission over MIMO systems. As is well-known, multimedia has two key characteristics: 1) for images or videos encoded by standard codecs, different parts of the bit-stream shall have unequal importance (a.k.a. different priorities); 2) delay sensitivity. In Section 2, we will introduce prioritized spatial multiplexing schemes for SU-MIMO systems with application to H.264 scalable video coding (SVC) video. Basically, these schemes focus on the unequal importance of the video sources. In section 3, a novel user scheduling approach is introduced with consideration of delay sensitivity of video sources. Finally, section 4 concludes the paper with a summary.

2. Prioritized Spatial Multiplexing for SU-MIMO systems with H.264 SVC Video Conventional spatial multiplexing techniques [1], [3]-[4] assume that all the data bits are equally important for the receiver. However, for SVC video, different bit-streams have different priorities. It is based on this observation that the prioritized spatial multiplexing schemes have been developed in order to improve the performance of wireless video transmission. Here, we will first demonstrate the characteristics of video bit-streams encoded by SVC in the next section. 2.1 SVC SVC is able to generate multi-layered video bit streams and has the potential to be properly mapped to spatial multiplexing parallel MIMO channels. Hence, it has been considered as the preferred option for video transmission in the next generation multimedia communications over MIMO systems. H.264 SVC compressed video consists of a base layer and several enhancement layers. The base layer contains all the necessary information to reconstruct a video sequence with minimum acceptable quality while the enhancement layers, if received, will improve the received video quality after combining with the data from the base layer. It is obvious that the importance of the each video layer is very different from each other and the base layer is the most important one as all enhancement layers shall rely on the base layer for successful decoding. From such layered data structure with different priorities, we can conclude that the decomposed MIMO sub-channels should also be prioritized to match a highest quality sub-channel to base layer in order to guarantee the best possible transmission of SVC base layer video data. 2.2 Prioritized Spatial Multiplexing Schemes in Close-loop SU-MIMO Systems with H.264 SVC Video In the case of close-loop MIMO, we assume that channel state information (CSI) is available to both transmitter and receiver. For close-loop SU-MIMO systems, several new schemes specifically designed for multimedia transmission were



proposed recently. To achieve the goal of prioritized delivery of layered video bit streams, elegant singular value decomposition (SVD)-based spatial multiplexing schemes for SVC-based video transmission over MIMO wireless systems were proposed in [5]-[6] . In particular, the authors developed an adaptive channel selection scheme in [5] to assign the best sub-channel obtained by SVD-based channelization to the base layer. However, even though the sub-channel with highest signal-to-noise-ratio (SNR) is assigned to the based layer in the transmitter after channelization, the received SNR of the base layer may not be able to reach the desired SNR in order to guarantee the required minimum acceptable video quality. Therefore, the inability of such conventional water-filling power allocation scheme to incorporate desired SNR information into its design has limited its potential to achieve optimal quality of the reconstructed video con-strained by the desired SNR for prioritized video layers. In [6], a novel alternative design was developed to maximize the system data throughput. In this scheme, a novel power allocation and adaptive QAM modulation method was introduced. The power allocation scheme presented in [6] incorporates the target bit-error-rate (BER) of each layer to the active adjustment the transmit power to achieve the desired SNR. However, this power allocation scheme was too complex for most RF hardware implementation. Furthermore, in fast fading channel, the delay caused by the iterative power reallocation using CSI feedback will lead to performance degradation for such MIMO systems. In [7], we propose a novel pre-coding scheme shown as Fig. 1. In this scheme, to match the characteristics of layered SVC video bitstreams so as to achieve optimal quality of the reconstructed video, different video layers are proactively assigned with different priorities based on their BER and SNR requirements in the receiver side, as well as the qualities of sub-channels. The power is

allocated to each SVD-based sub-channel corresponding to the priorities by augmented water-filling method. In particular, the base layer that is assigned the highest priority can utilize more resource via pre-coding to establish a sub-channel with desired SNR. Comparing with schemes in [5]-[6], the proposed pre-coder has low computational complexity and is suitable for rapid hardware implementation. Moreover, simulation results in [7] shows that the scheme can efficiently enhance the quality of SVC video streaming over SU-MIMO systems comparing to the conventional spatial multiplexing SU-MIMO systems. 2.3 Prioritized Spatial Multiplexing Scheme for fast moving terminals in open-loop SU-MIMO Systems with H.264 SVC Video In the previous section, we have discussed some QoS guaranteed video transmission schemes in close-loop SU-MIMO systems. Generally, close-loop designs are applied for static or slowly moving terminals. As for fast moving terminals, open-loop scheme (i.e. CSI is only available to the receiver) will be a better choice since it is impossible for the transmitter (i.e. base station) to obtain real-time CSI from the terminal with high moving speed. Currently, researchers and practitioners have mainly focused their designs on coping with static or slowly evolving traffic demands. Fast moving traffic conditions have not yet been properly taken into account in the design of broadband multimedia applications. In [8], we consider the challenging task of providing high quality multimedia services over SU-MIMO systems with fast moving terminals. With fast moving terminals, many existing approaches to MIMO channel estimation and equalization cannot be applied as most of them rely on the help of periodic pilot signals. When the terminals are travelling in high speed, the undesired Doppler effects prevent the existing approaches from efficient use of pilot signals for channel estimation and equalization. In [8], we develop a blind constant modulus algorithm (CMA)-based channel equalization (CMACE) scheme (shown as Fig. 2.) to track the fast time-varying channel without the requirement of pilot signals. We show that the transmitted signal in each sub-channel can be modulated with different symbol constellation, and delivered at different rate according to their importance. Simulation results show that the proposed scheme is effective for channel estimation/equalization for prioritized spatial multiplexing SU-MIMO system with fast moving terminals.

Fig. 1. SVC-based video transmission scheme in close-loop SU-MIMO.



M

Adaptive

DemodulationM

Base Layer

Enhance Layer 1

Enhance Layer Video

M

1tN−

Adaptive

modulationM

Base Layer

Enhance Layer 1

Enhance Layer Video

Sequence

M

1tN −

MappingMM

Channel

Equalization

Fig.2. Prioritized spatial multiplexing scheme for fast moving terminals in open-loop SU-MIMO systems.

3. QoS Guaranteed User Scheduling for MU- MIMO systems In the previous section, we demonstrate video transmission schemes in SU-MIMO systems, and in this section, we will present novel QoS guaranteed schemes in MU-MIMO systems. For MU-MIMO, it is not a simple transition of SU-MIMO from a single user to multiple users. Actually, MU-MIMO is substantially different from SU-MIMO and outperforms SU-MIMO in several aspects. One major difference is that MU-MIMO system needs a well designed scheduler to effectively coordinate multiple users within such systems. In downlink MU-MIMO systems, linear zero-forcing (ZF) pre-coding has been widely utilized in order to achieve the spatial multiplexing gain. By ZF pre-coding, the multiple antenna transmitter can null multi-user interference signals in wireless communications. If we assume each UE is equipped with only one receive antenna, then the maximum number of users can be served each time is equal to the number of transmit antennas in the BS. However, a downlink MU-MIMO system usually has much more users to be severed than the number of transmitting antennas [9]. Therefore, the BS has to dynamically select a best subset of users to serve at each time slot, demanding an appropriate user scheduling in such a MU-MIMO system. One crucial problem for user scheduling in MU-MIMO is how to balance quality of service (QoS) and fairness. This problem becomes even more challenging when we consider multimedia con-sumers together with regular data users. For example, since multimedia users are bandwidth hungry, the fairness design for the scheduling scheme should pay special attention to bandwidth assignment when the multimedia users are served. Another crucial issue needs to be considered is the delay constraints of individual video packets. Video packets are with deadlines (i.e. the time determines whether or not a video packet is still valid). A time-out video packet is equivalent to a loss packet as it is unavailable to video

reconstruction when needed. Therefore, a deadline-aware scheduling is necessary to guarantee QoS. In [10], we propose a fairness and QoS guaranteed user scheduling scheme for MU-MIMO BC channels. The contribution of this scheme lies in providing fair services to regular data users while guaranteeing the QoS of multimedia consumers. In this framework, we first estimate the transmission deadline of each video packet. The estimation of transmission deadline serves dual purpose: it is used to guarantee the transmission of any given video packet before it expires, and it also allows other unscheduled regular data users to access the channel when there is no pressing need to transmit any video packet for the video consumer. 4. Conclusions In this paper, we have presented several QoS-driven video transmission schemes for spatial multiplexing SU-MIMO and MU-MIMO systems. Simulation results show that all these schemes have been able to efficiently enhance the performance of video transmission over MIMO systems. Most importantly, it has been shown that high performance multimedia experience can only be achieved if the transmission schemes have sufficiently considered the characteristics of video sources. References [1] P. W. Wolniansky, G. J. Foschini, G. D. Golden, and R. A. Valenzuela, “V-BLAST: An architecture for realizing very high data rates over the rich-scattering wireless channel,” in Proc. ISSSE-98, Pisa, Italy, Sept. 29, 1998. [2] D. Gesbert, M. Kountouris, R. W. Heath, Jr., C.B. Chae, and T. Salzer, “Shifting the MIMO Paradigm: From Single User to Multiuser Communications,” IEEE Signal Processing Magazine, vol. 24, no. 5, pp. 36-46, Oct. 2007. [3] Y. Jiang, J. Li, and W. W. Hager, “The geometric mean decomposition,” Linear Algebra and Its Applications, vol. 396, pp. 373-384, Feb. 2005. [4] Y. Jiang, J. Li, and W. W. Hager, “Uniform channel decomposition for MIMO communications,” IEEE Trans. on Signal Processing, vol. 53, pp. 585-595, Nov. 2005. [5] D. Song and C. W. Chen, “Scalable H.264/AVC video transmission over MIMO wireless systems with adaptive channel selection based on partial channel information,” IEEE Trans. on Circuits and Systems for Video Technology, vol. 17, pp. 1218-1226, Sept. 2007. [6] D. Song and C. W. Chen, “Maximum-throughput delivery of SVC-based video over MIMO systems with time-varying channel capacity,” Journal of Visual Communication and Image Representation, vol. 19, pp. 520-528, Dec. 2008. [7] Q. Liu, S. Liu and C. W. Chen, “A novel prioritized spatial multiplexing for MIMO wireless system with



application to H.264 SVC video,” Proc. 2009 International Conf. Multimedia and Expo, July 2010, Singapore. [8] Q. Liu and C. W. Chen, “Blind Channel Equalization for Fast Moving Terminals in Prioritized Spatial Multiplexing MIMO Systems,” GLOBECOM 2010, pp.1-5, Dec. 2010. [9] M.A. Khan, R. Vesilo, I.B. Collings, and L.M. Davis, “Alpha-rule scheduling for MIMO broadcast wireless channels with linear receivers,” in Proc. of the 10th Australian Communications Theory Workshop, pp. 110-115, Mar. 2009. [10] Q. Liu and C. W. Chen, “Fairness and QoS guaranteed user scheduling for multi-user MIMO broadcasting channel,” accepted by ICIP 2011.

Qian Liu received B.S. and M.S. degrees from Dalian University of Technology in 2006 and 2009. She is now pursuing the Ph.D. degree in the Department of Computer Science and Engineering, State University of New York at Buffalo. Her current research

interests include mulitmedia transmission over MIMO systems and IEEE 802.11 wireless communication systems.

Chang Wen Chen (F’04) received the B.S. degree from the University of Science and Technology of China in 1983, the M.S.E.E. degree from the University of Southern California, Los Angeles, in 1986, and the Ph.D. degree from the University of Illinois

at Urbana-Champaign in 1992. He is a Professor of Computer Science and Engineering at the State University of New York at Buffalo. Previously, he has been Allen Henry Endow Chair Professor at the Florida Institute of Technology from 2003 to 2007, on the faculty at the University of Rochester from 1992 to 1996, on the faculty at the University of Missouri-Columbia from 1996 to 2003. Prof. Chen served as the Editor-in-Chief for IEEE Trans. Circuits and Systems for Video Technology for two terms from January 2006 to December 2009. He has been an Editor for numerous IEEE Transactions and Journals, including Proceedings of IEEE, IEEE Journal of Selected Areas in Communications, IEEE Trans. Multimedia, and IEEE Multimedia Magazine. He has also served as Conference Chair for several major IEEE, ACM and SPIE conferences related to mobile wireless video communications and signal processing. He has received numerous research, service, and best paper awards, including the 2003 Sigma Xi Excellence in Graduate Research Mentoring Award and 2009 Alexander von Humboldt Research Award. He is also an SPIE Fellow.



Optimizing Resource Allocation for Multimedia Cloud Computing Xiaoming Nan, Yifeng He and Ling Guan (IEEE Fellow), Ryerson University, Canada

[email protected], [email protected], [email protected]

1. Introduction Cloud computing is an emerging computing paradigm which can provide computation, communications, and storage resources as utilities in a scalable and virtualized manner. Multimedia services are growing more and more popular in the recent years. Many multimedia applications, such as image/video retrieval, 3D video rendering, require intensive computation and/or intensive storage, which may become burdens to the clients, especially the mobile clients who have limited computation and storage. Therefore, there are strong needs to introduce cloud computing into the multimedia applications. In multimedia cloud computing paradigm [1], cloud service providers deploy cloud resources as utilities to process multimedia requests and then deliver computing results to users. With multimedia cloud services, users do not need to own costly computing devices. Instead, they can request the powerful cloud servers to do multimedia processing by paying for the cloud resources that they occupy. There are two major concerns for multimedia cloud service providers: the service response time and the resource cost. The service response time in the data center is defined as the period from the time when the service request arrives at the data center to the time when the service result departs from the data center. Service response time is a significant Quality of Experience (QoE) factor to measure the performance of multimedia cloud service. A lower service response time will lead to a higher QoE. Thus, it is important for cloud service providers to meet customer’s requirement on service response time. The second concern is the cost of the allocated cloud resources. The cloud service can generally be divided into three consecutive phases: schedule, computation and transmission. Appropriate allocation of cloud resources is expected to greatly improve the cloud performance. However, it is challenging for cloud providers to optimally allocate resources among the three phases to minimize the service response time or minimize the resource cost. This letter focuses on performance analysis and optimization for multimedia cloud service providers. Specifically, we study the resource allocation problems in multimedia cloud by using queuing models. The remainder of the letter is

organized as follows. Section 2 discusses the related work. Section 3 presents the queuing models and the proposed methodologies. The conclusions and future directions are provided in Section 4. 2. Related Work The resource allocation strategies in cloud computing are studied in [2-6]. Lin et al. [2] develop a self-organizing model to manage cloud resources in the absence of centralized management control. Shi et al. [3] focus on the maximization of the steady-state throughput at data center by deploying cloud resources for the independent equal-sized tasks. Teng et al. present a resource pricing and equilibrium allocation policy based on the consideration of cloud users’ competition for limited resources [4]. Another research direction of resource management is to model the relationship between the service performance and the allocated resources. In [5], a video streaming workload analysis and prediction technique is proposed. With this method, multimedia workload can be examined from the user access perspective. The authors in [6] classify the application requests into disk- and CPU-bound ones and then construct scaling functions of system capacity accordingly to predict the required resources. 3. Methodologies The data center architecture generally consists of a master server, a number of computing servers and a transmission server. The queuing model of the data center is shown in Fig. 1. The master server maintains a schedule queue to receive all coming requests from customers, and then schedules the requests to the computing servers. The requests assigned to a computing server will be first stored in the corresponding computation queue, and then processed by the computing server. For each request, a service result will be generated by the computing server. All service results will be stored in a transmission queue before they are transmitted back to the customers. The resource cost in data center is calculated based on the utilized resources and the occupied time. The utilized resources include the resources at the master server, the computing servers, and the transmission server. The occupied time is defined



as the duration that the resources are occupied by the requests.

Fig. 1. Queuing model of the data center We will use the proposed queuing model to study the resource allocation problems in single-class service case and multiple-class service case, respectively. In single-class service case, there is only one kind of application service provided in the data center. Thus, all cloud customers request for the same kind of service with the same processing procedure. In such a case, the service response time depends on the allocated resources at the master server, the computing servers and the transmission server. There is a trade-off between the mean service response time and the utilized resource in the cloud. The resource allocation problem in the cloud can be formulated into the service response time minimization problem, which is stated as: to minimize the total service response time in the data center by optimizing the resources at the master server, the computing servers, and the transmission server, subject to the resource cost constraint. The optimal analytical solution can be obtained using the Lagrange multiplier method [8]. Alternatively, the resource allocation problem in the cloud can also be formulated into the resource cost minimization problem, which is stated as: to minimize the total resource cost in the data center by optimizing the resources at the master server, the computing servers, and the transmission server, subject to the requirement on the service response time. The resource cost minimization problem can also be solved analytically by using the Lagrange multiplier method [8]. In multiple-class service case, multiple kinds of application services are provided by the data centre. Each kind of service has a different processing procedure at the computing server, a different transmission time, and a different requirement on the service response time. In such a case, the mean service response time depends not only on the allocated resources in the three phases, but also on the arrival rate, the computation time, and the transmission time for each kind of service. The

service response time minimization problem in multiple-class service case is stated as: to minimize the total service response time in data center by optimizing the resources at the master server, the computing servers, and the transmission server, subject to the total resource cost constraint. The resource cost minimization problem in multiple-class service case is stated as: to minimize the total resource cost by optimizing the resources at the master server, the computing servers, and the transmission server, subject to the constraint on the service response time for each class of service. Both the service response time minimization problem and the resource cost minimization problem are convex optimization problems [7], which can be solved efficiently using the primal-dual interior-point methods [8].

Fig. 2. Comparison of the mean service response

time in multiple-class service case Fig. 2 shows the comparison of service response time in multiple-class service case between the proposed optimal allocation scheme, in which the allocated resources for schedule, computation and transmission are determined by solving the service response time minimization problem, and the equal allocation scheme, in which the resources for schedule, computation and transmission are allocated equally. From Fig. 2, we can see that the proposed optimal allocation scheme takes much less service response time than the equal allocation scheme under the same resource cost constraint. 4. Conclusions and Future Directions In this letter, we employ the queuing models to study the resource allocation in multimedia cloud. Specifically, we formulate the resource allocation problems in multimedia cloud into the service response time minimization problem or the resource cost minimization problem, for single-class service case and multiple-class service case, respectively. The evaluation result demonstrates



the superiority of the proposed optimal allocation scheme. There are numerous future research directions for resource allocation in multimedia cloud. 1) It is an open research problem on the integration of the resource allocation with the characteristics of specific multimedia application. 2) It is a challenging problem on dynamic resource allocation for multimedia cloud, considering the time-varying workload, the available cloud resources, and dynamic network conditions. 3) It is promising to apply the market-based resource allocation approaches to tackle the resource allocation problems for multimedia cloud computing. 4) It is a challenging task to maximize or minimize an end-to-end QoE metric by jointly optimizing the resources in the cloud, at the client, and along the transmission path between the cloud and the client. References [1] W. Zhu, C. Luo, J. Wang, and S. Li, “Multimedia cloud computing: Directions and applications,” Special Issue on Distributed Image Processing and Communications, IEEE Signal Processing Magazine, May 2011. [2] W. Lin and D. Qi, “Research on Resource Self-Organizing Model for Cloud Computing,” in Proc. IEEE International Conference on Internet Technology and Applications, 2010. [3] H. Shi and Z. Zhan, “An optimal infrastructure design method of cloud computing services from the BDIM perspective,” in Proc. IEEE Computational Intelligence and Industrial Applications, vol. 1, pp. 393–396, 2009. [4] F. Teng and F. Magoules, “Resource Pricing and Equilibrium Allocation Policy in Cloud Computing,” in Proc. IEEE International Conference on Computer and Information Technology, pp. 195–202, 2010. [5] H. Yu, D. Zheng, B. Zhao, and W. Zheng, “Understanding user behavior in large-scale video-on-demand systems,” ACM SIGOPS Operating Systems Review, vol. 40, no. 4, pp. 333–344, 2006. [6] L. Cherkasova and L. Staley, “Building a performance model of streaming media applications in utility data center environment,” in Proc. of ACM/IEEE Conference on Cluster Computing and the Grid (CCGrid), 2003. [7] X. Nan, Y. He and L. Guan, “Optimal resource allocation for multimedia cloud based on queuing model”, in Proc. of IEEE International Workshop on Multimedia Signal Processing (MMSP), Oct. 2011.

[8] S. Boyd and L. Vandenberghe, Convex Optimization, Cambridge, U.K., Cambridge University Press, 2004.

Xiaoming Nan received his M.S. degree in

Telecommunication Engineering from Beijing University of Posts &

Telecommunications, China, in 2010. He is currently a Ph.D. candidate at Ryerson University, Toronto, Canada. His research

interests include multimedia cloud computing and content-based video retrieval.

Yifeng He (M’09) received his Ph.D. degree in Electrical Engineering from Ryerson University, Canada in 2008. He is currently an Assistant Professor in the department of Electrical and Computer Engineering at Ryerson University, Canada. His research

interests include network resource optimization, multimedia cloud computing, Peer-to-Peer (P2P) video streaming, and wireless sensor networks. He is the recipient of 2008 Canada Governor General’s Gold Medal.

Ling Guan (M’90-SM’96-F’08) is a Tier I Canada Research Chair in Multimedia and Computer Technology, and a Professor of Electrical and Computer Engineering at Ryerson University, Toronto,

Canada. He received his Bachelor’s Degree from Tianjin University, China, Master’s Degree from University of Waterloo and Ph.D. Degree from University of British Columbia. Dr. Guan has been working on image, video and multimedia signal processing and published extensively in the field. He chaired the 2006 IEEE International Conference on Multimedia and Expo in Toronto, co-chaired the 2008 ACM International Conference



on Image and Video Retrieval in Niagara Falls, and served as the Founding General Chair of IEEE Pacific-Rim Conference on Multimedia in 2000. Dr. Guan is a Fellow of the IEEE, a Fellow of the Engineering Institute of Canada and a Fellow (Elected) of the Canadian Academy of

Engineering. He is an IEEE Circuits and System Society Distinguished Lecturer (2010-2011) and a recipient of the 2005 IEEE Transactions on Circuits and Systems for Video Technology Best Paper Award



Minimising Cell Transmit Power: Towards Self-organized Resource Allocation in OFDMA Femtocells

David López-Pérez and Xiaoli Chu, King’s College London, London, UK Athanasios V. Vasilakos, National Technical University of Athens Athens, Greece

Holger Claussen, Alcatel-Lucent Bell-Labs Dublin, Ireland {David.Lopez, Xiaoli.Chu}@kcl.ac.uk

Abstract With the introduction of femtocells, cellular networks are moving from the conventional centralised architecture to a distributed one, where each network cell should make its own radio resource management decisions, while providing inter-cell interference mitigation. However, realising this distributed cellular network architecture is not a trivial task. In this paper, we first introduce a simple self-organisation rule under which a distributed cellular network is able to converge into an efficient resource allocation pattern, then propose a novel resource allocation model taking realistic resource allocation constraints into account, and finally evaluate the performance of the proposed self-organisation rule and resource allocation model using system-level simulations. 1. BACKGROUND AND MOTIVATION Femtocells, which are low-cost, low-power cellular base stations (BSs) deployed by end-users in homes and offices, have been widely considered as a cost-effective solution to enhance indoor coverage and spectral efficiency in cellular networks. By reducing the distance between BSs and end-users and alleviating the traffic burden on macrocells, femtocells can potentially improve spatial reuse, allow higher user data-rates and provide energy savings. However, femtocell roll-outs are facing their own technical challenges. For instance, since the user-provided backhaul connection between an femto BS and the operator's core network usually has limited capacity, it is difficult to manage inter-cell interference using classic centralised network planning and optimization tools. This has led to a new inter-cell interference management problem that must be solved in a decentralised and distributed manner. However, to achieve good performance across the network with each cell taking its own radio resource management decisions is an intricate problem [1].

2. RADIO RESOURCE MANAGEMENT In orthogonal frequency division multiple access (OFDMA)-based networks, e.g., Long Term Evolution (LTE) and Wireless Interoperability for Microwave Access (WiMAX), the smallest resource unit that can be assigned is a resource block (RB). An RB is comprised of a set of adjacent sub- carriers in the frequency domain and OFDM symbols in the time domain. The scheduling question to be addressed by each cell is how RBs are to be allocated to users and how much transmit power is to be applied to each RB, so that network capacity is enhanced. This radio resource allocation problem is complex because users may have different quality of service demands and may experience various channel and interference conditions in each RB. Moreover, both LTE and WiMAX standards have scheduling constraints, e.g., when more than one RB are allocated to a user, all these RBs must utilise the same Modulation and Coding Scheme (MCS) [2]. 3. OUR SELF-ORGANISATION RULE For the independent and dynamic optimisation of radio resource assignments at each cell, it is necessary to have an objective that results in a good self-organising behaviour. Accordingly, our proposed cell optimisation rule is de_ned as: each cell assigns MCSs, RBs and transmit power levels to users independently, while minimising the cell total transmit power and meeting its users' throughput demands. The reasons for minimising the cell total transmit power are two: 1. A cell that aims at minimising its own transmit

power mitigates inter-cell interference to neighbouring cells, because less power is allocated to those users with good channel conditions or lower throughput demands. This is straightforward from Shanon-Hartley theorem.

2. A cell that aims at minimising its own transmit power tends to allocate those RBs that are not being used by its neighbouring cells, since less transmit power is required for a less

This paper first appeared in SIGCOMM’11, August 15–19, 2011, Toronto, Ontario, Canada. ACM 978-1-4503-0797-0/11/08.

IEEE COMSOC MMTC E


interfered/faded RB to get a targeted signal to interference plus noise ratio (SINR).

Following this rule, a cell will tend to allocate users that are closer to the BS or have lower datarate requirements (hence requiring lower MSCs as well as transmit powers) to RBs that are used by cell-edge users in neighbouring cells. 4. RESOURCEALLOCATIONMODELThe transmit power that femtocell assign to every subcarrier of RB downlink to user to achieve the SINR threshold

of its assigned MCS is

is the channel gain between femto BS

user , is the inter-cell interference power suffered by user in RB , and power. Femtocell i `know' and user measurement reports (MRs) [2]. Therefore, the joint MCS, RB and allocation problem in femtocellproposed self-organisation rule, can be formulated as the following integer linear problem:

subject to:

where (2g) is a binary decision variable that is equal to 1 if user u uses MCS otherwise, (2f) is a binary decision variable that is equal to 1 if user makes us0 otherwise, constraint (2b) makes sure thatis only assigned to at most one user (2c) and (2d) together guarantee that each user is allocated to at most one MCS, and constraint (2e)

MMTC E -Letter

http://www.comsoc.org/~mmc/ 14/60 Vol.6, No.9, September

targeted signal to (SINR).

rule, a cell will tend to allocate are closer to the BS or have lower data-

(hence requiring lower MSCs as to RBs that are used by

edge users in neighbouring cells.

RESOURCEALLOCATIONMODEL that femtocell should

to every subcarrier of RB assign in the to achieve the SINR threshold

is the channel gain between femto BS and cell interference power

, and is the noise and from its

user measurement reports (MRs) [2].

, the joint MCS, RB and transmit power following our

can be formulated problem:

(2g) is a binary decision variable that is in RB , or 0

binary decision variable makes use of MCS , or

0 otherwise, constraint (2b) makes sure that RB is only assigned to at most one user , constraints

and (2d) together guarantee that each user is at most one MCS, and constraint (2e)

ensures that each user obtains a throughput less than its requirement . Average channel quality in form of instantaneous

averaged over tens of user MRs is

compute , thus avoiding rapid

due to fast-fading, which may result in variations of radio resource assignments One possible approach to solve (2) is to apply integer linear programming (ILP) technreadily available in software packages. Hoalthough ILP solvers are able to optimality, their running times are unpre(exponential in the worst case), which renders theminappropriate for real-time use at femto BSs. Hence, to solve (2) in short time, we propodecomposition approach based on asearch and a minimum cost network Assuming that a MCS has been as

, i.e., is known a priori as part input, (2) can be reduced to a simpler form, i.e., a user-to-RB assignment problem weight

. This assignment problem can be formulated as a minimum cost problem, an optimally solved using a networksimplex algorithm. Thereafter, a heuristic search can be performed over the MCS assignment solution space, where for each MCS assignment to cell users, the optimal RB and power allocation can be found by using our network simSince network simplex runs fast, RB andallocations to users can be updated on a millisecond basis (dealing with fasvariations), while MCS assignmentupdated in a half-second basis (dealingload and mobility).

Figure 1: Simulation scenario. 5. PERFORMANCE AND CONCLUSIONS

September 2011

obtains a throughput no

Average channel quality in form of instantaneous averaged over tens of user MRs is used to

thus avoiding rapid fluctuations of

which may result in rapid assignments.

lve (2) is to apply ear programming (ILP) techniques

ware packages. However, solve (2) up to

, their running times are unpredictable (exponential in the worst case), which renders them

time use at femto BSs. Hence, (2) in short time, we propose a two-level

on a heuristics network flow problem.

has been assigned to user is known a priori as part of the

input, (2) can be reduced to a simpler form, i.e., a RB assignment problem weighted by

assignment problem can be network flow

problem, an optimally solved using a network simplex algorithm. Thereafter, a heuristic search

performed over the MCS assignment for each MCS assignment to

power allocation can found by using our network simplex algorithm.

Since network simplex runs fast, RB and power allocations to users can be updated on a

basis (dealing with fast channel assignments can be

second basis (dealing with traffic

Figure 1: Simulation scenario.

PERFORMANCE AND CONCLUSIONS



The performance of the proposed self- organisation rule has been investigated in an enterprise femtocell scenario (Fig. 1). Simulation results show that compared to existing radio resource management techniques in the literature, i.e., random [3], network listening mode (NLM) [4] and interference minimisation (IM) [5], our approach is able to significantly decrease the number of user outages, increase the average number of simultaneously transmitting users in the network (around 15%), and enhance the average network sum throughput (around 12%) (Tab. 1). This is because our approach (self-organization rule) achieves inter-cell interference coordination without the need of assigning orthogonal RBs among neighbouring cells. Instead, it allows all cells to allocate all RBs to their users in an intelligent manner: A cell that minimises its own transmit power assigns less power to those RBs allocated to users having good channel conditions or with lower data-rate demands. Thus, neighbouring cells will `see' low interference in such RBs and will allocate them to users having bad channel conditions or with large data-rate demands, thereby improving spatial reuse. To show this self-organising feature, Fig. 2 illustrate the transmit power allocated by 3 neighbouring femtocells at a given time in 8 available RBs. We can see that each femtocell tends to allocate higher power levels in RBs in which neighbouring femtocells assign lower power levels and vice versa. In this way, a same RB can be dynamically reused in neighbouring femtocells. Thus, our proposed se-organisation rule and resource allocation model effectively introduce an implicit coordination among independently operating femtocells via distributed dynamic radio resource allocations.

Figure 2: RB allocation.

Table 1: System-Level Simulation Results

REFERENCES [1] V. Chandrasekhar et. al. “Femtocell Networks: A

Survey,” IEEE Comm. Mag., vol. 46, no.9, pp. 59-67, Sep. 2008.

[2] E. Dahlman et. al. “3G Evolution. HSPA and LTE for Mobile Broadband,” Elsevier, 2 edition, Aug. 2008.

[3] V. Chandrasekhar et. al. “Spectrum Allocation in Tiered Cellular Networks,” IEEE Transactions on Communications, vol. 57, no. 10, pp. 3059-3068, Oct. 2009.

[4] J. Ling et. al. “On Resource Allocation in Dense Femto-Deployments,” in Proc. IEEE International Conference on Microwaves, Communications, Antennas and Electronics Systems (COMCAS), Nov. 2009.

[5] J. Zhang and G. de la Roche, “Femtocells: Technologies and Deployment,” John Wiley and Sons, Jan. 2010.

David Lopez-Perez is Research Associate at King's College London, UK, and currently an Invited Researcher at DOCOMO USA labs, Palo Alto, CA. He received his BSc and Master MSc degrees in Tele-communication from Miguel Hernandez University, Spain,

in 2003 and 2006, respectively, and his PhD title from University of Bedfordshire, UK, in 2011. In February 2005, he was with Vodafone Spain working netwrok planning and optimization tasks. In February 2006, he was with Cork Institute of Technology, Ireland. In May 2007, he was awarded with a PhD Marie-Curie fellowship at the Centre for Wireless Network Design (CWiND) at University of Bedfordshire, UK, where he investigated femtocellular networks and interference coordination techniques. He is editor and/or author of several cellular network related books, i.e., "Heterogeneous Cellular Networks: Theory, Simulation and Deployment" Cambridge University Press, 2012, "Femtocells – Technologies and Deployment", Wiley 2010, and "Femtocell Networks: Deployment, PHY Techniques, and Resource Management", Cambridge University Press, 2012. He is/has also been co-chair of several journal SIs and workshops, e.g., 2nd IEEE 2011 GLOBECOM Workshop on Femtocell Networks (FEMnet‘11).



Xiaoli Chu is a Lecturer within the Centre for Tele-communications Research at King’s College London (KCL). She received the PhD degree in Electrical and Electronic Engineering from the Hong Kong University of Science and Technology in 2005. From 2005 to 2006, she was a Research Associate

in the Division of Engineering at KCL. She received the UK EPSRC Cooperative Awards in Science and Engineering for New Academics in 2008, the UK EPSRC First Grant in 2009, and the RCUK UK-China Science Bridges Fellowship in 2011. She is Guest Editor for the ACM/Springer Journal of Mobile Networks & Applications Special Issue on Cooperative Femtocell Networks. She is TPC Co-Chair of the IEEE GLOBECOM 2011 Workshop on Enabling Green Wireless Multimedia Communications, the 4th International Workshop on Femtocells, and the IEEE WCNC 2012 Workshop on Broadband Femtocells: Paving the way to HetNets. She is currently Chair of the Graduates of the Last Decade Affinity Group and Secretary of the Computer Chapter within the IEEE United Kingdom and Republic of Ireland Section.

Athanasios V. Vasilakos is currently Visiting Professor at National Technical University of Athens (NTUA), Athens, Greece. He has authored or co-authored over 200 technical papers in major international journals and conferences. He is author/coauthor of 5 books

and 20 book chapters in the areas of communications. Prof. Vasilakos has served as General Chair, Technical Program Committee Chair for many international conferences. He served or is serving as an Editor or/and Guest Editor for many technical journals, such as the IEEE TRANSACTIONS ON NETWORK AND SERVICES MANAGEMENT, IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART B:CYBERNETICS, IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, ACM TRANSACTIONS ON AUTONOMOUS AND ADAPTIVE SYSTEMS, IEEE JSAC special issues of May 2009, Jan. 2011, March 2011, IEEE Communications Magazine, ACM/Springer Wireless Networks (WINET), ACM/Springer Mobile Networks and Applications (MONET). He is founding Editor-in-Chief of the International Journal of Adaptive and Autonomous Communications Systems (IJAACS, http://www.inderscience.com/ijaacs) and the International Journal of Arts and Technology (IJART). He is General Chair of the Council of Computing and Communications of the European Alliances for Innovation.

Dr. Holger Claussen is head of the Autonomous Networks and Systems Research Department at Bell Labs, Alcatel-Lucent in Ireland. His work in the area small cell networks has been commercialized as Alcatel-Lucent’s BSR-Femto product. Dr. Claussen is author of more than 40 publications and 70 filed patent applications. He is

senior member of the IEEE, member of the IET, and the Alcatel-Lucent Technical Academy.



Cross Layer WiMAX Scheduler for Scalable Video Coding Shehzad Amir and Young-June Choi, AJOU University, Woncheon-dong, Yeongtong-gu,

Suwon, 443-749, Korea ([email protected], [email protected])

Kyungtae Kim, NEC Laboratories America, Princeton, NJ 08540, USA ([email protected])

1. Introduction The advancement on the Internet and exponential use of multimedia data has created demands of efficient multimedia transmission over wireless networks. In wireless networks, reception at an end user is greatly influenced by the wireless bandwidth fluctuation as well as error and delay. To support video services with versatile types of devices having different bandwidth and processing capabilities, SVC (Scalable Video Coding) is a promising solution that provides scalability for quality of service (QoS).

SVC generates multi-layer data with different priorities, catering spatial and temporal quality of video. A SVC stream consists of a base layer and one or a few enhancement layers, where the base layer has the highest priority and enhancement layers are used for temporal and spatial quality of video. The quality and rate parameters of a video stream is adapted in real time for the end user by managing the scalable bit stream at any intermediate node or the encoder at the transmitter by feedback from the receiver or by the channel condition estimated at the transmitter. Fig. 1 shows the basic structure of SVC [1].

Bit-rate scalable media inherently has the prioritization capability and wireless standards such as IEEE 802.16 WiMAX also have QoS profiles which enable to adapt the data rate. In order to prevent serious QoS degradation for video service, we need to support a minimum data rate by adapting the data rate through a SVC encoder and/or MAC/PHY layers of WiMAX systems. In this paper, we propose new cross-layer scheduling algorithms which maintain a minimum data rate regardless of the channel condition. 2. Background 2.1WiMAX System The MAC layer of WiMAX systems provides data rate adaptation capability through various QoS

profiles and modulation/coding schemes. Each service is mapped to an individual 16-bit connection identifier (CID) in the MAC layer. WiMAX systems support scheduling of downlink and uplink based on the bandwidth request from a subscribe station (SS).In this paper, we only consider a downlink service at a base station (BS).

Fig. 1. SVC Layered structure

2.1 Related Work In past few years, cross-layer optimization that couples MAC, PHY and application layers for WiMAX schedulers has gained much interest. In [3] and [5], the authors propose schedulers that integrate QoS parameters of layer2 and layer3. In [4], a cross-layer scheduler is proposed to optimize the queue length from feedback of MAC and PHY layers. In [2], the authors propose a cross-layer scheduler coupling MAC and application layers. When a data rate decreases below a certain level, MAC layer changes the parameters of multimedia encoding. Existing management messages are used, but overhead is significantly increased. 3. Proposed Cross Layer Scheduler 3.1 Objective The objective of this paper is to maintain the minimum data rate requirement of the multimedia service by dropping low priority data (P or B frames) in the base or enhancement layer in addition to adaptive modulation and coding,



thereby confining channel errors to one frame instead of spreading it to all parts of the stream. 3.2 Algorithms

The temporal prediction in SVC is exhibited in

Fig. 2 in which there are two layers. In our algorithm, when a BS discovers that the data rate is below the required level in a downlink, it drops some selective packets from the transmission queue. The selection of packets is based on their priority in the stream and their type. Our algorithm is classified into three cases as follows.

Fig. 2. SVC temporal prediction structure [8]

1) Case1: This case drops some selective packets of the enhancement layer, hence making the data rate lower as shown in Fig. 3.Note that the adaptive modulation and coding scheme is applied, apart from our algorithm, by shifting to a more robust modulation scheme in this case. If the requirement of the minimum data rate is not met in Case 1, the data rate is further reduced as in Case2.

Fig.3. Frame dropping from the enhancement layer

of SVC

2) Case2: This case drops packets such as P and B frames before the next resetting point of the video (I frame) in the base layer. In Fig. 4, I frames are the resetting spot for the video and packets followed by I frames are chosen for the drop.

3) Case3: This case is triggered when the system

still does not meet the minimum rate requirement of the video. Two packets are chosen to be dropped each from base and enhancement layers. This case inherently activates both Case1 and Case2 at the same time and hence decreases the data rate of the

video, as shown in Fig. 5.

Fig.4: Frame dropping from the base layer of SVC before the next I frame.

Fig.5: Frame dropping from the base layer of

SVC before the next I frame and Frame dropping from the enhancement layer

Table 1. Summary of proposed algorithms

Packet drops

Base layer Enhancement layer

Case 1 No Yes

Case 2 Yes No

Case 3 Yes Yes

By dropping the packets, we can decrease the data rate while the MAC layer also shifts to a more robust modulation and coding scheme. Though dropping those packets, the video quality will not be degraded much, as the errors are confined to P frames. If the rate requirement is still not met after Case3, more intermediate packets and even I frames in the enhancement layer may be dropped during a certain period. When the channel condition gets better, the MAC layer stops dropping the packets.

Our algorithm exploits the inherent benefit of the layered structure of SVC by confining the errors to a few dropped frames which are less critical to the video quality. 4. Conclusion

IEEE COMSOC MMTC E


Our work contributes to exploit the scalability of SVC for wireless systems such as WiMAX systems by the cross-layer approach. We propose some algorithms that drop selective packets to guarantee the minimum data rate to cope with fluctuating channel conditions of the wireless link. References [1] Thomas Schierl, Cornelius Hellge, Shpend

Mirta, Karsten Grüneberg and Thomas Wiegand Fraunhofer; ”H.264/AVCScalable Video Coding (SVC) for Real Time Streaming in Wireless IP Networks” for Telecommunication. – Institute Einsteinufer 37, DGermany

[2] Alaeddine Abdallah, Toufik Ahmed, Raouf Boutaba, “Cross Layer Optimization Architecture for Video Streaming in WIMAX Networks” in the ComputCommunications (ISCC), 2010 IEEE Symposium , June 2010

[3] Yi-Ting Mai, Chun-Chuan Yang, and YuHsuan Lin, "Cross-Layer QoS Framework in the IEEE 802. 16 Network," Advanced Communication Technology, The 9th International Conference on , vol209, 12- 14 Feb. 2007

[4] Noordin, K.A. and Markarian, G., "CrossLayer Optimization Architecture for WiMAX Systems," Personal, Indoor and Mobile Radio Communications, 2007. PIMRC 2007. IEEE 18th International Symposium on , pp. 1 7 Sept. 2007

[5] Y. T. Mai, C. C. Yang and Y. H. Lin, "Design of the Cross-Layer QoS Framework for the IEEE 802. 16 PMP Networks," IEICTransactions on Comm., vol. E911360-1369, May 2008

[6] D.-K. TriantafYlIopoulou, N. Passas and A. Kaloxylos, "A Cross Layer Optimization Mechanism for Multimedia Traffic over IEEE 802. 16 Networks", European Wireless 2007, Paris, France, Apr. 2007

[7] Thomas Schierl, Thomas StockhammerThomas Wiegand, “Mobile Video Transmission Using Scalable Video Coding” ,

MMTC E -Letter


Our work contributes to exploit the scalability of SVC for wireless systems such as WiMAX systems

layer approach. We propose some algorithms that drop selective packets to guarantee

ata rate to cope with fluctuating channel conditions of the wireless link.

Thomas Schierl, Cornelius Hellge, Shpend Mirta, Karsten Grüneberg and Thomas Wiegand Fraunhofer; ”H.264/AVC-based Scalable Video Coding (SVC) for Real Time

Wireless IP Networks” for Heinrich-Hertz-

Institute Einsteinufer 37, D-10587 Berlin,

Alaeddine Abdallah, Toufik Ahmed, Raouf Boutaba, “Cross Layer Optimization Architecture for Video Streaming in WIMAX

Computers and Communications (ISCC), 2010 IEEE

Chuan Yang, and Yu-Layer QoS Framework in

the IEEE 802. 16 Network," Advanced Communication Technology, The 9th International Conference on , vol 5, pp.200-

Noordin, K.A. and Markarian, G., "Cross-Layer Optimization Architecture for WiMAX Systems," Personal, Indoor and Mobile Radio Communications, 2007. PIMRC 2007. IEEE 18th International Symposium on , pp. 1 - 4, 3-

Y. T. Mai, C. C. Yang and Y. H. Lin, "Design Layer QoS Framework for the

IEEE 802. 16 PMP Networks," IEICE , vol. E91- B, no. 5, pp.

K. TriantafYlIopoulou, N. Passas and A. yer Optimization

Mechanism for Multimedia Traffic over IEEE 802. 16 Networks", European Wireless 2007,

Thomas Schierl, Thomas Stockhammer and “Mobile Video

Transmission Using Scalable Video Coding” ,

IEEE Transactions on Circuits and Systems for Video Technology, vol. 17, no. 9, Sept. 2007

Shehzad Amir degree in Information and Communication System from the NUST University in 2008, Pakistan and He was in MOBILINK Pakistan from 2008 through 2011 as Network Engineer. He is studying in AJOU University, Korea in

Computer Engineering MS and working in Wireless Net Lab over SVC streaming on WIMAX.

Young-June Choi M.S., and Ph.D. degrees from the Department of Electrical Engineering & Computer Science, Seoul National University, in 2000, 2002, and 2006, respectively. From Sept. 2006 through July 2007, he was a

postdoctoral researcher at the University of Michigan, Ann Arbor, MI, USA. From 2007 through 2009, he was with NEC Laboratories America, Princeton, NJ, USA, as research staff member. He is currently an assistant professor at Ajou University, Suwon, Korea. His research interests include fourthwireless networks, radio resource management, and cognitive radio networks.

Kyungtae Kim degree in Department of Computer Science from the Columbia University in 2000, NY and his Ph.D. degree in the department of Electrical and Computer Engineering at Stony Brook University in 2006. He is working for NEC Laboratori

America Inc during 9 years until now in the areas of multimedia interactive communications over the wireless network, cognitive radio, wireless mesh networks, and mobility management over the heterogeneous networks including 802.16/802.11 networks.

Vol.6, No.9, September 2011

ns on Circuits and Systems for Video Technology, vol. 17, no. 9, Sept. 2007.

received B.E. degree in Information and Communication System from the NUST University in 2008, Pakistan and He was in MOBILINK Pakistan from 2008 through 2011 as Network Engineer. He is studying in AJOU University, Korea in

king in Wireless Lab over SVC streaming on WIMAX.

June Choi received B.S., M.S., and Ph.D. degrees from the Department of Electrical

Computer Science, Seoul National University, in 2000, 2002, and 2006, respectively. From Sept. 2006 through July 2007, he was a

postdoctoral researcher at the University of Michigan, Ann Arbor, MI, USA. From 2007 through 2009, he

America, Princeton, NJ, USA, as research staff member. He is currently an assistant professor at Ajou University, Suwon, Korea. His research interests include fourth-generation wireless networks, radio resource management, and

received a M.S. degree in Department of Computer Science from the Columbia University in 2000, NY and his Ph.D. degree in the department of Electrical and Computer Engineering at Stony Brook University in 2006. He is working for NEC Laboratories

America Inc during 9 years until now in the areas of multimedia interactive communications over the wireless network, cognitive radio, wireless mesh networks, and mobility management over the heterogeneous networks including 802.16/802.11

IEEE COMSOC MMTC E


Multimedia Streaming in Mobile Sensor NetworksEnvironmental Energy Harvesti

Lei Shu, Takahiro Hara, Shojiro Nishio

{lei.shu, hara,

Fig. 1 The complicated research

1. Introduction Multimedia streaming in wireless multimedia sensor networks (WMSNs) has been identified as an important research issue since 2005 in [1], [2] and [3]. Various approaches havefor facilitating multimedia data transmission and gathering in WMSNs in recent years, e.g., [4], [5], and [6]. However, most of these researches only consider the static WMSNs, in which all network elements, i.e., multimedia source nodes, scalar sensor nodes, sink nodes, are fixed. In our previous work [7] and [8], we first time presented the concept of mobile multimedia sensor networks (MMSNs), in which part/all of the network elements can move in the network field, e.g., mobile multimedia source nodes.

Different from multimedia streaming in static WMSNs, multimedia streaming in MMSNs faces a number of challenging issues as listed in follows:

1) Multimedia streaming in sensor networkgenerally needs multiple stable pattransmission [9], if either the multimedia node or the base station can move, i

E-Letter

http://www.comsoc.org/~mmc/ 20/60 Vol. 6, No.9, September

aming in Mobile Sensor Networks with Duty-cycleEnvironmental Energy Harvesting

Lei Shu, Takahiro Hara, Shojiro Nishio, Department of Multimedia Engineering, Osaka University, Japan

{lei.shu, hara, nishio}@ist.osaka-u.ac.jp

The complicated research problem is presented in 5 different layers (aspects).

Multimedia streaming in wireless multimedia sensor networks (WMSNs) has been identified as an important research issue since 2005 in [1], [2]

have been developed ing multimedia data transmission and

gathering in WMSNs in recent years, e.g., [4], [5], and [6]. However, most of these researches only consider the static WMSNs, in which all network elements, i.e., multimedia source nodes, scalar

sink nodes, are fixed. In our previous work [7] and [8], we first time presented the concept of mobile multimedia sensor networks (MMSNs), in which part/all of the network elements can move in the network field, e.g.,

rent from multimedia streaming in static

WMSNs, multimedia streaming in MMSNs faces a issues as listed in follows:

sensor networks generally needs multiple stable paths for

ultimedia source node or the base station can move, it will be very

hard to maintain stable multiple transmission paths in the sensor network.

2) Large-scale wireless sensor networks generally request sensor nodes can wake up and sleep to conserve energy [10] even when the sensor nodes are equippedcapability to harvest environmental energy, e.g., solar power cells. The time-varyingtopology caused by duty-cycled sleep scheduling also prevents the stable multipath transmission MMSNs.

3) Furthermore, if increasing the number of sleeping nodes (sleep rate) in a WSN, the number of explored transmission paths will be seriously affected [11], since the number of useable sensor nodes are reduced. A balance should be found between the sleep rate and the multipath transmission in MMSNs based on the transmission requirements from the multimedia applications.

4) Data transmission in duty-cycled WSNs with environmental energy harvesting should consider selecting the sensor nodes with menergy to balance the energy consumption in the

September 2011

cycle and

Engineering, Osaka

is presented in 5 different layers (aspects).

hard to maintain stable multiple transmission paths

scale wireless sensor networks request sensor nodes can dynamically

erve energy [10] even equipped with the

capability to harvest environmental energy, e.g., varying network

cycled sleep scheduling also prevents the stable multipath transmission in

3) Furthermore, if increasing the number of sleeping nodes (sleep rate) in a WSN, the number of explored transmission paths will be seriously affected [11], since the number of useable sensor nodes are reduced. A balance should be found

the sleep rate and the multipath transmission in MMSNs based on the transmission requirements from the multimedia applications.

cycled WSNs with environmental energy harvesting should consider selecting the sensor nodes with more residential energy to balance the energy consumption in the


http://www.comsoc.org/~mmc/ 21/60 Vol. 6, No.9, September 2011

networks. In this E-letter, the main problem is how to

successfully and efficiently send the multimedia data produced by the source node to the base station while the multimedia source node moves in the network. We analyze and break down this complicated research problem into a number of sub-problems. Solutions for these identified sub-problems have been published in our previous work. As the unique contribution in this E-letter, we propose a new cross-layer optimized framework, which provides the system level overview about our solution for multimedia streaming in mobile multimedia sensor networks.

2. Problem Statement and Analysis In Fig. 1, we give an overview about the research problem and break down the research problem into 5 layers. 2.1 Layer 1: Network Model In this research work, we mainly consider the simplest scenario that having only one mobile multimedia source node in the outdoor network, while all other network elements are static. Both the mobile multimedia source node and the sink node are assumed with unlimited power supply. Each static normal sensor node is powered by rechargeable batteries with the capability to harvest solar power to recharge the batteries from one additional solar power cell. The energy harvesting rate and energy consumption rate in each individual sensor node are different and unpredictable, since many factors can contribute to this and affect the residential energy level, e.g., the unstable local weather, the different number of 1-hop neighbor nodes, and the unexpected query tasks from users. All sensor nodes including the mobile multimedia source node have the same radio transmission range. Each sensor node knows its own location information. 2.2 Layer 2: Energy Constraint Since the energy consumption rate and the energy harvesting rate are not predictable, there is still high possibility that some sensor nodes can consume all their energy before getting recharging. Although these sensor nodes may alive again after harvesting energy for recharging the batteries, their temporally dead status may still affect the service that provided by the WSN, e.g., the temporally dead status of a critical node may result in the partition of the whole network. Therefore, the sleep scheduling strategy should be still applied in WSNs with energy harvesting. A certain number of sensor nodes with less residential energy should

turn to sleep for having the opportunity to recharge the energy.

In our previous publication [10], we proposed a new sleep scheduling algorithm called Energy Consumed uniformly Connected K-Neighborhood (EC-CKN), in which sensor nodes with more residential energy are selected to be awake in each epoch and all selected awake nodes are connected to guarantee the full network connectivity. In this EC-CKN algorithm, there is an important parameter K, which is used for studying the K-connected neighborhoods problem. By changing the value of K, EC-CKN algorithm can dynamically change the sleep rate of a given WSN.

The EC-CKN algorithm is applied in this layer for handling the uncertainty of remaining energy level that caused by both environmental energy harvesting and package transmission. The parameter K is also considered as a method for controlling the sleep rate of the WSN to support multiple transmission paths exploration.

2.3 Layer 3: Mobility Model Many mobility models had already been identified for mobile elements in WMSNs [8], which play an extremely important role for affecting the multimedia streaming performance in WMSNs. In this research work, we again mainly consider the simplest scenario that the mobile source node moves towards the sink node with certain fixed speed along a predefined route. 2.4 Layer 4: Storage Constraint Since the memory storage in sensor is normally small in size, the mobile source node cannot carry all its produced multimedia data by itself until reaching the sink node. Thus, it is necessary for the mobile source node to spread the produced multimedia data to other scalar sensor nodes on the way to the sink node.

Assuming that the mobile source node needs to divide its multimedia data into n pieces in each time slot, and spread these n pieces to n different 1-hop neighbor nodes in that time slot. During the spreading phase, a number of problems can happen:

1) The mobile source node does not have enough number of 1-hop neighbor nodes in certain time slot when moving towards the sink node, since some nodes are sleeping at that time.

2) Some sensor nodes may receive more than one piece of multimedia data within different time slots.

3) Some sensor nodes received the multimedia data may be further from the sink node than the mobile multimedia source node.

IEEE COMSOC MMTC E


4) The unreliable wireless communication with interference may cause failure during the transmission when spreading the multimedia

All these problems require the applying of erasure coding [12] in the mobile multimedia source node. 2.5 Layer 5: Routing

Once the distributed caching of multimedia data is successfully conducted, the next step is how to efficiently forward these multimedia data to the sink node in a duty-cycled WSN.

In our latest publication [13], we proposed a new geographic multipath routing algorithm in dutycycled WSNs named TPGFPlus. Different fromour previous research TPGF in [9], TPGFPlus chooses the next hop node based on 2neighbor information, in which the node that is closest to the sink node is selected. The 1intermediate node then can be chosen based on its remaining energy. Fig. 2 illustrates the basic geographic forwarding policy of TPGFPlus.

Fig. 2 2-hop geographic forwarding example: Node a chooses 2-hop neighbor g as its nextwhich is closest to the sink among all and 2-hop neighbors. Once choosing selects node b as an intermediate 1 This research work revealed two important

insights: 1) Geographic routing in dutyWSNs should be 2-hop based, but not 1because in most existing sleepalgorithms it is mandatory for gathering 2neighborhood information. The gathering of 2neighbors is not an additional overhead for TPGFPlus algorithm; 2) 2-hop based geographic routing allows more nodes to sleep in the network while achieving the same desired average numbeof paths, compared with that of 1algorithm.

The TPGFPlus geographic routing algorithm can be applied in this layer for sending the distributed multimedia data to the sink node.

However, in [13], we also reveal that the sleep rate of the WSN can strongly affect the transmission performance, e.g., the routing path can be longer if the WSN has a higher sleep rate.

MMTC E -Letter

org/~mmc/ 22/60 Vol. 6, No.9, September

4) The unreliable wireless communication with interference may cause failure during the

multimedia data. the applying of

erasure coding [12] in the mobile multimedia

Once the distributed caching of multimedia data conducted, the next step is how to

efficiently forward these multimedia data to the

In our latest publication [13], we proposed a new geographic multipath routing algorithm in duty-cycled WSNs named TPGFPlus. Different from our previous research TPGF in [9], TPGFPlus chooses the next hop node based on 2-hop neighbor information, in which the node that is closest to the sink node is selected. The 1-hop

node then can be chosen based on its illustrates the basic

forwarding policy of TPGFPlus.

hop geographic forwarding example: Node

neighbor g as its next-hop node which is closest to the sink among all a’s 1-hop

hop neighbors. Once choosing g, node a intermediate 1-hop node.

This research work revealed two important insights: 1) Geographic routing in duty-cycled

hop based, but not 1-hop based, because in most existing sleep-scheduling

for gathering 2-hop neighborhood information. The gathering of 2-hop neighbors is not an additional overhead for

hop based geographic routing allows more nodes to sleep in the network while achieving the same desired average number of paths, compared with that of 1-hop based

The TPGFPlus geographic routing algorithm can be applied in this layer for sending the distributed

However, in [13], we also reveal that the sleep

an strongly affect the transmission performance, e.g., the routing path can be longer if the WSN has a higher sleep rate.

3. The Proposed Framework Based on foregoing analysis, we can easily draw the conclusion that to successfully transmit and gather multimedia data from mobile source node in MWSNs, the collaboration among different layers is necessary.

Fig. 3 Parameters from different layers should be jointly considered

Fig. 3 shows the proposed framework in which 5 important parameters are extracted from the complicated problem for consideration:

1) Remaining energy information of each individual sensor node: this parameter can directly affect the sleep rate of the network.

2) Sensor node’s channel capacity and memory storage capacity: this parameter can directly affect the number of nodes that are requested for spreading the divided multimedia data.

3) Sleep rate of the network: this parameter can directly affect the number of awoke nodes for distributed multimedia data distribution and the geographic routing performance.

4) Moving speed of source node: this parameter can directly affect the distributed spreading of the multimedia data, since the multimedia data may bedistributed to the same sensor nodes is the moving speed is very slow, which may overload the 1neighbor nodes.

5) Actual number of distribution of multimedia data: this parameter can highly affect the transmission load in the network as well as the quality of final gathered multimedia data, since the loosing of different piece of distributed multimedia data may serious affect the quality. Acknowledgement This research work in this paper was supported by

September 2011

Based on foregoing analysis, we can easily draw the conclusion that to successfully transmit and

timedia data from mobile source node in MWSNs, the collaboration among different layers

Fig. 3 Parameters from different layers should be

Fig. 3 shows the proposed framework in which 5 important parameters are extracted from the complicated problem for consideration:

1) Remaining energy information of each individual sensor node: this parameter can directly

2) Sensor node’s channel capacity and memory

storage capacity: this parameter can directly affect the number of nodes that are requested for spreading the divided multimedia data.

3) Sleep rate of the network: this parameter can umber of awoke nodes for

distributed multimedia data distribution and the

4) Moving speed of source node: this parameter can directly affect the distributed spreading of the multimedia data, since the multimedia data may be distributed to the same sensor nodes is the moving speed is very slow, which may overload the 1-hop

5) Actual number of distribution of multimedia data: this parameter can highly affect the transmission load in the network as well as the uality of final gathered multimedia data, since the

loosing of different piece of distributed multimedia

This research work in this paper was supported by



Grant-in-Aid for Scientific Research (S)(21220002) of the Ministry of Education, Culture, Sports, Science and Technology, Japan.

Lei Shu is the corresponding author. References [1] Eren Gurses, Ozgur B. Akan. Multimedia Communication in Wireless Sensor Networks. In Ann Telecommun, 60(7-8), pp.799-827, 2005. [2] Ian F. Akyildiz, Tommaso Melodia, Kaushik R. Chowdhury. A Survey on Wireless Multimedia Sensor Networks. In Computer Networks, 51(4), pp.921-960, 2007. [3] Satyajayant Misra, Martin Reisslein, Guoliang Xue. A Survey of Multimedia Streaming in Wireless Sensor Networks. In IEEE Commun Surv Tutor, 10(4), pp.18-39, 2008. [4] Daniel G. Costa, Luiz Affonso Guedes. A Survey on Multimedia-Based Cross-Layer Optimization in Visual Sensor Networks. In Sensors, 11, pp.5439-5468, 2011. [5] Lei Shu, Yan Zhang, Zhangbing Zhou, Manfred Hauswirth, Zhiwen Yu, Gearoid Hynes. Transmitting and Gathering Streaming Data in Wireless Multimedia Sensor Networks within Expected Network Lifetime. In ACM/Springer Mobile Networks and Applications, 13(3-4), pp.306-322, 2008. [6] Lei Shu, Manfred Hauswirth, Yan Zhang, Jianhua Ma, Geyong Min, Yu Wang. Cross Layer Optimization on Data Gathering in Wireless Multimedia Sensor Networks within Expected Network Lifetime. In Springer Journal of Universal Computer Science (JUCS), 16(10), pp.1343-1367, 2010. [7] Lei Shu, Min Chen. Multimedia over Sensor Networks. In IEEE COMSOC MMTC E-Letter, 5(5), pp. 59-61, 2010. [8] Lei Shu, Takahiro Hara, Shojiro Nishio, Yuanfang Chen, Manfred Hauswirth. The New Challenge: Mobile Multimedia Sensor Networks. In International Journal of Multimedia Intelligence and Security, 2(2), pp.107-119, 2011. [9] Lei Shu, Yan Zhang, Laurence T. Yang, Yu Wang, Manfred Hauswirth, Naixue Xiong. TPGF: Geographic Routing in Wireless Multimedia Sensor Networks. In Telecommunication Systems,

44(1-2), pp. 79-95, 2010. [10] Zhuxiu Yuan, Lei Wang, Lei Shu, Takahiro Hara, Zenquan Qin. A Balanced Energy Consumption Sleep Scheduling Algorithm in Wireless Sensor Networks. In IWCMC’11: Proceedings of the 7th International Wireless Communications & Mobile Computing Conference , Istanbul, Turkey, July, 2011. [11] Lei Shu, Zhuxiu Yuan, Takahiro Hara, Lei Wang, Yan Zhang. Impacts of Duty-cycle on TPGF Geographical Multipath Routing in Wireless Sensor Networks. In IWQoS’10: Proceedings of the 18th International Workshop on Quality of Service, Beijing, China, June, 2010. [12] Hakim Weatherspoon and John D. Kubiatowicz. Erasure Coding vs. Replication: A Quantitative Comparison. In IPTPS’02: Proceedings of the 1st International Workshop on Peer-to-Peer Systems, March, 2002. [13] Yuhui Dong, Lei Shu, Guangjie Han, Mohsen Guizani, Takahiro Hara. Geographic Multipath Routing in Duty-cycled Wireless Sensor Networks: One-hop or Two-hop? In MobiCom’11: Proceedings of the 17th Annual International Conference on Mobile Computing and Networking, Las Vegas, Nevada, USA, September 19-23, 2011.

Lei Shu received the B.Sc. degree in Computer Science from South Central University for Nationalities, China, in 2002 and the M.Sc. degree in Computer Engineering from Kyung Hee University, Korea, in 2005 and the Ph.D. degree in Digital Enterprise

Research Institute, from National University of Ireland, Galway, Ireland, in 2010. He is currently a Specially Assigned Research Fellow in Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University, Japan. He has published over 90 papers in related conferences, journals, and books. He had been awarded the MASS 2009 IEEE TCs Travel Grant and the Outstanding Leadership Award of EUC 2009 as Publicity Chair, the Globecom 2010 Best Paper Award. He has served as editors for Wiley, European Transactions on Telecommunications, IET Communications,



KSII Transactions on Internet and Information Systems (TIIS), Journal of Communications, Inderscience, International Journal of Sensor Networks, Wiley, Wireless Communications and Mobile Computing. He has served as more than 30 various Co-Chair for international conferences/workshops, e.g., IWCMC, ICC, ISCC; TPC members of more than 100 conferences, e.g., ICC, Globecom, ICCCN, WCNC, ISCC. His research interests include wireless sensor network, sensor network middleware, multimedia communication, and security. He is a member of IEEE and IEEE ComSoc.

Takahiro Hara received the B.E, M.E, and Dr.E. degrees from Osaka University, Osaka, Japan, in 1995, 1997, and 2000, respectively. Currently, he is an Associate Professor of the Department of Multimedia Engineering, Osaka University. He has published more than 100

international Journal and conference papers in the areas of databases, mobile computing, peer-to-peer systems, WWW, and wireless networking. He served and is serving as a Program Chair of IEEE International Conference on Mobile Data Management (MDM'06 and 10) and IEEE International Conference on Advanced Information Networking and Applications (AINA'09). He guest edited IEEE Journal on Selected Areas in Communications, Sp. Issues on Peer-to-Peer Communications and Applications. He served and is serving as PC member of more than 120 international conferences such as IEEE ICNP, WWW, DASFAA, ACM MobiHoc, and ACM

SAC. His research interests include distributed databases, peer-to-peer systems, mobile networks, and mobile computing systems. He is a member of five learned societies, including senior members of ACM and IEEE.

Shojiro Nishio received the B.E., M.E., and Ph.D. degrees from Kyoto University, Kyoto, Japan, in 1975, 1977, and 1980, respectively. He has been a full professor at Osaka University since August 1992. He served as a Vice President and Trustee of

Osaka University from August 2007 to August 2011. He also acted as the Program Director in the Area of Information and Networking, Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan from April 2001 and March 2008. His research interests include database systems and multimedia systems for advanced networks such as broadband networks and mobile computing environment. Dr. Nishio has co-authored or co-edited more than 55 books, and authored or co-authored more than 560 refereed journal or conference papers. He served as the Program Committee Co-Chairs for several international conferences including DOOD 1989, VLDB 1995, and IEEE ICDE 2005. He has served and is currently serving as an editor of several international journals including IEEE Trans. on Knowledge and Data Engineering, VLDB Journal, ACM Trans. on Internet Technology, and Data & Knowledge Engineering. He is also a fellow of IEICE and IPSJ, and is a member of six learned societies, including ACM and IEEE.



DSP Techniques for Low-cost Mobile Terminals Hai Lin, and Katsumi Yamashita, Graduate School of Engineering, Osaka Prefecture University

([email protected], [email protected]) 1-1 Gakuen-cho, Sakai, Osaka, Japan 599-8531

1. Introduction Nowadays, the increasing smartphone subscribers bring significant challenges to wireless communication systems. In contrast to the traditional mobile phone only for voice communication, the smartphone is also capable of handling multimedia contents. For example, with cloud computing technology, we can now keep our emails, photos, music, and even video clips in the cloud server, and use smartphone to access them from any place covered by wireless hotspots or 3G/4G cellular systems.

In order to deal with heavy traffics introduced by these multimedia contents, current wireless systems have adopted advanced modulation schemes, such as, orthogonal frequency division multiplexing (OFDM), high-order QAM, and turbo coding, to achieve high spectrum efficiency and near-capacity transmission. To enable such complicated modulation schemes, radio frequency (RF) and analog devices in the mobile terminals are expected to be perfect. However, such near-ideal devices are usually quite expensive and an immediate result is the high manufacturing cost of the mobile terminal. Instead of high precision RF devices, a cost-effective solution is realized recently by combining low-cost RF front-end, like direct conversion receiver (DCR), with digital signal processing (DSP) based compensation techniques [1]. This paper will provide an overview of DSP-based compensation methods for RF/analog impairments in the DCRs. 2. RF/Analog Impairments in the DCR Without intermediate frequency (IF) hardware, the received RF signals are directly down-converted to baseband signals in the DCR. Theoretically, the DCR is only made up of one carrier-frequency local oscillator (LO), two mixers, and two low-pass filters. Compared to the super-heterodyne receiver with an IF band, the merits of the DCR include small circuit-board size, low cost, and low power consumption. However, due to several technical challenges in practical implementation, the DCR was not widely adopted until recently, despite it was actually invented around 80 years ago.

Two dominant impairments in the DCR are direct current (DC) offset and I/Q imbalance [2]. The DC-offset arises from signals self-mixing. The I/Q imbalance in a narrow-band system indicates non-90o phase difference and unequal amplitudes between the In-phase (I) and Quadrature-phase (Q) branches, which are constant over the bandwidth thus frequency-independent. In a wideband system, besides the frequency-independent I/Q imbalance, the mismatched frequency responses of branch components further introduce frequency-dependent imbalance [3]. On the other hand, frequency instability of the LO is another critical impairment for OFDM systems. Due to the overlapping spectrum, a carrier frequency offset (CFO) will cause inter-carrier interference (ICI) and then severe performance degradation [4]. Analog compensation for these impairments is difficult or costly.

3. DSP-based Compensation Methods All of these RF imperfections need to be considered in the design of a low-cost OFDM DCR. Since the DC-offset will saturate the subsequent circuits, it is usually eliminated beforehand via inexpensive ac-coupling [1] [2]. As a result, the problem can be simplified to develop an effective digital compensation scheme for CFO and I/Q imbalance.

The CFO estimation in OFDM systems has been well-studied, see [5] [6] and references therein. Also, the compensation for I/Q imbalance in OFDM systems can be found in [7] [8], where the compensation is incorporated into frequency domain channel equalization as pair-wise demodulation. Obviously, it is relatively easy to compensate for one of these two impairments, when another one is absent. However, the coexistence of these impairments is usual in practical receivers. The joint compensation for CFO, frequency-independent and frequency-dependent I/Q imbalance is known as a challenging problem. The frequency domain joint compensation is quite complicated, considering the large number of interference tones [9]. Bearing in mind that the CFO can be corrected easily after I/Q imbalance compensation, current trend is to develop low-



complexity time domain compensator. In the literature, there are three types of time domain I/Q imbalance compensator, where the real-valued filter (RVF) [10-12], the complex-valued filter (CVF) [13], and the complex-valued separation filter (CVSF) [14], are employed, respectively. Among three compensators, the RVF-based one only requires real arithmetic thus has less computational complexity.

Meanwhile, according to the estimation algorithms for their own coefficients, these time domain approaches also can be categorized as pilot-aided or blind. Based on a modified periodic pilot, the authors proposed a joint estimation algorithm in [10], where the CFO estimation is formulated as a nonlinear least squares problem without closed-form solution. In [11], the method in [10] is extended to the context of the IEEE 802.11a Wireless LAN, where two periodic pilots in the preamble are used. A low complexity estimation algorithm with analytic solution was proposed in [12] for the RVF-based compensator, with the assistance of a generalized periodic pilot. It has been shown that the pilot-aided approaches can achieve fast convergence and satisfied performance.

On the other hand, the blind estimation algorithms usually rely on the statistical properties of the signal. A common assumption is that the signal without I/Q imbalance is proper [13] [14], which is true for systems with complex signal constellation [15]. Noteworthy, even distorted by the channel and the CFO, the signal is still proper. Therefore, these blind algorithms are robust against the CFO. The block-type blind estimation in [14] for the CVSF-based compensator has no closed-form solution and requires a gradient descent algorithm with prohibitive high computational complexity. In [13], the authors proposed a blind block-type algorithm and a blind adaptive algorithm for the CVF-based compensator. A closed-form solution is given for the block-type algorithm, which is an approximation depending on the image attenuation (IA) quality of the analog front-end. The blind adaptive algorithm in [13] is a heuristic approach with dynamic performance. Except [11], the existing pilot-aided approaches require special pilots, which however are unavailable in current wireless standard. In most cases, the standard-independent blind approach may be our only option. Clearly, it is important to choose a suitable compensator and the corresponding estimation algorithm in practice.

4. Conclusions In this paper, the existing DSP-based compensation techniques for RF/analog impairments in the DCRs have been reviewed. Current studies show the digital compensation is a promising solution to low-cost mobile terminals. Future research topics in this area may include more effective blind estimation algorithms for the RVF-based and the CVF-based compensators.

References [1] W. Namgoong, T. H. Meng, "Direct- conversion RF receiver design," IEEE Trans. Commun., vol. 49, no. 3, pp. 518-529, Mar. 2001. [2] B. Razavi, "Design considerations for direct-conversion receivers," IEEE Trans. Circ. Syst. II, vol. 44, no. 6, pp. 428-435, June 1997. [3] M. Valkama, M. Renfors, V. Koivynen, "Compensation of frequency-selective I/Q imbalance in wideband receivers: models and algorithms," in Proc. IEEE SPAWC'01, Taoyuan, Taiwan, Mar. 2001, pp. 42-45. [4] T. Pollet, M. Bladel, M. Moeneclaey, “BER sensitivity of OFDM systems to carrier frequency offset and wiener phase noise," IEEE Trans. Commun., vol. 43, pp. 191-193, Feb./Mar./Apr. 1995. [5] Z. Zhang, W. Jiang, H. Zhou, Y. Liu, J. Gao, "High accuracy frequency offset correction with adjustable acquisition range in OFDM systems," IEEE Trans. Wireless Commun., vol. 4, pp. 228-237, Jan. 2005. [6] J. Choi, J. Lee, Q. Zhao, H. Lou, “Joint ML estimation of frame timing and carrier frequency offset for OFDM systems employing time-domain repeated preamble,” IEEE Trans. Wireless Commun., vol. 9, no. 1, pp. 311-317, Jan. 2010. [7] A. Schuchert, R. Hasholzner, P. Antoine, "A novel IQ imbalance compensation scheme for the reception of OFDM signals," IEEE Trans. Consumer Electron., vol. 47, no. 3, pp. 313-318, Aug. 2001. [8] A. Tarighat, R. Bagheri, A. H. Sayed, "Compensation schemes and performance analysis of IQ imbalance in OFDM receivers," IEEE Trans. Signal Process., vol. 53, no. 8, pp. 3257-3268, Aug. 2005. [9] I. Barhumi, M. Moonen, "IQ-imbalance compensation for OFDM in the presence of IBI and carrier-frequency offset," IEEE Trans. Signal Process., vol. 55, no. 1, pp. 256-266, Jan. 2007. [10] G. Xing, M. Shen, H. Liu, "Frequency offset and I/Q imbalance compensation for direct-conversion receivers," IEEE Trans. Wireless Commun., vol. 4, no. 2, pp. 673-680, Mar. 2005. [11] H. Lin, T. Nakao, K. Yamashita, "Joint compensation of frequency-selective I/Q imbalance and



CFO in OFDM-based WLAN," in Proc. IEEE CCNC'08, Las Vegas, NV, Jan. 2008. [12] H. Lin, X. Zhu, K. Yamashita, "Low complexity pilot-aided compensation for carrier frequency offset and I/Q imbalance," IEEE Trans. Commun., vol. 58. no. 2, pp. 448-452, Feb. 2010. [13] L. Anttila, M. Valkama, M. Renfors, "Circularity-based I/Q imbalance compensation in wideband direct-conversion receivers," IEEE Trans. Veh. Technol., vol. 57, no. 4, pp. 2099-2113, July 2008. [14] Y. Tsai, C. Yen, X. Wang, "Blind frequency-dependent I/Q imbalance compensation for direct-conversion receivers," IEEE Trans. Wireless Commun., vol. 9, no. 6, pp. 1976-1986, June 2010. [15] P. J. Schreier, L. L. Scharf, "Second-order analysis of improper complex random vectors and processes," IEEE Trans. Signal Process., vol. 51, no. 3, pp. 714-725, Mar. 2003.

Hai Lin received his B.E. degree from Shanghai JiaoTong University, China, in 1993, and his Dr. Eng. degree from Osaka Prefecture University, Japan, in 2005. Since 2000, he was a

research associate in the Graduate School of Engineering, Osaka Prefecture University, where now he is an assistant professor. His research interests are in general area of signal processing for communication systems.

Dr. Lin is an Editor of the IEEE Transactions on Wireless Communications,

and an Associate Editor of the IEEE Transactions on Vehicular Technology. He has acted many times as session chair and technical program committee member for the IEEE ICC, GLOBECOM, WCNC, VTC, PIMRC, as well as TPC Co-Chair for the Wireless Communications Symposium, IEEE ICC 2011. Dr. Lin is a member of the IEEE and the IEICE, and is currently serving as secretary of the IEEE Communications Society, Signal Processing for Communications Electronics technical committee.

Katsumi Yamashita received the B.E. degree from Kansai University, the M.E. degree from Osaka Prefecture University, and the Dr. Eng. degree from Osaka University in 1974, 1976 and 1985,

respectively, all in electrical engineering. In 1982, he became an assistant professor in University of the Ryukyus, where he was a professor in 1991. Now he is a professor in the Graduate School of Engineering, Osaka Prefecture University.

His current interests are in digital communication and digital signal processing. Dr. Yamashita is a member of the IEEE, the IEICE, and the IEEJ.



Network Planning for DTV Broadcasting to Mobile and Portable Devices José Antonio Arenas, Telefónica Móviles 28013, Madrid, Spain

Unai Gil and Pablo Angueira, Bilbao Engineering College, University of the Basque, Spain

[email protected]

1. Introduction Mobile and digital television services are nowadays one of the challenges that broadcasters are facing. This work focuses on obtaining reference planning values and network architectures for UHF broadcast mobile services. This work presents different scenarios proposed and a cost function that will provide the broadcaster with a tool to evaluate coverage objectives as a function of:

• Reception type: indoor, outdoor, mobile, in vehicle, etc.

• Network type :SFN or MFN • Number of transmitter sites • Transmitted service features

The first step has been the definition of reference scenario environments for this type of services. The classification obtained is shown on Table 1.

TABLE I. Reference scenario environments

Type of Environment

Big City – Urban expansion area

Big City – Old part of town

Big City – Residential area

Big City – Single-family residential area

Big City – Roads

Medium City

Village

Roads

Then, the planning parameters which depend on the environment and that could be important in order to define a deployment strategy have been identified. Specifically, it is important to reduce the cost of the network deployment in an urban environment taking advantage of using two different network infrastructures: the DTV broadcast network and the mobile 2G/GSM or 3G/UMTS network infrastructure. The final result will be a cost function and

associated set of parameters in order to allow an efficient evaluation of the convenience of each deployment strategy in different reception scenarios. The factors that have to be accounted for:

• Coverage. Transmitted Power. • Available data rate and service features • Service Scenarios • The number of transmitters to be used. • Frequency management SFN or MFN. • Network Topologies: MISO, MIMO • Field Strength Spatial variability

Section 2 summarizes different basic concepts to be taken into account in order to evaluate possible reception scenarios where the cost function could be applied, such as, coverage requirements, transmitted signal configuration or classification of reception. Section 3 deals with network infrastructure analyzing the possibility of using broadcast networks, cellular networks or both. Section 4 and 5 present studies related to the alternative of using SFN or MFN networks and MISO or MIMO strategies respectively. Those sections intend to provide the gain that could be achieved using different strategies and therefore, different transmitting scenarios. Finally, in Section 6, it is summed up, a set of results obtained from the field strength location variability study carried out for network planning. 2. Coverage Basis and service features The deployment strategy for network planning depends on the reception required for the transmitted service, which in this case is digital terrestrial television. So two possible scenarios have been taken into account: Portable reception – The receiver stays fixed or moving at a very low speed (walking). It is usually considered that the portable receiver is not moved during reception Error! Reference source not found.. The portable reception occurs mainly in urban areas. There are two types of portable reception: • Class A: Outdoor reception, when the receiver • Class A: Outdoor reception, when the receiver

is located at no less than 1.5 meters above ground level.



• Class B: Ground floor indoor reception where the receiver is located at no less than 1.5 meters above floor level, with a window in an external wall.

Mobile reception – The receiver moves at medium to high speed (no walking speed). There are two types of mobile reception: • Class C: Outdoor reception, when the receiver

is located at no less than 1.5 meters above ground level (for instance, antenna integrated in a car).

• Class D: The receiver is located inside a vehicle (bus or train), at no less than 1.5 meters above ground level.

This study applies the definition of coverage described Error! Reference source not found.. According to this specification, the coverage area is composed of 100m x 100m cells, named “small area”. A small area is composed of 0.5m x 0.5m areas, designated as “receiving location”. A receiving location is regarded as covered if the required C/N and C/I values are achieved for the 99% of the time. The coverage of a small area is classified as: • Good – If at least 95 % of receiving • Good – If at least 95 % of receiving locations at the edge of the area are covered for portable reception and 99 % of receiving locations within it are covered for mobile reception. • Acceptable - If at least 70 % of locations at the edge of the area are covered for portable reception and 90 % of receiving locations within it are covered for mobile reception. Regarding the physical layer, it is necessary to achieve a trade off between the maximum distance between different transmitters and the maximum reception speed, so, the configuration of the transmitted signal is a factor take into account for network planning and in order to determine the cost function. 3. Network Infrastructure One of the key aspects for a successful business model in mobile multimedia broadcasting relies on the fast achievement of a decent coverage percentage with a reasonable cost. A fundamental problem to this success is the cost and effort required to acquire and set up of new transmitting sites. It is essential then to propose deployment strategies that re-use existing infrastructure, mainly DTV and cellular network sites.

• Broadcast Network: It uses high power transmitters in DTV sites. This option will provide

a poor coverage for indoor coverage. Some enhancement might be expected increasing the robustness of the system through modulation and coding schemes. In any case this option would serve rural areas but would not be definitely the solution for urban coverage.

• Hybrid Broadcast/Cellular Network This option would be interesting if the sites of a cellular operator could be used, either a subset of the sites database or the full site list. This architecture would be the best solution for urban environments. In this case the objective of the study for an optimal deployment would consist of obtaining the required density of cellular sites that would complement the DTV site coverage as a function of the coverage objectives and the types of antennas (omni or sector). There are several technical possibilities for delivering DTV to mobile and portable Among them, the use of 3G cellular networks, such as UMTS/HSDPA [2] and MBMS, is one of the worth mentioning. Nevertheless the massive access of users to the mobile television services by cellular networks would imply the demand of an excessive amount of resources. Nowadays, the scale of the cellular networks that would be necessary for providing this service makes this solution unfeasible at a reasonable cost. A more feasible choice can be based on the use of the existing broadcasting networks [3] where the necessary resources, and therefore, the budget, are not so directly dependent on the number of users. The standard DVB-T[2], successfully used for digital television broadcasting services in many countries, has proved the capacity of providing services for mobile reception to certain extent, especially when reception diversity is included. However, this standard has shown some limitations. The main objections lie, first, in the excessive power consumption of the batteries employed in today’s terminals and second, the limited robustness in indoor or mobile reception for receivers without reception diversity. It should be also mentioned the low efficiency of current antennas on handheld devices. Specifically, it is important to reduce the cost of the network deployment in an urban environment taking advantage of using hybrid infrastructures. In order to save cost and time in the future standard for mobile broadcasting DVB-NGH network deployment, it will be interesting to start using DVB-T existing infrastructure for DVB-NGH transmitter installation. Usually, DVB-T sites are



located at high spots and a first coverage layer can be achieved from them. This primary coverage can be completed with sites that today are used by cellular networks. These networks are usually very dense in urban areas in order to supply 3G services (300-500 m coverage radius in an urban area), with less dominant sites (height not much higher than building roofs), and can supply a continuous coverage layer. This mixed solutions has been already suggested in the literature [4]. Both networks have different characteristics which can be used in a cooperative way in order to improve the coverage vs. cost ratio. The coverage provided by those infrastructures for mobile receivers have been analyzed on the basis of simulations corresponding to a real network using different prediction models. The causes of coverage failure and shadow areas have been analyzed in order to highlight the critical factors that affect to mobile DTV services when broadcasted from current DTV sites. At second stage, this coverage have been complemented with small power transmitters following a cellular planning approach in order to fill the gaps left from the broadcast infrastructure. In order to obtain compatibility between both networks, power and frequency restrictions are required not to cause mutual interference. The coverage simulation results are shown in the figure below.

Figure 1. Coverage achieved for different reception types as function of network resources (results for DVB-H). In a first stage, percentages below 80% of the locations will be covered, and this value will be gradually increased by increasing the density of transmitter sites. 4. Spectrum Management: SFN Gain For broadcast networks, the Single Frequency Network mode is an alternative to well known Multiple Frequency Network mode, where instead of transmitters operating at different frequencies, all base stations use the same frequency. Besides

the optimal frequency reuse, it is usually expected that more homogeneous distribution of received signal strength reception in an SFN will improve the quality of service. The SFN gain can be used for coverage planning of future broadcast networks. Two different effects are associated to SFN operation. Some locations will show a contribution of the SFN effect to the coverage (positive gain) and at other locations the SFN effect will be degrading the coverage (negative gain). Only limited research about SFN gain is available [5]. Recently, in [6] an unambiguous methodology to analyze the SFN gain has been proposed, this analysis provides more insight into this so far rather unclear phenomenon. Comparison with previous works shows that results that seemed contradictory (SFN gains between -7 and 3 dB) are explained taking into account that the gain is negative on locations very close to one transmitter, but since the MER is high enough, it has no real negative influence on the reception quality. In general no relevant SFN gain is observed, but on locations where at least two transmitters provide the receiver with a similar signal strength, the SFN gain is positive and the reception quality is improved. An SFN gain of 1.1 dB with a standard deviation of 3.3 dB is obtained for locations on the border between network cells. 5. Network Topologies Finally, the impact of different MIMO schemes should be considered. In principle, the study cases will not depend on the specific coding used for the MIMO solution. The objective of this work is to research on the optimal topologies for different MIMO schemes in mobile broadcast scenarios. The optimization will target the following parameters: energy (EIRP) and number of transmitters. Further work needs to be carried out in order to relate for example the MIMO gain to ratio between received power from different transmitting antennas and delay from the transmitters. This way, the physical layer simulations are required for some number of different observed power balanced and delays between the transmitters. The MIMO gain could be estimated in the whole network basing on the corresponding parameters obtained by network simulations. It has been defined a deployment strategy using MISO or MIMO, and cellular or mixed (cellular+broadcast) networks. So it is necessary to identify urban environment types and simulate the wideband channel for each deployment strategy for

IEEE COMSOC MMTC E


each environment. Finally it is necessary to provide results for each strategy providing a BER curve or simply a gain/degradation curve with respect an ideal situation. It is important to consider polarization, spatial or both diversities in the cases of MISO and MIMO schemes. This is a work to be done in the future because further studieneeded. 6. Field strength location variabilityThe location variability inside the service area in a radio communication network is a key parameter in the infrastructures dimensioning, specifically in portable and mobile services with an important influence of the multipath propagation. The network architecture and radiating power need to be dimensioned as a function of the possible variations of the received field strength level. From the link budget point of view, the location variability effect is considered by means of a lognormal location correction factor introduction that ensured a desired coverage. The location variability of the received field strength derived from the statistical analysis of empirical data has been based on the empirical references of standards defined on DVB (Digital Video Broadcasting) Consortium. At present, the location correction factors are based on ITUP.1546-1 [7] that states a standard deviation of 5.5 dB. The main conclusion is that current values are pessimistic. Most available recommendations regarding the location variability are exclusively referred to cell size resolution of 100x100 meters, which is the typical resolution in broadcast services planning. However, this consideration is no longer valid for network deployment in urban areas oriented to mobile and portable reception where a resolution of 5x5 meters is needed. Nowadays, available cartography data allows fulfilling the analysis proposed in this paper. It has been calculated that for different recenvironments and cell sizes below 100 meters, the standard deviation of the received field strength values are always lower than 3 dB. Those values are clearly below de 5.5 dB value recommended by both, ITU-R and ETSI. Also, the median value of the outdoor location standard deviation has been found to be almost independent of the cell size: 1.5 dB for a cell size of 5x5 meters and 2.2 dB for a cell size of 25x25 meters as shown in the figure for different cities.

MMTC E -Letter


each environment. Finally it is necessary to provide ding a BER curve or

simply a gain/degradation curve with respect an ideal situation. It is important to consider polarization, spatial or both diversities in the cases of MISO and MIMO schemes. This is a work to be done in the future because further studies are

Field strength location variability The location variability inside the service area in a radio communication network is a key parameter in the infrastructures dimensioning, specifically in portable and mobile services with an important nfluence of the multipath propagation. The network architecture and radiating power need to be dimensioned as a function of the possible variations of the received field strength level. From the link budget point of view, the location

considered by means of a log-normal location correction factor introduction that

The location variability of the received field strength derived from the statistical analysis of empirical data has been based on the empirical references of standards defined on DVB (Digital Video Broadcasting) Consortium. At present, the location correction factors are based on ITU-R

1 [7] that states a standard deviation of 5.5 dB. The main conclusion is that current values are

Most available recommendations regarding the location variability are exclusively referred to cell size resolution of 100x100 meters, which is the typical resolution in broadcast services planning. However, this consideration is no longer valid for

twork deployment in urban areas oriented to mobile and portable reception where a resolution of 5x5 meters is needed. Nowadays, available cartography data allows fulfilling the analysis

It has been calculated that for different reception environments and cell sizes below 100 meters, the standard deviation of the received field strength values are always lower than 3 dB. Those values are clearly below de 5.5 dB value recommended by

R and ETSI. Also, the median value of outdoor location standard deviation has been

found to be almost independent of the cell size: 1.5 dB for a cell size of 5x5 meters and 2.2 dB for a cell size of 25x25 meters as shown in the figure for

Figure 2. Statistical values vs cell size. It should be noted that nowadays, the available cellular network planning tools work typically using DTM (Digital Terrain Model) information with a resolution of 5 meters in urban applications and 25 meters for suburban environments. Therefore, it would be suitable to use the values of the location standard deviation obtained for calculating the location correction factor in order to avoid over dimensioning of the network. Finally it has also concluded that a Logapproximation fits well the empirical data. 7. Conclusions This paper summarizes the results obtained for different factors to take into account in network planning for delivering digital terrestrial television to mobile and portable devices. Results gathered in this paper, lead to a quantification of different gains or aspects regarding the network deployment strategies which should be taken into account for network planning depending on the reception scenarios. That is, those results are useful in order to develop a cost function which can allow the broadcaster to choose the more efficient service scenario and configuration to achieve good coverage within a given service area. Nevertheless, further studies are ongoing in order to obtain a complete characterization of the link buddigital television delivering networks. References [1] “Digital Video Broadcasting (DVB): DVBimplementation guidelines” ETSI TR 102 377 v1.1.1 2005 [2] ETSI, “EN300 744 V1.5.1. Digital Video Broadcasting (DVB); Framing structure, channel coding and modulation for digital terrestrial television” European Tlecommunications Standards Institute, Nov2004 [3] F. Hartung et al “Delivery of Broadcast Services in 3G Networks” IEEE Trans. Broadcasting Vol. 53 Issue 1 Pp.188-199 [4] Bria, A.; Gomez-Barquero, D.; “Scalability of DVBH deployment on existing wireless infrastructure”

September 2011

It should be noted that nowadays, the available cellular network planning tools work typically using DTM (Digital Terrain Model) information with a resolution of 5 meters in urban applications and 25 meters for suburban environments.

it would be suitable to use the values of the location standard deviation obtained for calculating the location correction factor in order to avoid over dimensioning of the network. Finally it has also concluded that a Log-normal

e empirical data.

This paper summarizes the results obtained for different factors to take into account in network planning for delivering digital terrestrial television to mobile and portable devices. Results gathered in

to a quantification of different gains or aspects regarding the network deployment strategies which should be taken into account for network planning depending on the reception scenarios. That is, those results are useful in order

n which can allow the broadcaster to choose the more efficient service scenario and configuration to achieve good coverage within a given service area. Nevertheless, further studies are ongoing in order to obtain a complete characterization of the link budget in digital television delivering networks.

[1] “Digital Video Broadcasting (DVB): DVB-H implementation guidelines” ETSI TR 102 377 v1.1.1

[2] ETSI, “EN300 744 V1.5.1. Digital Video Broadcasting (DVB); Framing structure, channel coding and modulation for digital terrestrial television” European Tlecommunications Standards Institute,

[3] F. Hartung et al “Delivery of Broadcast Services in 3G Networks” IEEE Trans. Broadcasting Vol. 53 Issue 1

“Scalability of DVB-H deployment on existing wireless infrastructure”.



Personal, Indoor and Mobile Radio Communications, 2005. PIMRC 2005.IEEE 16th International Symposium on Volume 2, 11-14 Sept. 2005 Pp. 716-720 Vol. 2

[5] G. Santella et al. “SFN Planning for digital terrestrial television and radio broadcast services: the Italian frequency plan for t-dab” Vehicular Technology Conference 2004 vol. 4 pp.2307-2311

[6] D. Plets et al “Methodology for Calculating SFN Gain in Digital Broadcasting Systems” IEEE Transactions on Broadcasting. Accepted for publication. [7] ITU-R Recommendation P.1546 “Method for point-to-area predictions for terrestrial services in the frequency range 30 MHz to 3000 MHz” Radiocommunications Study Group

Jose A. Arenas received the M.S. in

telecommunication engineering at the University of the Basque Country (Spain) in 2000. He has worked in Telefónica Móviles as radio engineer for 9 years in mobile

communications network planning and optimization. He also has been working part time as Professor in the Dpt. of Electronics of the University of Extremadura at Merida Faculty of Engineering for the last 6 years. His current research interests include several aspects of digital broadcast planning. He is working toward his Thesis on mobile multimedia broadcasting network planning

Unai Gil received the MSc. degree in

Telecommunication Engineering in the year 2004 and Phd in 2011 from the Bilbao Engineering College, University of the Basque Country. He

has been involved in several research projects during the last four years in the area of Digital terrestrial TV and Digital Radio broadcasting with the University of the Basque Country. His current research interests include several aspects of Digital Radio Mondiale signal processing applied to Simulcast configuration.

Pablo Angueira (IEEE Member) received the M.S. and PhD in

Telecommunication Engineering at the University of the Basque Country (Spain) in 1997 and 2002 respectively. Prof. Pablo Angueira joined the Dpt. of Electronics and

Telecommunications of the University of the Basque Country in 1998. Within the TSR Research Group, he has been involved for 10 years in research projects to evaluate digital broadcasting systems in different frequency bands. His current research activities are related to DRM and new broadcast technologies (DVB-T2 and NGH). Pablo Angueira is co-author of a wide list of international peer reviewed journals, conference presentations related to digital terrestrial broadcast planning. He has supervised several PhD dissertations in this field. He has also co-authored contributions to the ITU-R working groups WP6 and WP3. Dr. Angueira is an Associate Editor of the IEEE Transactions on Broadcasting journal and Chairs the Strategic Planning Committee of the Broadcast Technology Society Administrative Committee.



TECHNOLOGY ADVANCES ADVANCED VIDEO COMMUNICATIONS

Guest Editor, Honggang Wang, University of Massachusetts Dartmouth, USA [email protected]

The growing needs of video content delivery have posed significant challenges in both video processing and networking, especially when the network resources (e.g., bandwidth and energy) are limited and the requirements of video applications are strict. Today, many handheld devices such as Smartphone have been widely used to support video streaming over cellular networks. In such networks, video applications do not only require intensive bandwidth, but also involve user quality of experiences (QoE). The QoE becomes a new measurement method to replace the traditional Quality of Services (QoS) for video communication applications in recent years. There have been significant research efforts in improving the QoE driven system performance of video communications. In addition, with the rapid growth of the 3D-TV and 3D mobile application in the world, the 3D video processing and communications have drawn much attention from research communities. It is the fact that many multimedia networking techniques have been employed to carry the streams of video content to heterogeneous user ends. The major challenges for delivering video streams over heterogeneous networks include unreliable wireless channels, stringent delay and jitter constraints, degradation of video quality, security assurance, energy efficiency and embedded computing power, etc.. Therefore, it is essential to develop novel wireless multimedia protocols, algorithms, and methodologies to satisfy the future needs of video applications. There are still many emerging issues that need to be addressed, such as video collaborations, power management, adaptations, flexible delivery, and dynamic access, etc., This E-Letter is to bring together the state-of-the-art research contributions that describe the new techniques in advanced video communications. The first paper titled “Multi-Channel Peer-to-Peer Multimedia Streaming” contributed by M. Wang et al., tackles the problems of providing high quality video streaming services caused by bandwidth heterogeneity and proposed a multi-channel P2P streaming system. In this article, they focus on technical trends and solutions in three areas: Multi-channel bandwidth allocation; the influence of

channel popularity; watching videos with heterogeneous end devices. Today, running multimedia applications over mobile devices such as Smartphone has faced significant challenges in both signal processing and networking domains. In the second paper titled “Mobile Multimedia Technology for Internet of Things,” R. Li and M. Chen address some of these challenges by proposing a novel CMDA communication system to support mobile multimedia applications in internet of things. The system provides comprehensive functions in terms of flexible radio coverage, multimedia support, and All-IP communications, etc. In the third paper titled “Downlink VBR Video Scheduling in Cellular Networks with Orthogonal Channels,” S. Mao, et al, highlight the technology trends of using intelligent handheld devices and smart phones to support video in cellular networks. However, the capacity of traditional cellular networks will be greatly stressed due to the characteristics of video applications. M. Shi et al. proposed the joint design and optimization of networking and control mechanisms across multiple layers to improve the overall performance of the video communication system. J. Fang et al. in the fourth paper titled “Motion-Based Region of Interest (ROI) Transmission with Variable Frame Rate over Wireless Multimedia Sensor Networks,” state that large quantity of data and real-time constraint on latency and jitter are two major problems of transmitting multimedia in sensor network. To address these problems, they propose a video streaming scheme which segments and only transmits the Region of Interest (ROI) in a frame. This mechanism greatly reduces the quantity of the video data to be transmitted thus decreases transmission bit rates. Furthermore, the Frame Per Second (FPS) rate of the transmitted ROI frames is adjusted based on the moving velocity of the detected object, which further decreases the transmission bit rate. The cooperative communication techniques have been proposed at multiple network layers to improve the wireless system performance. In the fifth paper titled “ Cross Layer QoS Provisioning for Cooperative Video Transmission,” the authors



studied the QoS provisioning for real time video transmission with cross layer cooperation techniques. The importance and interaction of different parameters among physical, MAC and network layers have been evaluated for different QoS requirements (e.g., bit error rate, throughput, energy allocation, delay and data loss). This multiple layer parameter optimization can be taken as a non-linear optimal problem. This work in the sixth paper titled “ Real-time Video Surveillance for Large Scenes,” stitches the real time video streams into a panorama view in a distributed and hierarchical architecture. The authors mitigate the synchronization problem in distributed video surveillance while providing a convenient, effective and efficient way for security administrators. This work can find its contribution especially in large surveillance scenes. The P2P multimedia applications have gained increasing popularity over the past few years. However, the scheduling algorithms of P2P multimedia applications in the Internet assume perfect channel conditions, which lead to degraded video performance when directly used in the wireless environment. In the paper titled “ Cross-layer Based P2P Scheduling in Wireless Multimedia Networks,” H. Luo et al. propose a novel P2P scheduling algorithm based on distributed cross-layer optimization, which can effectively utilize the limited network resources residing at different network layers and thus significantly improve the user-perceived video performance.

Honggang Wang is an Assistant Professor at the University of Massachusetts, Dartmouth. He received his Ph.D. in Computer Engineering at University of Nebraska-Lincoln in 2009. His research interests include

multimedia communications,

networking, wireless sensor networks, multimedia security and E-Healthcare. He has published more than 50 papers in his research areas. He is the co-winner of the Best Paper Award of the 2008 IEEE Wireless Communications and Networking Conference (WCNC). He serves Associate Editor of Wiley's Security and Communication Networks

(SCN) Journal, KSII Transactions on Internet and Information Systems and Editor board of International Journal of Handheld Computing Research. He also serves as TPC Co-Chair of ACM IWCMC 2010 Multimedia over Wireless Symposium, TPC Vice Chair of 13th IEEE International Conference on Computational Science and Engineering (Advanced Networking and Applications), and TPC Co-Chair of Second IEEE Globecom 2011 Multimedia Communication &Services (MCS) Workshop. He was the TPC members for IEEE ICC 2011, IEEE Globecom 2010-2011, IEEE WiMob 2008, 2009. He currently serves as a Board member of IEEE MMTC (Technical Committee on Multimedia Communications) Services and Publicity. He is a member of IEEE communication society and IEEE Computational Intelligence Society.



Multi-Channel Peer-to-Peer Multimedia Streaming Miao Wang, Zhipeng Ouyang, Lisong Xu and Byrav Ramamurthy, Department of

Computer Science & Engineering, University of Nebraska, Lincoln, NE, 68588, USA {mwang, zouyang, xu, byrav}@cse.unl.edu

1. Introduction Peer-to-Peer (P2P) streaming technologies have been widely used for delivering videos to millions of Internet users (e.g., UUSee[1], one of the top three P2P video streaming service providers in China, claims to serve millions of users watching more than 10,000 channels). Users watching the same channel are organized into the same overlay [2], which is referred to as a channel in this article. Therefore, multiple channels coexist in current P2P streaming systems (referred to as multi-channel P2P streaming systems). In multi-channel P2P streaming systems, bandwidth demands and supplies are different among different channels [3], which is mainly due to the following reasons: 1) P2P streaming users access Internet with various technologies (e.g., DSL, Cable etc.), which provide heterogeneous access bandwidth. 2) Channel popularities are different among multiple channels (i.e., some popular channels attract more users than other less popular channels). 3) Tablet computers and smart phones enable users to watch P2P videos with diverse resolutions and bandwidth. For example, recently, PPStream [4] has released versions for iPhone and iPad users watching videos via its P2P streaming system, where these users have different bandwidth demand and supply compared to normal Internet users. To provide high-quality streaming service, multi-channel P2P streaming systems should be aware of the bandwidth heterogeneity. In this article, we introduce research advances and trends in multi-channel P2P streaming systems, which are in the following areas: 1) Multi-channel bandwidth allocation among different channels. 2) The influence of channel popularity on multi-channel P2P streaming systems. 3) Watching videos with heterogeneous end devices. 2. Bandwidth allocation in multi-channel P2P streaming systems In this section, we focus on bandwidth allocation among different channels in a multi-channel P2P streaming system (i.e., inter-channel bandwidth allocation). Early P2P streaming studies have noticed the bandwidth heterogeneity among different channels (i.e., some channels have

insufficient upload bandwidth for sustaining all peers in these channels, while others still have surplus bandwidth), although they mainly focus on the design of a single-channel system. Wu et al. [5] first proposed a bandwidth competition problem in multi-channel P2P streaming systems, where some users watch multiple channels at the same time and thus the overlay of different channels might overlap with each other. Furthermore, they developed a distributed bidding algorithm for solving the inter-channel bandwidth allocation problem and used a network coding based streaming protocol for utilizing allocated bandwidth for video stream delivery. In our previous work [6] , we solved the inter-channel bandwidth allocation problem with a divide-and-conquer (DAC) strategy. Compared with [5] , the inter-channel bandwidth allocation problem is solved at the channel level, instead of solved at each individual peer. Specifically, each channel selects some peers to represent that channel for participating bandwidth allocation and distributing bandwidth allocation results. Our approach provides a flexible framework for integrating any existing single-channel P2P streaming design into a multi-channel design. In our prototype, we use bootstrap servers for running the bandwidth allocation algorithm. Our DAC protocol should work very well with current Internet infrastructure, in that data centers and clouds are providing reliable and powerful server resources. Both [5] and [6] solved the inter-channel bandwidth allocation problem based on peers watching multiple channels. However, if there is only a small fraction of peers watching multiple channels, there will be insufficient bandwidth to be allocated among these channels, which implies that bandwidth heterogeneity still exists in the multi-channel P2P streaming system. Wu et al. [7] proposed a View-Upload-Decoupling (VUD) approach to further balance bandwidth supply among different channels, which allow users to subscribe to unwatched channels as helpers. VUD design can maximally utilize extra bandwidth in the multi-channel system with the cost of managing peers to join unwatched channels. To compare multi-channel P2P streaming system designs from



the perspective of bandwidth utilization efficiency and implementation complexity, our previous work [8] establishes tractable linear programming models for capturing the characteristics of inter-channel bandwidth allocation, which can be efficiently solved with high-end desktop computers. Based on our large-scale numerical simulations, we find that VUD design is required in some specific application scenarios. 3. Influence of channel popularity on multi-channel P2P streaming systems Channel popularity is another cause of bandwidth supply and demand heterogeneity. In a multi-channel P2P streaming system, most of users watch a small number of popular channels [9] . Our previous study [10] on channel popularity of user generated content websites also shows high skewness on video popularity, which has a Zipf like distribution. Intuitively, channel popularity skewness implies bandwidth demand and supply imbalance among different channels. Multi-channel P2P streaming systems design should consider channel popularity skewness. However, existing research [9] [10] mainly focuses on static analysis of video popularity, in that the dynamic characteristics are difficult to retrieve, which means that it is difficult to precisely predict video popularity with historical data. Hence, traffic monitoring based approaches have recently been proposed to predict bandwidth demand and supply in P2P streaming systems. Niu et al. [11] proposed a learning framework for estimating peer bandwidth demand and supply changes with simple machine learning algorithms in peer-assisted Video-on-Demand (VOD) systems. Furthermore, the same authors apply time-series based models to forecast peer demands and predict P2P VOD system performance [12] , which shows that it is feasible to apply classic data analysis and prediction models on adaptively improving streaming quality of all channels. The problem of improving multi-channel P2P streaming system performance via channel popularity prediction is still open, in that it requires collecting data of real systems and depends greatly on large-scale data mining techniques. It will be a good research topic on multi-channel P2P streaming systems. 4. Watching P2P videos with heterogeneous devices As previously mentioned, users are able to watch P2P videos from mobile devices, such as smart

phones and tablet devices, which introduce more bandwidth supply and demand heterogeneity due to the following reasons: 1) Mobility introduces more bandwidth heterogeneity. The upload and download capacity varies based on network coverage and types (e.g., 3G, WiFi). 2) Mobile devices require different video stream resolutions, since they have different display resolutions. It is very challenging to design a P2P streaming system that can provide high quality of service in a heterogeneous environment. We have begun to study this problem from an application perspective, where users might watch videos of different resolutions [13] [14] . Our motivation is to enable cooperation between peers watching high resolution videos and those watching low resolution videos. Specifically, we solve the window resizing problem (i.e., peers can dynamically change their watching video resolutions) with the goal of minimizing the window resizing delay (i.e., the playback delay after a window resizing) and minimizing the bandwidth consumption. We propose a partial forwarding scheme based on layered coding to let peers watching low resolution videos to forward the enhancement layer of high resolution videos. As future work, mobility models, network dynamics, video coding schemes and power consumptions should be considered in designing multi-channel P2P streaming systems with mobile devices. 5. Conclusion In this article, we briefly discussed the advances in multi-channel P2P streaming systems from the perspective of bandwidth demand and supply heterogeneity. Then, we introduced some trends in P2P streaming systems, where solving heterogeneity problem is the key. References [1] http://www.uusee.com/ [2] Y. Liu, Y. Guo, C. Liang, A survey on peer-to-peer video streaming systems, Journal of Peer-to-Peer Networking and Applications 1 (1) (2008) 18–28. [3] X. Hei, C. Liang, J. Liang, Y. Liu, K. Ross, A measurement study of a large-scale P2P IPTV system, IEEE Transactions on Multimedia 9 (8) (2007) 1672–1687. [4] http://ppstream.com/ [5] C. Wu, B. Li, Z. Li, Dynamic bandwidth auctions in multi-overlay P2P streaming with network coding, IEEE Transactions on Parallel and Distributed Systems 19 (6) (2008) 806–820.



[6] M. Wang, L. Xu, B. Ramamurthy, A flexible divide-and-conquer protocol for multi-view peer-to-peer live streaming, in: Proceedings of IEEE P2P, 2009, pp. 291–300. [7] D. Wu, C. Liang, Y. Liu, K.W. Ross, View-upload decoupling: a redesign ofmulti-channel p2p video systems, in: Proceedings of IEEE INFOCOM, Mini-Conference, Rio de Janeiro, Brazil, 2009, pp. 1–6. [8] M. Wang, L. Xu, and B. Ramamurthy, “Linear programming models for multi-channel P2P streaming systems,” in Proceedings of IEEE INFOCOM Mini-Conference, 2010. [9] C. Wu, B. Li, and S. Zhao, “Multi-channel live P2P streaming: Refocusing on servers,” in Proceedings of IEEE INFOCOM, Phoenix, AZ, April 2008. [10] G. Li, M. Wang, J. Feng, L. Xu, B. Ramamurthy, W. Li and X. Guan, “Understanding User Generated Content Characteristics : A Hot-Event Perspective”, in Proceedings of IEEE ICC, 2011, Japan, pp. 1-5. [11] D. Niu, B. Li, and S. Zhao. “Self-Diagnostic Peer-Assisted Video Streaming through a Learning Framework,” in the Proceedings of ACM Multimedia 2010, Florence, Italy, October 25-29, 2010. [12] D. Niu, Z. Liu, B. Li, and S. Zhao. “Demand Forecast and Performance Prediction in Peer-Assisted On-Demand Streaming Systems,” in Proceedings of IEEE INFOCOM 2011 Mini-Conference, Shanghai, China, April 11-15, 2011. [13] Z. Ouyang, L. Xu, and B. Ramamurthy, "Partial Forwarding Scheme for Dynamic Window Resizing in Live P2P Streaming Systems", in Proceedings of IEEE GLOBECOM, New Orleans, LA, November, 2008. [14] Z. Ouyang, L. Xu, and B. Ramamurthy, "A Cooperative Scheme for Dynamic Window Resizing in P2P Live Streaming", in Proceedings of IEEE ICC, Germany, June, 2009

Miao Wang received his Ph.D. degree from the University of Nebraska-Lincoln in 2011. He is currently a research software engineer at A10networks, San Jose, CA. His research interests include Peer-to-Peer computing, multimedia streaming and protocol design. Zhipeng Ouyang received his B.E. and M.E. degrees in Computer Science from the East China Normal University Shanghai in 2003 and 2006, respectively. He is currently a Ph.D. student

with Professor Lisong Xu and Professor Byrav Ramamurthy at the University of Nebraska-Lincoln.

Lisong Xu received his B.E. and M.E. degrees in Computer Science from the University of Science and Technology Beijing in 1994 and 1997, respectively. He received his Ph.D. degree in Computer Science from North Carolina State

University in 2002. From 2002 to 2004, he was a Post-Doctoral Research Fellow at North Carolina State University. He is currently an associate professor in Computer Science and Engineering at the University of Nebraska-Lincoln. He is a recipient of NSF CAREER Award (2007), UNL CSE Student Choice Outstanding Teaching Award (2006, 2007, 2008, 2010, 2011), and UNL College Distinguished Teaching Award (2011).

Byrav Ramamurthy is currently a Professor in the Department of Computer Science and Engineering at the University of Nebraska-Lincoln (UNL). He is the author of the book "Design of Optical WDM Networks - LAN, MAN and WAN

Architectures" and a co-author of the book "Secure Group Communications over Data Networks" published by Kluwer Academic Publishers/Springer in 2000 and 2004 respectively. He serves as the Chair of the IEEE Communication Society's Optical Networking Technical Committee (ONTC). He serves as the IEEE INFOCOM 2011 TPC Co-Chair. His research areas include optical and wireless networks, peer-to-peer networks for multimedia streaming, network security and telecommunications. His research work is supported by the U.S. National Science Foundation, U.S. Department of Energy, U.S. Department of Agriculture, AT&T Corporation, Agilent Tech., Ciena, HP and OPNET Inc.



Mobile Multimedia Technology for Internet of Things Roy Lai, Confederal Networks Inc

Min Chen, School of Computer Science and Engineering, Seoul National University, Korea

([email protected] , [email protected])

The Internet of Things, the ultimate in ubiquitous networking, will greatly impact everyone’s life in every facet, including entertainment, socialization, business, healthcare and education [1]. In order to enhance the end-user quality of experience, an important challenge is to support mobile multimedia applications in internet of things [2]. Especially, in emergency situations, along with other scalar context-aware data, real-time mobile multimedia is critical to elaborate complicated on spot events and phenomena, facilitating command center or monitoring staff to make quick and accurate decision and actions. Figure 1 shows two recent examples of emergency situation, such as tsunami in Japan, and riot in Libya.

(a) Tsunami in Japan (b) Riot in Libya Figure 1. Illustrative Examples of Emergency Situation. Recently, the typical “last mile” coverage technologies for terminal networks in Internet of Things includes wireless sensor networks, Zigbee, Bluetooth, WLAN, WiFi, RFID, and wireless mesh networks, etc [3]. However, they have the intrinsic characteristics of low data rate [4], line-of-sight constraints, and/or limited radio coverage, which make them hard to support mobile multimedia/video applications. Furthermore, due to the short distance between context-aware data collectors and the access point, the terminal networks easily become “information islands” when emergency situations happen, such that the gateway/access points may fail to work or become disable due to the nearby destructive events.

Figure 2 System Architecture for All-IP communication technology and Internet of things [5] In order to address above challenges from practical point of view, we are developing a novel CMDA communication system for the use in Internet of Things, which can provide comprehensive functions in terms of flexible radio coverage, multimedia support, and All-IP communications, etc. The basic system architecture is shown in Figure 2.

Figure 3 Potable CDMA Base Transceiver Station (BTS) [5] Previously, we have developed KT-28 CDMA communication system for coal mine (System V1.0), which is the only mobile communication which can cover the whole underground tunnel and satisfy all of the requirements proposed by coal mines industry. In order to further support mobile multimedia applications with low cost, adaptive coverage and flexible deployment, we have to further tailor our BTS into small one (e.g. Confederal Potable BTS in Figure 3), and develop a novel SIP CDMA platform to replace traditional



CDMA core network in circuit switched domain.

(a) Simple architecture (b) video camera product

(c) 3G Mobile Video Management System

Figure 4. Illustrative Example for 3G Mobile Video System. Currently, our solution has been adopted by wide range of applications in the area of Internet of Things, such as intelligent video surveillance for ships, mobile surveillance system for policeman (e.g., Fig. 4), coal mines, and environment emergency response, etc. References [1] L. Atzori, A. Iera, G. Morabito, “The Internet of Things: A survey”, Computer Networks, Vol. 54, No. 15, pp. 2787-2805, 2010. [2] M. Chen, M. Guizani, M. Jo, “Mobile Multimedia Sensor Networks: Architecture and Routing”, MobiWorld 2011, Shanghai, China, April 10-15, 2011. [3] M. Chen, S. Gonzalez, A. Vasilakos, H. Cao and V. Leung, “Body Area Networks: A Survey,” ACM/Springer Mobile Networks and Applications, Vol. 16, No. 2, pp. 171-193, April 2011. [4] M. Chen, S. Gonzalez, H. Cao, Y. Zhang, and S. Vuong, "Enabling Low Bit-rate and Reliable Video Surveillance over Practical Wireless Sensor Networks", Journal of Supercomputing, DOI: 10.1007/s11227-010-0475-2 [5] http://www.confednet.com/

Roy “Xiaorong” Lai is the founder and CEO of Confederal Technology Corporation. Mr. Lai was the General Manager of China Potevio CDMA Division. Potevio is one of three CDMA

network providers in China. In 2006, he successfully introduced CDMA communication system into China coal mine industry by cooperated with CARI, with subcontractors as UTStarcom Canada and Arasor International, USA. He was the technical Manager of Direct Paging Dept., UTStarcom from July 1997 to Dec. 1998. He was the Managing Director of Telecom Institute of Concord Group USA when he was 26 years old, managed a team of more than 100 engineers.

Min Chen is an assistant professor in School of Computer Science and Engineering at Seoul National University (SNU). He was the R&D Director for Confederal Networks Inc. He has worked as a Post-Doctoral Fellow in Dept. of

Electrical and Computer Engineering at UBC for three years since Mar. 2009. Before joining UBC, he was a Post-Doctoral Fellow at SNU for one and half years. He has published more than 130 technical papers. Dr. Chen received the Best Paper Runner-up Award from QShine 2008. He serves as editor or associate editor for Wiley I. J. of Wireless Communication and Mobile Computing, Wiley I. J. of Security and Communication Networks, Journal of Internet Technology, KSII Transactions on Internet and Information Systems, IJSNet, etc. He is a managing editor for IJAACS. He was a TPC co-chair of BodyNets 2010. He is a symposia co-chair and workshop chair of CHINACOM 2010. He was co-chair of MMASN-09 and UBSN-10. He was the TPC chair of ASIT-09, ASIT 2010, TPC co-chair of PCSI-09 and PCSI-10. He is a symposia co-chair for IEEE ICC 2012 and IEEE ICC 2013. He is an IEEE senior member.



Downlink VBR Video Scheduling in Cellular Networks with Orthogonal Channels Shiwen Mao, Yingsong Huang, and Yihan Li, Auburn University, Auburn, AL, USA

[email protected], [email protected], [email protected]

1. Introduction With the wide availability of intelligent handheld devices and smart phones, there is a compelling need to support video in traditional cellular networks. Video applications not only are bandwidth intensive, but also involve user quality of experiences. It is foreseeable that the capacity of traditional cellular networks will be greatly stressed, while joint design and optimization of networking and control mechanisms across multiple layers would be necessary. Among various forms of compressed video, variable bit rate (VBR) videos can offer constant and better quality over constant bit rate (CBR) videos given the same bit budget; a highly desirable feature for video applications. However, VBR videos also pose great challenges to network scheduling and control, due to the high variability and complex autocorrelation structure in VBR video traffic [1]. In an interesting work [2], Salehi, et al. applied majorization [3] to VBR video smoothing and developed a smoothness optimal algorithm, assuming a constant rate channel. In [4, 5], the authors study the problem of VBR video over wireless channels, both focusing on one VBR stream over a given time-varying wireless channel. In our prior work [6], we investigate the problem of supporting multiple VBR videos in a multi-cell network, where capacity is limited by inter-cell interference. We develop a reformulation-linearization technique (RLT) approach as well as a distributed algorithm based on dual decomposition. In this work, we focus on the downlink of a cell with orthogonal channels, where the base station (BS) streams multiple VBR videos to mobile users in the cell. We consider buffer underflows (causing stalled display) and overflows (causing missing frames and error propagations in the following received and decoded frames) as user viewing performance measure, and aim to minimize the total power consumption at the BS. In particular, we adopt a deterministic model for VBR video traffic that incorporates video frame and playout buffer characteristics, and formulate

a constrained stochastic optimization problem. We show that the problem fits well with majorization theory, which concerns with partial ordering of real vectors and order-preserving functions [3], and develop a majorization-based solution framework. For the case of a single VBR video session with relaxed peak power constraint, a power optimal algorithm with low complexity is developed. We prove the power optimality of the proposed algorithm and the uniqueness of the global optimum, and demonstrate that the proposed algorithm is also smoothness optimal. For the case of multiuser VBR video streaming, we develop a heuristic algorithm that selectively suspends some video sessions when the peak power constraint is violated. The proposed algorithms are evaluated with trace-driven simulations, and are shown to achieve considerable power savings and improved video quality over a conventional “lazy” scheme. The remainder of this letter is organized as follows. In Section 2, we introduce the system model and problem formulation. The main results are summarized in Section 3. Section 4 concludes this letter. 2. System Model and Problem Formulation We consider the downlink of a cellular network. There are N mobile users in the cell. A BS streams multiple VBR videos to the mobile users. Each user occupies a downlink channel, which is a spectral/time resource slot, the nature of which depends on the specific multiple access technique adopted. We assume that the downlink channels within a cell are orthogonal, due to perfect synchronization of the spreading codes or the use of guard times or frequencies. We adopt a deterministic model for VBR videos that considers frame sizes, frame intervals, and playout buffers [2]. Let Dn(t) denote the cumulative consumption curve of the n-th user, representing the cumulative amount of bits consumed by the decoder at time t. Given the playout buffer size bn, we can derive a cumulative overflow curve for user n, Bn(t), which is the maximum number of bits that can be received at time t without overflowing user n’s playout buffer. Finally, we define cumulative transmission curve Xn(t) as the total amount of bits



transmitted to user n at time t. The three curves are illustrated in Fig. 1.

Fig. 1 Deterministic model for VBR video scheduling. Generally Xn(t) is a piece-wise linear curve. For a linear segment, its slop is the transmission rate for that period of time, which depends on the signal-to-noise ratio (SNR), or, the BS transmit power allocation to the session. A feasible transmission schedule should cause neither playout buffer underflow nor overflow, i.e., satisfying Dn(t) ≤ Xn(t) ≤ Bn(t), for all t, n. Therefore, the problem is to find the optimal feasible transmission schedules {Xn(t), 0 < t ≤ Tn}, for each user n, such that the total transmit power consumption can be minimized. We then formulate a constrained stochastic optimization problem, to minimize the expectation of the total transmit power under buffer underflow, overflow and peak power constraints [7]. 3. Main Results We find the formulated problem fits well with majorization theory, which provides order preserving results for inequality problems [3]. With such an interpretation, the problem is to find a rate vector that is majorized by all other feasible rate vectors. We first prove that the objective function of the formulated problem is Schur-convex with the order-preserving property [3]. Secondly, we investigate the case of a single VBR video session with relaxed peak power constraint, for which a majorization-based power optimal algorithm with low complexity is developed. We prove the power optimality of the proposed algorithm and the uniqueness of the global optimum. In addition, we demonstrate that the proposed algorithm is also optimal with respect to smoothness [2].

Thirdly, we investigate the case of multiuser VBR streaming, where power allocations for the users are coupled with the peak power constraint. We develop a heuristic algorithm that selectively suspends some video sessions, which will not incur underflow in the next time slot, when the peak power constraint is violated. The proposed algorithms are evaluated with trace-driven simulations, using VBR video traces from the Video Trace Library at Arizona State University, and are shown to achieve considerable power savings over a conventional “lazy” scheme. This finding is somewhat counter-intuitive, since the lazy scheme seems to be energy efficient by only transmitting the minimal amount of video data in each time slot. Interested readers are referred to [7] for details. 4. Conclusions In this paper, we studied the problem of downlink multiuser VBR video streaming in cellular networks. We formulate a constrained stochastic optimization problem aiming to minimize the BS power consumption and to avoid playout buffer overflow or underflow. We developed majorization-based algorithms to solve the formulated problem. The superior performance of the proposed algorithms over a conventional scheme is validated with trace-driven simulations. Acknowledgment: This work is supported in part by the US National Science Foundation (NSF) under Grants CNS-0953513, ECCS-0802113, and IIP-1127952. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the foundation. References [1] M. W. Garrett and W. Willinger, “Analysis, modeling

and generation of self-similar VBR video traffic,” ACM SIGCOMM Comput. Commun. Rev., vol.24, no.4, pp.269–280, 1994.

[2] J. Salehi, Z.-L. Zhang, J. Kurose, and D. Towsley,

“Supporting stored video: reducing rate variability and end-to-end resource requirements through optimal smoothing,” IEEE/ACM Trans. Networking, vol.6, no.4, pp.397–410, Aug. 1998.

[3] A. W. Marshall and I. Olkin, Inequalities: Theory of

Majorization and Its Applications. New York, NY: Academic Press, 1979.

[4] T. Stockhammer, H. Jenkac, and G. Kuhn, “Streaming



video over variable bit-rate wireless channels,” IEEE Trans. Multimedia, vol.6, no.2, pp.268–277, Apr. 2004.

[5] G. Liang and B. Liang, “Balancing interruption

frequency and buffering penalties in VBR video streaming,” in Proc. IEEE INFOCOM’07, Anchorage, AK, May 2007, pp.1406–1414.

[6] Y. Huang and S. Mao, “Downlink power control for variable bit rate video over multicell wireless networks,” in Proc. IEEE INFOCOM’11, Shanghai, China, Apr. 2011, pp.2561–2569.

[7] Y. Huang, S. Mao, and Y. Li, “Downlink power

control for VBR video streaming in cellular networks: A majorization approach,” in Proc. IEEE GLOBECOM’11, Huston, TX, Dec. 2011.

Shiwen Mao (S’99-M’04-SM’09) is an Associate Professor in the Department of Electrical & Computer Engineering at Auburn University, Auburn, AL. Before joining Auburn University, he was a research scientist with Virginia Tech from

2003 to 2006 and Polytechnic Univesity in 2006. His research interests include cross-layer design and optimization of wireless networks and multimedia communications, with current focus on cognitive radio networks, 60GHz mmWave networks, and free space optical networks. He is on the Editorial Board of IEEE Transactions on Wireless Communications and several other journals. He chairs the Interest Group on Cross-layer Design for Multimedia Communications under IEEE Communications

Society's Multimedia Communications Technical Committee. Dr. Mao received the NSF CAREER Award in 2010 and the 2004 IEEE Communications Society Leonard G. Abraham Prize in the Field of Communications Systems.

Yingsong Huang received the B.S. and M.S. degrees from Chongqing University, Chongqing, P.R. China, in 2000 and 2003, respectively. From 2003 to 2007, he joined Advantech Co., Ltd. R&D center, Beijing Branch, P.R. China.

Since 2007, he has been a graduate student pursuing Ph.D. degree in the Department of Electrical and Computer Engineering, Auburn University, Auburn, AL, USA. His research interests include network congestion control and video communications.

Yihan Li received the Ph.D. degree in Electrical and Computer Engineering in 2004 from Polytechnic University, Brooklyn, NY in 2000. Currently, she is a Visiting Assistant Professor in the Department of Electrical and Computer Engineering, Auburn University, Auburn, AL. She

was a Research Scientist in the Department of Electrical and Computer Engineering at Polytechnic University from 2004 to 2006, and a postdoctoral fellow in the Department of Electrical and Computer Engineering at Auburn University from 2006 to 2009. Her research interests include Scheduling in wired and wireless networks, wireless ad hoc networks and high-speed packet switching.



Motion-Based Region of Interest (ROI) Transmission with Variable Frame Rate over Wireless Multimedia Sensor Networks

Jian Fang and Alvin Lim, Auburn University, Auburn, AL 36849, USA Qing Yang, Montana State University, Bozeman, MT 59717, USA

1{fangjia, limalvi}@auburn.edu, [email protected]

1. Introduction Wireless Sensor Networks (WSN) has attracted great interest in research community in recent years due to its potential support for a wide range of applications. However, most of the WSNs deployed today can only collect and transmit scalar data of physical phenomenon like temperature, speed, or pressure due to the limited capability of sensors. With the advent of miniaturized and less expensive hardware like CCD cameras and sensitive microphones, sensor nodes (SN) may be augmented to form a Wireless Multimedia Sensor Networks (WMSN) that can deliver real-time video/audio information about the monitored environment, and this may greatly expand the realm of usability of WSNs [1]. However, WMSN poses great challenges to the system design and computing algorithms. Multimedia data has two inherent properties which are different from that of scalar data: large quantity and real-time constraint on latency and jitter. Both properties require high bandwidth of the transmission channel and high on-node or in-network data processing capability. However, this requirement is diametrically opposite to the characteristics of WSN, because although their functionalities are constantly being improved, SNs in a typical WSN usually have irreplaceable low power battery, limited computing capability and less memory. The wireless channel is also more prone to interference, multi-path fading, and congestion than wired networks. Thus, to implement a WMSN, the requirements for real-time multimedia streaming, which is often defined as Quality of Service (QoS), must be tuned to the characteristics of the WMSN. Most of the current research efforts in WMSN focus on a specific layer of the network protocol stack to improve multimedia steaming QoS, e.g., distributed source coding is applied at the application layer [2]; multi-path transmission is used at the network layer to guarantee reliability [3]; and contention-free protocol is used at the MAC layer [4] to reduce latency and increase throughput. Excellent surveys on multimedia streaming techniques and algorithms at different network protocol stack layers of WMSN can be

found in [1] [5]. Since large quantity of data and real-time constraint on latency and jitter are two major problems in WMSN, in this work, we focus on video data reduction and decrease of latency and jitter. Actually, high data redundancy often exists in a WMSN, for instance, the background information in a video surveillance WSN with static cameras is largely redundant and should not be transmitted frequently. Therefore, we propose a video streaming scheme for a WMSN which segments and transmits only the Region of Interest (ROI) in a frame. This will greatly reduce the quantity of the video data to be transmitted thus decrease transmission bit rate. Furthermore, the Frame Per Second (FPS) rate of the ROI frames transmitted is adjusted based on the moving velocity of the object detected, which further decreases transmission bit rate. Compared with existing techniques for multimedia streaming in WMSN, our approach has the following advantages:

(1) The sensor node transmits only the ROI and background information is transmitted only intermittently.

(2) The ROI frame is independent of any inter-frame or intra-frame information. Thus, a specific packet loss will not affect the received video quality as does by the loss of a key frame such as the I-frame in MJPG encoding.

(3) Most of the existing encoding algorithms with variable compression bit rate are based on the feedback from the network or the receiver [5]. However, in our approach, we adjust the bit rate based on the video source information, which is usually the very reason for congestion and packet loss in the network.

2. Related Works Reducing data to be transmitted but providing low quality loss, reliable and error-resilient application data is the responsibility of encoders at the application layer. A wide variety of encoding techniques exist in the research literature. In [6], a low-complexity video compression algorithm is presented which is based on change/difference



detection. In this method, a frame is dived into 8x8 blocks as in JPEG. A change detection algorithm is run against the frame and marks those blocks showing changes with respect to the reference frame. These marked blocks are encoded by means of a JPEG like technique. The method is claimed to achieve a compression quality similar to MPEG-2 at a much less computational cost. However, the disadvantages of this algorithm are, to mark all the blocks with changes, four passes are needed to run against the frame and the position information for marked bocks may incur a high overhead for the data transmitted, especially when the marked area is of irregular shapes. A widely studied category of encoding algorithm, Wyner-Ziv encoding, is presented in [7]. This algorithm can achieve highly efficient lossy compression by exploring source side information at the decoder. The most significant characteristic of this algorithm is it reverses of complexity of encoding and decoding at the encoder and decoder sides. In traditional compression techniques, encoding is much more complex and computationally expensive than decoding. However, the Wyner-Ziv algorithm has a much less computing overhead at the encoder. This makes it more appropriate for WSNs because of the limited power and computing capability in sensor nodes. Since it leverages source side information for decoding, algorithms similar to Wyner-Ziv algorithm are also called distributed source coding. However, this algorithm is theoretically sound but difficult to implement in practice because it requires overlapping of Fields of View (FOV) of sensors and the overlapping FOVs need to be identified at the joint decoder for decoding. A Multi-layer coding with JPEG 2000 based wavelet compression algorithm is proposed in [8]. This algorithm uses wavelet based decomposition to create multiple encoded bit-streams and the wavelet coefficients are grouped into trees and encoded separately. So an image can be encoded as scalable bit streams with multiple layers. JPEG 2000 encoding with multiple layers makes ROI encoding possible. In [9], a JPEG 2000 compatible encoding method is presented to encode and transmit ROIs. This algorithm is similar to ours in terms of ROI segmentation. However, it requires background frames to be transmitted frequently for image decoding. The major disadvantages of multi-layer coding based on wavelet compression are the coding algorithm is complex and may consume energy quickly, and heavy computations of this

algorithm may increase latency and disrupt real-time video transmissions. 3. Proposed Scheme We propose a noval real-time video transmission scheme in a WMSN for video surveillance. The WMSN consists of multiple sensors. Some sensors are equipped with video cameras. When a moving object is detected, the SN detecting the object needs to transmit the captured video in real-time to the sink node which may make a decision for further actions base on the video, e.g., instructing other sensor devices to track the object. Before presenting the proposed scheme in detail, we make the following assumptions:

(1) The WMSN is a video surveillance WSN with static cameras. Thus, the content of the background of the video captured is relatively stable and has little variations over time.

(2) The ROI covers the moving object completely, which is a person in our experiment. This will reduce visual noise in integration of ROI with the background at the receiver.

(3) Only one moving object will appear in the view of the camera.

In the proposed WMSN, when an object enters the monitored area, multiple cameras may be activated and start transmitting video frames at the same time. The abrupt surge of data in transmission may cause severe signal interference, network congestion and packet loss although at this very moment the sink node needs the video information the most. Data transmitted in this scenario has a high level of redundancy. For a static camera monitoring a stable background, it will waste energy and bandwidth to transmit background frames frequently. Further, it’s not necessary either to transmit the video frames with high FPS when the detected object moves slowly because of the limitation of our visual acuity. We take two steps to reduce these data redundancies. First, we transmit only the frame segment including the moving object detected (the ROI in Fig. 1). The ROI is identified by running a motion detection algorithm at the server node. Since the background is stable with little variations, the background frame will be transmitted infrequently, e.g., once every five minute.



To integrate the ROI with the background at sink node (the Client node in Fig. 1), it needs to know the size and the relative position of the ROI within the background frame. The position and size of ROI is determined by the motion detection algorithm running on the server node.

Figure 1. ROI Transmission and Running Threads in the Server and Client Nodes. Once the client node gets the ROI video data and the size and position of the ROI, the integration of the video frame can be done by frame embedding with the ROI as the frame mask which replaces the corresponding pixels in the background frame. The video is reconstructed on the client node by embedding the ROI within the background frame on a frame-by-frame basis. The video reconstruction process and multiple threads running on the server and client nodes are illustrated in Fig. 1. In transmission of ROI data, compression is not used. The system transmits raw image data captured by the camera, which is in the YUYV format in our experiments. The reason we do not use compression for ROI data is that coding with compression, such as MPEG-2 compression, will create inter/intra frame dependencies and the block matching of MPEG compressed frames make the ROI integration at the client side difficult. Transmitting raw data for the ROI segmentation may relatively increase transmission bit rate, however, the tradeoff is energy can be saved without running complex compression algorithms, and latency and jitter can be reduced due to removal of the computing overhead of compression. Another gain of transmitting raw image is the packet header overhead can be avoided since the ROI data transmitted is on the pixel level and no high-level metadata of the ROI is required on the client node. Only the position and size information of the ROI is needed. To transmit this information, we append six bytes in front of each ROI data packet. Thus, the ROI data packet has only the ROI

video data plus 6 bytes. Second, to reduce the data redundancy due to high FPS when an object moves slowly, we use the following algorithm: The center of the ROI is computed after the ROI is identified. We use the velocity of the moving ROI center as the parameter for adjusting the FPS for ROI transmission, e.g., if the time difference between two frames is 1ms and the Euclidean distance traveled by the ROI center is 5 pixels, the velocity will be 5 pixels/ms. The FPS for ROI data transmission is experimentally determined based on the value of the ROI moving velocity. The maximum FPS can be achieved on our system is 15, thus the FPS can be adjusted between 1 and 15. Finally, for simplicity and to further reduce bit rate, we convert the ROI identified into a single channel gray scale image. The background frame is converted in a similar fashion. One frame of the reconstructed video is shown in Fig. 3. 4. System Implementation To verify the real-time video streaming scheme proposed, we implemented a prototype system as shown in Fig. 2.

Figure 2. The Prototype System for Testing the Proposed Real-time Video Streaming Scheme Each sensor node is a PC104 based compact system manufactured by Aaeon, which has a 533MHz VIA Mark processor. It has Four USB1.1 ports and a SDRAM-SODIMM socket for up to 512 megabytes of RAM and supports type I compact flash cards. The camera used is Logitech QuickCam Orbit AF pan/tilt web camera which connects to the PC104 system through a USB2.0 port. The camera supports video capture up to 30 FPS of 2 mega pixel video frames (1600x1200). The software is installed on a 1GB SanDisk compact flash card. The OS is Slax 7.01 and the



video processing library used is OpenCV 2.0.0. The prototype system has four sensor nodes, as shown in Fig. 2. Node A is the server node which does the video monitoring and motion detection. Node C is the client which is the sink node for the video streams. ROI segmentation is conducted on Node A and transmitted to Node C through Node B. Node D is used to introduce traffic to Node B to generate congestion on the transmission path. 5. Preliminary Results and Future Work We have completed the prototype implementation and some preliminary data results are collected. As shown on Fig. 3, the moving person is identified on the server node and the ROI (covered by the red rectangle on the right screen in Fig. 3) is transmitted to the sink node and the video stream is successfully reconstructed on the receiver.

Figure 3. ROI Identified on the Server Side and Reconstructed Video Stream on the Client Side. We plan to use this platform to do more experiments to test the performance of our proposed scheme for real-time multimedia streaming under various conditions, such as impact of changing channel conditions on packet loss, latency and jitters. We also plan to compare the proposed multimedia streaming approach with other ones using different video compression standards. References [1] I. Akyildiz, T. Melodia, and K. R. Chowdhury, “A survey of wireless multimedia sensor networks,” Computer Networks, 51:921-961, 2007. [2] R. Wagner, R, R. Nowak, and R. Baraniuk, “Distributed image compression for sensor networks using correspondence analysis and super-resolution,” in Proceedings of 2003 International Conference on Image Processing (ICIP), Vol. 1, pp. 597-600, September, 2003. [3] B. Deb, S. Bhatnagar, and B. Nath, “ReInFroM: Reliable information forwarding using multiple paths in sensor networks,” in Proceedings of 28th Annual IEEE International Conference on Local Computer Networks, pp. 406-415, Bonn, Germany, Octomber, 2003. [4] P. Guangyu and C. Chien, “Low power TDMA in

large wireless sensor networks,” in Proceedings of Military Communications Conference (MILCOM), Vol. 1, pp. 347-351, Octomber, 2001. [5] S. Misra, M. Reisslein, G. L. Xue, “A survey of multimedia streaming in wireless sensor networks,” in Communications Surveys & Tutorials, IEEE, Vol. 10 Issue 4, pp. 18-39, 2008. [6] E. Magli, M. Mancin, and L. Merello, “Low-complexity video compression for wireless sensor networks,” in Proceedings of International Conference on Multimedia and Expo (ICME ’03), pp. 585-588, Baltimore, MD, USA, July 2003. [7] A. Wyner, J. Ziv, "The rate-distortion function for source coding with side information at the decoder," IEEE Trans, Informat. Theory 22, pp. 1-10, January, 1976. [8] H. Wu and A. Abouzeid, “Energy efficient distributed JPEG 2000 image compression in multihop wireless networks,” in Proceedings of the 4th Workshop on Applications and Services in Wireless Networks (ASWN), pp. 152-160, August, 2004. [9] V. Sanchez, A. Basu and M. K. Mandal, "Prioritized Rgion of Interest Coding in JPEG2000," IEEE Trans. on CSVT, Vol. 14(9), pp. 1149-1155, September 2004.

Jian Fang is currently a doctoral student of the Department of Computer Science and Software Engineering at Auburn University. His research interests include mobile and distributed systems, target tracking, and video processing

and transmission over sensor networks. He received his bachelor degree from the University of Shanghai for Science and Technology and master degree from the University of Alabama in Huntsville. He is a student member of IEEE.

Alvin Lim is currently an associate professor of computer science and software engineering at Auburn University. He received his Ph.D. degree in computer science from University of Wisconsin at Madison in 1993. His

research interests include self-organizing sensor networks, mobile and pervasive computing, network security, wireless networks, reliable and dynamically reconfigurable distributed systems, complex distributed systems, mobile and distributed databases, distributed operating systems,



and performance measurement and analysis. He has published widely in journals and conferences in these networking and distributed systems areas. He is a subject area editor of the International Journal of Distributed Sensor Networks. His work had been supported by the National Science Foundation, the DARPA SensIT program, U.S. Air Force Research Lab and the U.S. Army.

Qing Yang is a RightNow Technologies Assistant Professor in the department of Computer Science at Montana State University. He obtained his Ph.D degree in the Computer Science and Software Engineering at Auburn University. His

current research interests lie in the areas of wireless networks, vehicular networks and network security and privacy. Qing Yang has received various awards and scholarships including the Graduate Research Scholars Program Scholarship and Vodafone-U.S. Foundation Fellowship. He has participated in various research projects supported by the National Science Foundation, Alabama EPSCoR, U.S. International Traffic in Arms Regulation, U.S. Army and USDA’s National Institute of Food and Agriculture.



Real-time Video Surveillance for Large Scenes Hanyu Liu, Chong Tang, and Shaoen Wu, School of Computing, University of Southern

Mississippi, Hattiesburg, MS 39401, USA { hanyu.liu, chong.tang, shaoen.wu }@usm.edu

1. Introduction With the advances in networking technologies, more and more Internet protocol (IP) cameras are being adopted to support a variety of requirements and applications, such as security surveillance, traffic control [1] and online courses [2]. Compared to the traditional analog closed circuit television (CCTV) cameras, IP cameras can exchange data via wired or wireless computer networks. Therefore, the captured real-time video can be easily shared with remote clients. However, constrained by optical technologies and costs, cameras can only provide limited angles of view, namely covers only partial of a scene, especially when it is large. A widely used solution to the above limitation is to deploy multiple IP cameras and generate the whole scene in separate displays or viewing windows in a large display. This intuitive solution stems from the traditional CCTV surveillance system, which connects all cameras to a central control where there are a number of displays [3]. However, two problems exist in this solution. The first is the difficulty of synchronizing the frames from different cameras on a packet-switching network [4]. The other is that the separation in display of views from different cameras provides offensive user experience and undermines the exploitation of useful detail information from those cameras [5] [6]. In this article, we briefly present a novel framework, which takes the advantages of recent advances in Computer Science and other disciplines, to address the problems mentioned above. In the rest of the article, we first describe the architecture of the system. Then, discussed is the hierarchical real-time video stitching. Finally, a novel working set oriented user interface is presented, which provides end users a more friendly and efficient way to accomplish their work with multiple IP cameras. 2. Environment The distributed system is depicted in Figure 1, Panasonic residential IP network cameras, BL-C210, are used to capture video. The cameras support H.264, MPEG-4 and Motion JPEG compression format and can stream compressed

video via RTSP [7], a popular protocol used for establishing and controlling media sessions. The embedded device consists of an x86 based AMD embedded CPU and the PC has an Intel i3 processor with 4G RAM. All those devices are connected via gigabit Ethernet or 802.11n Wi-Fi wireless network.

Figure 1. The System Architecture. All IP cameras are connected to Internet. The Embedded Device preforms stitching remotely and streaming to end users. 3. Video Stitching To provide a panorama view of a monitored scene, we introduce a Computer Vision technique, video stitching. This technique exploits the image stitching that has been investigated for decades [8] and a number of software have been released [9]. With video stitching, the video frames from different cameras covering an entire scene are combined together. In addition to the details from each camera, the user can have a panorama video of these cameras, which could support a better overall understanding the entire target scene. We implemented the video stitching based on homography [10], a concept wildly used in Computer Vision that can be viewed as an invertible transformation from one projective plane to another. Denote pa as a point p in frame (or projective plane) Fa and pb in frame Fb. The transformation can be described as follows:

Where Hba is the homography mapping matrix. Ka and Kb are the intrinsic parameter matrices of the cameras. Before calculating remapping matrix, we apply Scale-Invariant Feature Transform (SIFT) [11] to extract associating points and Random Sample Consensus (RANSAC) [12] to eliminate



outliners. To address the synchronization problem among videos from multiple cameras in complicated networking environments, we design the video stitching system with hierarchical implementation and optimization. In some cases where the networking topology is simple and stable, the videos streamed from the IP cameras can be directly fetched by the computer and then stitched for further use. Otherwise, in the cases where frame synchronization from the IP cameras cannot be guaranteed, or the target mission is critical, the embedded device is topologically deployed nearby the related cameras. Then the video is stitched first at the front-end embedded device and then delivered to the back-end users. 4. Working Set Oriented User Interface As each camera could only capture a small portion of the whole target scene, it is crucial develop a novel user interface to assist a user to manage the cameras and accelerate the processing of those captured videos. The traditional grid TV wall lacks scalability and capability to customize. We present a novel user interface that encapsulates basic video displaying and camera managing functions into an “item” [13] for each camera. An example of the interface is shown in Figure 2. Each “item” can be freely placed, resized and grouped on top of a flat panel. The user can easily build up a highly customized, full functional user interface to meet different requirements. If the user wants to stitch two videos, s/he can simply drag one “item” to the other with mouse or figures on touch screens. In addition, our user interface is also adaptive to different sizes of display. For large and high-resolution display, multiple “items” can be added and stitched. On the other hand, for small displays or hand hold devices, one “item” can be initialized and support basic video displaying function. 5. Conclusion and Future Works In our preliminary test, our video stitching on the embedded device works an average of 12 FPS output with two 640x480 video inputs. The user interface can easily boost a 30 FPS output with four 640x480 video inputs via RTSP on a typical IPv4 network. In addition, we conducted a small-scale pilot test and simulated both online course and parking area surveillance scenarios; all test takers respond positively. In the next step, it would be interesting and meaningful to test the performance in practice and collect formal reposes from professional users.

Figure 2. User interface with a floating control panel and three video “items”. The bottom-right item is generated from top two. The user could freely manipulate and manage those items. References [1]. Industrial Video & Control (IV&C) official website,

http://www.ivcco.com. [2]. Netcom Learning official website,

http://www.netcomlearning.com. [3]. CCTV SecurityPros official website,

http://www.cctvsecuritypros.com. [4]. James F. Kurose and Keith W. Ross, Computer

Networking: A Top-Down Approach (5th Edition), Addison Wesley, 2009.

[5]. Trouvain, B, Schlick, C. “A study of audio and visual context switch indicators in a mutirobot navigation task”, IEEE International Conference on System, Man and Cybernetics, 2004.

[6]. Cinvertino, G.; Chen, J.; Yost, B.; Ryu, Y. and North, C. “Exploring context switching and cognition in dual-view coordinated visualizations”, International Conference of Coordinated and Multiple Views in Exploratory Visualization, 2003.

[7]. RFC 2326, Real Time Streaming Protocol, IETF, 1998.

[8]. Szeliski, Richard, “Image Alignment and Stitching”, 2005.

[9]. Hugin official site, http://hugin.sourceforge.net. [10]. Richard Hartley and Andrew Zisserman, “Multiple

View Geometry in Computer Vision”, Cambridge University Press, 2003.

[11]. Lowe, David G. “Object recognition from local scale-invariant features”, Proceedings of the International Conference of Computer Vision, 1999.

[12]. Martin A. Fischler and Robert C. Bolles, “Random Smaple Consensus”, Comm. Of the ACM, 1981.

[13]. Andrew Bragdon, Robert Zeleznik, Steven P. Reiss, Suman Karumuri, William Cheung, Joshua Kaplan, Christipher Coleman, Ferdi ANdepitra, and Jeseph J. LaViola Jr. “Code Bubbles, A Working Set-based Interface for Code Understanding and Maintanence”, International Conference on Human Factors in Computing System, 2010.



Hanyu Liu received his B.S. degree in Computer Science from Tianjin University (TJU), China 2009. Currently he is a Ph.D. student at School of Computing, the University of Southern Mississippi, US. His research interests include computer vision, interaction

and computer graphics.

Chong Tang received Bachelor’s degree from Xi’an Institute of Post Telecommunications. After that, he has worked as a Software Engineer in NEC. He is presently pursuing his master degree in the School of Computing at University

of Southern Mississippi. His current researches focus on Rate Adaptation, Media Access and Cognitive Radio.

Shaoen Wu received a Ph.D. in Computer Science in 2008 from Auburn University. He is presently an Assistant Professor of School of Computing at the University of Southern Mississippi. He has held a researcher scientist position at ADTRAN Inc. and a senior software engineer position at Bell

Laboratories. His current research is in the areas of wireless networking, mobile computing and cyber security. He is a recipient of Best Paper Award of IEEE ISCC 2008 and ANSS 2011. He has served on the chairs and the committees of several conferences and as editor for several journals. More information is available at http://www.cs.usm.edu/~swu.



Cross Layer QoS Provisioning for Cooperative Video Transmission Kun Hua, Department of Electrical and Computer Engineering, Lawrence Tech

University, Southfield, MI Yin Wang, Department of Mathematics and Computer Science, Lawrence Technological

University

Wei Wang, South Dakota State University Shaowen Wu, University of Southern Mississippi

([email protected], [email protected] , [email protected], [email protected])

1. Introduction Real time wireless Video transmission, such as online gaming and video phone calls, is fast emerging, under the prevalence of 4G wireless network and smart phones [1],[2]. It demands quality of service (QoS) guarantees such as throughput, delay bounds and battery life (energy cost) for real time transmission. However, transmitting video over mobile phone with desired throughput and acceptable delay and losses is particularly challenging since the wireless channels are subject to multipath fading, and therefore the channel condition may be changing from frame to frame. Due to its high spectral efficiency (high data rates) and good diversity characteristics (low bit error rates (BER)), cross layer Cooperative diversity is quite well-suited to satisfy the above mentioned requirements. Instead of impractical antenna array of traditional MIMO techniques, cooperative communications form a virtual antenna array through multiple nodes. Several relays duplicates the video frames from the source and combines them to obtain a more reliable estimate of the transmitted signal at the receiver as well as higher data rates and low interference [3]. The cooperation layers include the physical, MAC and network layer. With different channel environments, adaptive convolutional coding rate and constraint length, as well as the number of cooperative relay paths and hops in correlated channels can be chosen in physical layaer for different QoS requirements, e.g., BER, throughput, Energy allocation or delay. Similarly, the minimum contention window size, MAC-frame size, retransmission times and data rate are also adjusted in MAC layer. Furthermore, the optimal routing strategy will be selected in network layer. In different channel condition, the cross-layer control parameters from multiple layers could have different level impact or sensitivities on the network performance such as throughput and delay. It is important to identify key control parameters to reduce the cross-layer control and optimization

overheads. In this letter, we propose a parameter adaptive cross-layer cooperative video transmission architecture for QoS provisioning. A Choquet-based learning algorithm for key parameters selection is applied for searching important parameters. The significance and interaction among these cross-layer parameters are evaluated through this algorithm. The proposed cross layer cooperative scheme exploits spatial diversity and time diversity simultaneously without an expansion of bandwidth or a power increase. It requires additional burden of video stream encoding and decoding in the relays and base/mobile stations. 2. Proposed Cross-Layer Design Frame works At physical layer, the constraint length of the Convolutional codes K and code rate R are two factors which are adjustable for QoS provisioning. Larger K and smaller R will protect the data stream as expected (better BER performance), but will cause the larger delay time. K and R can be calculated with distance d. Path number M and Hop number N are other two important issues for QoS. Lager M and N will promise the quality but Energy allocation and delay time will be an issue to consider. With different channel environment, adaptive K,R,M,N can be chosen for different QoS requirements for BER, throughput, Energy allocation or delay. The receiver employs Viterbi decoding with hard decision. The proposed system allows the receiver to exploit the additional time diversity as well as the spatial diversity inherent in relay systems [4]. MAC layer parameters include the data rate S, contention window size W, MAC-frame size L, and retransmission times T, Customer number C, etc. For example, under a severe channel condition, the BER will be reduced by decreasing the transmission rate S, similarly, the frame error probability will be reduced by decreasing the frame size L, and seems the throughput could be improved. But on the other hand, the throughput



will be decreased with a lower transmission rate S and extra overhead cost for a smaller frame size L and contention window size W. Therefore there are tradeoffs between delay bounds and throughput [5], [6]. Retransmission time T is relative to the performance of delay and Customer number C is proportional to the energy cost. The network throughput is obtained under different S,W,L,T and C. Obviously, to achieve desired throughput, delay bounds and energy efficiency, an international study need to be done for both parameters {S,W,L,T,C} at MAC layer and {K,R,M,N} at physical layer. Routing is a way of cooperation among nodes. In wireless networking, cooperative diversity is being widely exploited as a spatial diversity technique in the physical layer. Therefore, the communication between a pair of nodes of a hop along a selected routing path with cooperative diversity possibly consists of more than one distinguished links. Consequently, the cooperative diversity at lower layer impacts the routing path selection in the networking layer. Therefore a cross-layer approach has to be considered for routing with cooperative diversity, called Cooperative Routing, which has been a hot topic in recent years. In this letter, we propose a comprehensive cross-layer link cost with cooperative transmission to achieve a combination of optimization goals on throughput, delay and energy. This link cost is proportional to the required transmission energy, the transmission delay, and throughput and inversely proportional to the transmission rate. The required transmission energy is a function of transmission distance, with the decomposed of each link into several cooperative transmissions with respective SNRs and power attenuation factors. It should be noted that delay of a specific link cannot simply take the inverse of its bit rate because it involves possible queuing delay and retransmission. Eventually, the cooperative routing goal is to identify a path with the minimum path cost. 3. Multi-Tier Analysis for Cross-Layer Design for Cooperative Wireless In wireless environment, channel noise and interferences are burst, time varying and location dependent. It is hard to model the real time video network due to the time varying and uncertainty of cross layer control parameters. Unfortunately, all traditional cross-layer approaches usually are

exhaustively looking for all possible strategies. In particular, when there are a number of parameters from multiple equivalent layers, the optimization process becomes difficult due to the associated complexity (more control, information exchange, and computation overheads). In the cross-layer design, the parameters at each equivalent layer can impact the overall system performance. There are trade-offs or dependences among these parameters. It is important to discover these interactions and determine optimized cross-layer control strategies. At the network layer, routing algorithms should find the highest quality paths to deliver data in multi-hop scenarios. At the data link layer, maximum ARQs (Automatic Repeat Requests) and adaptive frame size help the transmission system maintain a low packet loss rate. Adaptive modulation and power control lead to trade-offs between energy consumption and bit error rate at the physical layer. However, the control of these parameters from multiple layers increases the complexity, which may hinder a wide application of the proposed architecture. There is a strong need to select a few key parameters to optimize the data transmission performance in an integrated manner. 4. Choquet-based Learning for Significant Parameters Traditional linear measurement might be inappropriate under such complex environment. We propose a nonlinear statistical fuzzy measure model based on non-additive integrals for the key control parameters and their interactions without establishing an exact mathematical model [7]. The distinguishing feature of this model is that the interaction among system parameters toward the performance metrics of interest can be properly measured through the nonlinear integral such as Choquet integral [8]. The Choquet-based learning algorithm for key parameters selection is described as following. In this algorithm, the significance and interaction among these cross-layer parameters are evaluated. Choquet-based learning algorithm for cross-layer

key parameter selection Input : Observed data at physical, MAC and Network layer. Output : Optimal link cost and Cross-layer parameters significance value Step 1: Input the network training data, coding and decoding the unknown parameters;



Step 2: Population and chromosomes. Each chromosome denotes a possible interaction among cross-layer parameters { K,R,M,N } at physical layer , { S,W,L,T,C } at MAC layer and optional routing paths at Networking layer, which can be represented as a gene string; Step 3: The function of evaluating the performance of cross-layer -- Chromosomes’ fitness. The fitness function is determined by the target of the optimization; Step 4: Parent selection and producing new chromosomes. Step 5: Renew population. Step 6: Stop the controller and determine significant parameters based on the important index. Output the corresponding values of parameters and choose important index. The minimum link cost is achieved with the optimal parameters. References [1] A.A. Khalek, Z. Dawy, “Energy-Efficient Cooperative Video Distribution with Statistical QoS Provisions over Wireless Networks”, IEEE Transactions On Mobile Computing, May 2011 [2] Z. Chen, D. Wu, “Rate-Distortion Optimized Cross-layer Rate Control in Wireless Video Communication”, IEEE Transactions on Circuits and Systems for Video Technology. [3] Kun Hua, Shaoen Wu, Wei Wang, Honggang Wang, “Cognitive Cross-layer Design with QoS Provisioning for Cooperative Wireless Networking”, 44th Annual Simulation Symposium (ANSS2011), Boston, USA, Apr. 2011. [4] Kun Hua, Won Mee Jang and Lim Nguyen “Convolution Coding in Cooperative Relay Systems with Spread Spectrum” IEEE Fourth International Conference on Networked Computing and Advance Information Management,2008. [5] D. Qiao, S. Choi, and K. Shin. Goodput Analysis and Link Adaptation for IEEE 802.11a Wireless LANs [J]. IEEE Trans. on Mobile Computing. 2002, vol. 1, no. 4, pp. 278–292, 2002. [6] D. Skordoulis, Q. Ni and H. H. Chen. IEEE 802.11n MAC frame aggregation mechanisms for next-generation high-throughput WLANS [J]. Wireless Communications, 2008, 15(1): 40-47. [7] Fang, H., Rizzo, M. L., Wang, H., Espy, K. A., “A new nonlinear classifier with a penalized signed fuzzy measure using effective genetic algorithm," Pattern Recognition (Elsevier), 2010. [8] Z. Wang. A new model of nonlinear multiregressions by projection pursuit based on generalized Choquet integrals [J]. IEEE International Conference on Fuzzy Systems, 2002: 1240–1244

Kun Hua is an Assistant Professor of Electrical and Computer Engineering Department, Lawrence Technological University, Southfield, MI, USA since 2010. He received the B.Sc. (with First Class Honors) and M.Sc. degrees from Electrical and Computer

Engineering, Xi’an Jiaotong University, China, in 1999 and 2004 respectively. He earned his PhD degree in computer and electronic engineering from the University of Nebraska Lincoln, Nebraska, USA in 2008. He continued his research in the University of Nebraska Lincoln as a Post-doctorial researcher in 2009. His current research interests are in the areas of wireless communication and multimedia signal processing. Prof. Hua is a senior member of IEEE. He served as an Associate Editor of Wiley Security and Communication Networks Journal.

Yin Wang received the Ph.D. degree from Department of Computer Science, University of Kentucky, in 2010. He is currently an Assistant Professor in Department of Mathematics and Computer Science, Lawrence Technological University.

His major research interests are scientific high performance computing, computer simulation and data mining in social networks.

Wei Wang is an Assistant Professor with the department of Electrical Engineering and Computer Science, South Dakota State University, Brookings, SD, USA. He joined South Dakota State University in January 2010. He received

his B.S. degree in Computer and Information Engineering from Xian Jiaotong University, China, 2002, and M.S. degree in Information and Communication Systems from Xian Jiaotong University, China, 2005. He received his Ph.D. degree in Computer Engineering from University



of Nebraska - Lincoln, USA, 2009. His major research interests include wireless sensor networks, multimedia computing, information security, and embedded systems. He is the winner of the Best Paper Award in IEEE Wireless Communications and Networking Conference (WCNC), Las Vegas, NV, 2008. He served as an Associate Editor of Wiley Security and Communication Networks Journal, as the Program Chair of IWMMN 2010, and as the vice chair of NGN track in IEEE ICCT 2011.

Shaoen Wu received a Ph.D. in Computer Science in 2008 from Auburn University. He is presently an Assistant Professor of School of Computing at the University of Southern Mississippi. He has held a researcher scientist position at ADTRAN Inc. and a senior software engineer

position at Bell Laboratories. His current research is in the areas of wireless networking, mobile computing and cyber security. He is a recipient of Best Paper Award of IEEE ISCC 2008 and ANSS 2011. He has served on the chairs and the committees of several conferences and as editor for several journals. More information is available at http://www.cs.usm.edu/~swu



Cross-layer Based P2P Scheduling in Wireless Multimedia Networks Haiyan Luo, Song Ci, Dalei Wu, Wei An, Department of Computer and Electronics

Engineering, University of Nebraska-Lincoln, Lincoln, NE 68588, USA [email protected], [email protected], [email protected], [email protected]

1. Introduction Over the past few years, P2P multimedia applications have enjoyed tremendous growth. However, huge challenges still exist before its wide deployment in wireless networks. The majority of the current research and commercial products of P2P video applications, which are based on the overlay network architecture, cannot achieve satisfactory results when directly applied to wireless networks. They explicitly or implicitly assume that the network layers below the P2P overlay networks are in perfect condition by either ignoring the lower layers or assuming an error-free Internet environment [1]. However, in wireless networks, due to the time-varying channel characteristics and high heterogeneity, this assumption greatly affects the user-perceived video quality at the receiver end. Generally, wireless P2P multimedia networks have the characteristics of time-varying channel conditions, higher network heterogeneity, intricate interactions among different network layers, and the tendency of more peers joining and leaving activities [2], [3]. To meet these challenges, we present our current research project on wireless P2P multimedia applications to achieve significantly-improved user-perceived video quality while the computational overhead is evenly distributed to each peer node. To achieve this, a cross-layer based distortion-delay framework is implemented as the integral and essential part of the proposed distributed P2P scheduling algorithm at each peer node, where functions provided by different network layers can be holistically optimized for the P2P video applications in wireless networks. The performance impact from various network parameters residing at different layers of the P2P wireless environment is jointly evaluated to effectively improve the overall video performance. In addition, the joint optimization is employed in a distributed fashion, decreasing the computational complexity at each node and increasing the possibility of deployment. The experimental results demonstrate that significant user-perceived video quality improvement can be achieved from the perspective of end users. In this article, we briefly discuss how different network layers interact with each other to affect the

overall P2P video performance and how they can be jointly optimized into a distributed scheduling algorithm to significantly improve the overall video performance for P2P end users over wireless multimedia networks. 2. Interactions of network functions and the cross-layer design In wireless P2P multimedia networks, the interactions of network functions residing at different network layers jointly affects the user-perceived video quality, which can be taken into consideration in the design of the scheduling algorithm. For instance, the fed-back channel quality affects the choice of Modulation and Channel Coding (MCC) scheme, while the video encoding determines the packet length by using the appropriate encoding parameters such as quantization step size (QP) q and intra- or inter prediction mode. Then, the used MCC scheme and packet length jointly determine the packet loss rate ρ. Further, the transmission rate r and ARQ are affected by the MCC scheme and packet loss rate ρ, respectively. Also, the transmission delay can be achieved by the joint impact of transmission rate r and retransmission, while the estimated video distortion is jointly affected by packet loss rate ρ and video encoding. At last, the transmission delay and estimated video distortion E[D] interact with each other to achieve the final video quality. Thus, by fine-tuning the system parameters residing at different layers in a systematic way during scheduling to achieve the optimal combination of parameter values, video performance can be greatly improved. Based on this observation, we propose a cross-layer based optimized scheduling algorithm for wireless P2P multimedia networks, where the core is the distortion-delay cross-layer optimization that is distributed to and performed at each neighboring node. As shown in Figure 1, we consider video encoding, ARQ, and link adaptation. The scheduler is able to adjust the network behavior by equipping it with key system parameters of each network layer, such as the quantization step size at the application layer, the retransmission at the data link layer, and the modulation and channel coding (MCC) scheme at the physical layer. Here, the cross-layer optimization algorithm is distributed



and implemented at each network node [4].

Figure 1. The architecture of the proposed quality-driven distributed scheduling for P2P multimedia applications over wireless networks. 3. Cross-layer optimized P2P scheduling Overall, the proposed optimized P2P scheduling works as follows. 1) When a given node detects a missing segment,

it initiates a request to the neighboring nodes with a delay requirement.

2) The neighboring nodes either perform cross-layer optimization to determine the best network parameter values of different layers, or just forward the request to the other neighboring nodes when the requested segment is not available.

3) The given node gathers the responses from the neighboring nodes and chooses the best node with the highest video quality from which to fetch the missing segment.

4) The data connection is established between the given node and the chosen neighboring node to transmit and receive the missing segment.

For the proposed scheduling algorithm, the core is the distributed cross-layer optimization as explained in step 3), which can be formulated into a delay-distortion framework, where the expected video distortion is calculated by using the ROPE algorithm [5], [6], while the expected delay can be formulated by jointly considering the video encoding, queuing delay and the channel modulation [4]. In this way, network functions are jointly considered and optimized in a holistic way. Further, the problem can be effectively solved by using dynamic programming [7]. 4. Simulation results We randomly distribute 80 nodes in an area of

1000m × 1000m and set the transmission radius of every node to be 150m. Further, we allow the nodes to randomly join and leave the P2P overlay networks. The proposed scheduling algorithm runs at every P2P node. At the streaming server, we use the QCIF (176 × 144) [8] sequence “Foreman” as the video streaming content, which is coded at a frame rate of 30 frames/second. Each communication link has an average SNR γ. The propagation delay on each link is set to 10µs, while the average bandwidth is randomly set to 150K, 200K and 250K (symbols/second) to simulate a heterogeneous network environment. To avoid prediction error propagation, a 10% macroblock level intra-refreshment is used. The motion vector of a missing MB is estimated as the median of the motion vectors of the nearest three MBs in the preceding row. If that row is also lost, the estimated motion vector is then set equal to zero. The pixels in the previous frame, pointed by the estimated motion vector, are used to replace the missing pixels in the current frame. At the streaming server, the values of QPs are chosen from 1-50. Different AMC schemes are available during the optimization [6], [9].

(a) (b)

Figure 2. User-perceived video quality improvement of the proposed cross-layer based P2P scheduling algorithm in wireless multimedia networks. We present the visual quality difference in Figure 2. In this figure, (a) is a video frame that uses the existing P2P scheduling algorithm while (b) is the same video frame achieved by using proposed distributed P2P scheduling algorithm. It can be concluded that the existing system experiences more packet losses and thus more frequent use of error concealment, leading to degraded video quality. Our experiments also show that this user-perceived performance enhancement can be even more significant when the playback deadline is more stringent or/and the channel quality is worse, thus, the proposed scheduling algorithm works even better in wireless environments with limited resources. References [1] E. Setton and B. Girod, Peer-to-Peer Video



Streaming. Springer Science+Business Media, LLC, 2000. [2] P. Si, F. R. Yu, H. Ji, and V. C. M. Leung, “Distributed sender scheduling for multimedia transmission in wireless mobile peer-to-peer networks,” IEEE Trans. Wireless Commun., vol. 8, no. 9, pp. 4594–4603, Sep. 2009 [3] M. Conti, J. Crowcroft, F. Delmastro, and A. Passarella, “P2P support for Group-Communication Applications: a Cross-Layer Approach for MANET Environments,” IEEE INFOCOM, pp. 23–29, Apr. 2006 [4] H. Luo, S. Ci, and D. Wu, “A cross-layer optimized distributed scheduling algorithm for peer-to-peer video streaming over multi-hop wireless mesh networks,” IEEE SECON, pp. 1–9, Jun. 2009 [5] R. Zhang, S. L. Regunathan, and K. Rose, “Video Coding with Optimal Inter/Intra-Mode Switching for Packet Loss Resilience,” IEEE J. Select. Areas Commun., vol. 18, no. 6, pp. 966–976, Jun. 2000 [6] Q. Liu, S. Zhou, and G. Giannakis, “Cross-layer combining of adaptive modulation and coding with truncated ARQ over wireless links,” IEEE Trans. Wireless Commun., vol. 3, no. 5, pp. 1746–1755, Sep. 2004 [7] G. M. Schuster and A. K. Katsaggelos, Rate-Distortion Based Video Compression: Optimal Video Frame Compression and Object Boundary Encoding. Norwell, MA: Kluwer, 1997. [8] “Draft ITU-T Recommendation and Final Draft International Standard of Joint Video Specification (ITU-T Rec. H.264łISO/IEC 14496- 10 AVC),” ftp://ftp.imtc-files.org/jvt-experts/2003-03-Pattaya/JVT-G50r1.zip, May 2003 [9] H. Luo, S. Ci, D. Wu, H. Tang, Quality-driven Cross-layer Optimized Video Delivery over LTE, IEEE Communications Magazine, vol. 48, no. 2, pp. 102-109, Feb. 2010

Haiyan Luo (S’09) received his B.S. degree from Dalian Jiaotong University (formerly Dalian Railway Institute), M.S. degree from Dalian University of Technology, Dalian, China and Ph.D. from University of Nebraska-Lincoln, Lincoln,

USA. He is currently associated with Cisco Systems. Dr. Luo’s research interests include multimedia networks and distributed system. He has served as a referee for IEEE Transactions on Wireless Communication, IEEE Transactions on Vehicular Technology, IEEE Transactions on Multimedia, IEEE Transactions on Network and Service Management and TPC for several conferences such as IEEE GLOBECOM and IEEE CSE and so on.

Song Ci (S’98–M’02–SM’06) is an Associate Professor of Computer and Electronics Engineering at the University of Nebraska-Lincoln. He is the Director of the Intelligent Ubiquitous Computing Lab (iUbiComp Lab) and holds a courtesy appointment of

UNL Ph.D. in the Biomedical Engineering Program. He is also affiliated with Nebraska Biomechanics Core Facility at the University of Nebraska at Omaha and Center for Advanced Surgical Technology (CAST) at University of Nebraska Medical Center, Omaha, NE. His research interests include dynamic complex system modeling and optimization, green computing and power management, dynamically reconfigurable embedded system, content-aware quality-driven cross-layer optimized multimedia over wireless, cognitive network management and service-oriented architecture, and cyber-enable e-healthcare. Prof. Ci is an Associate Editor of the IEEE Transactions on Vehicular Technology (TVT) and serves as a Guest Editor of the IEEE Transactions on Multimedia Special Issue on Quality-Driven Cross-Layer Design for Multimedia Communications; Guest Editor of the IEEE Network Magazine Special Issue on Wireless Mesh Networks: Applications, Architectures and Protocols; and Associate Editor of four other international journals. He serves as the TPC co-chair of IEEE ICCCN 2009 and as the TPC member for numerous conferences. He is the Vice Chair of the Computer Society of the IEEE Nebraska Section and is a member of the ACM and the ASHRAE. He is the SIG chair of the Quality-Driven Cross-Layer Design of the Multimedia Communications Technical Committee (MMTC) of the IEEE Communications Society (COMSOC).

Dalei Wu (S’05) received his B.S. and M.Eng. degrees in electrical engineering from Shandong University, Jinan, China, in 2001 and 2004, respectively, and Ph.D. degree in computer engineering from the University of Nebraska-



Lincoln in 2010. His research interests include wireless multimedia networking, green computing, cyber-physical systems, complex dynamic system modeling and optimization.

Wei AN received his Master degree in Applied Mathematics from School of Science, East China University of Science and Technology, Shanghai, China, in 2008. He is currently pursuing his Ph.D.

degree at school of Information Science and Engineering, East China University of Science and Technology. Now he is under the advising of Dr Song Ci in the Department of Computer & Electronics Engineering, University of Nebraska-Lincoln, from Sept. 2009 as a Joint Ph.D. Candidate. His research interests include wireless sensor network and complex systems.



NEWS

MMTC Virtual Lab Tour Talk 3DRPC IG is glad to launch the 1st conference call based program "Research Orientation", which intends to introduce the research facilities of and research projects conducted by IG key members to other key members and interested individuals in MMTC, to faciliate potential research collaboration and the sharing of experimental findings. Pradeep Atrey (http://www.acs.uwinnipeg.ca/pkatrey/) along with Xenophon Zabulis ( http://www.ics.forth.gr/~zabulis/) will be coordinating these talks. The details are as follows: Date and Time: September 27, 2011 at 4:00 p.m. Central European Time Speaker: Dr Ing. Carl James Debono, Senior Lecturer and Deputy Dean Department of Communications and Computer Engineering University of Malta, Msida MSD 2080 Malta Email: [email protected] Homepage: http://www.eng.um.edu.mt/~cjdebo Research Group Link: The Multimedia and Wireless Communication Systems (MMWCS) Research Group: http://www.eng.um.edu.mt/~cjdebo/ The slide can be downloaded from http://www.eng.um.edu.mt/~cjdebo/labtour.pdf . For any question, please contact MMTC 3DRPC IG chair: Dr. Irene Cheng ([email protected])



E-LELLER EDITORIAL BOARD DIRECTOR CO-DIRECTOR Chonggang Wang Kai Yang InterDigital Communications Bell Labs, Alcatel-Lucent USA USA EDITOR Mischa Dohler Takahiro Hara CTTC Osaka University Spain Japan Kyungtae Kim Vijay Subramanian NEC Laboratories America Hamilton Institute USA Ireland Jian Tan Weiyi Zhang IBM T. J. Watson North Dakota State University USA USA Xiaoqing Zhu Cisco USA

MMTC OFFICERS

CHAIR Haohong Wang TCL Corporation USA VICE CHAIRS Madjid Merabti Bin Wei Liverpool John Moores University AT&T Labs Research UK USA Jianwei Huang The Chinese University of Hong Kong China

Documents

E-LETTER - Multimedia Communications Technical Committeemmc.committees.comsoc.org/files/2016/04/E-Letter-September11.pdf · Multimedia streaming in wireless multimedia sensor networks