Energy Efficient VoIP Communication Using WSN Clustering Approach

8/9/2019 Energy Efficient VoIP Communication Using WSN Clustering Approach

1/6

Energy Efficient VoIP Communication Using WSN

Clustering Approach

Manish TembhurkarStudent IV Sem M.E.(Mobile Technology)

Department of Computer Science & Engineering

G.H.Raisoni College of Engineering

Nagpur, India

[email protected]

[email protected]

Dr. Latesh MalikHead of the Department



Nagpur, India

[email protected]@rediffmail.com

Nekita ChavhanAssociate Professor



Nagpur, [email protected]

[email protected]

Abstract — VoIP refers to communications services — voice

and/or voice-messaging applications — that are transported via

the Internet, rather than the public switched telephone network

(PSTN). Rate of evolution of mobile services (such as data

transfer service (GPRS, EDGE, 3G, and 4G LTE), audio & video

player service, camera with higher resolution, GPS) is much

higher than that of energy resources or energy conservation. The

wide deployment of Voice-over-IP (VoIP) over IEEE 802.11

wireless LAN causes higher rate of energy consumption which

is a major issue both in wireless sensor network as well as in

mobile networks. Researchers have found out several solutions to

overcome this issue such as Greencall Algorithm, algorithm

through trace-driven simulations as well as experiments on

commodity hardware/software [Energy-Efficient VoIP over

Wireless LANs]. More efficient & effective solution is needed.

Keywords — Voice over IP, Wireless Sensor Network, Energy

Efficiency, Clustering Approach, Mobile Communication, Android

I. I NTRODUCTION

Mobile services are evolving with much higher rate than

that of energy resources or energy conservation which causes

energy consumption as a major issue not only in wireless

sensor network but in mobile networks. The common factor between these two networks is that they have limited resources

in terms of energy and power.

We cannot ignore this need and be stationary at one place

for continuous power source for our devices. The issue of low

battery life [1] in mobile devices such as mobile phones andlaptops, cause a major concern in certain scenarios such as

war, military applications, medical emergencies.

In previous works, however, assume-stations were always

awake during a call till in 2005, the Wi-Fi Alliance proposed a

power saving mode extension which allows stations to retrieve

packets from the Access Point (AP) in any mode, at any time.

This helps to conserve energy by staying in sleep mode during

VoIP call over WLAN (Wireless Local Area Network)

(Greencall algorithm [2]: reduces considerable energy

consumption using PSM (Power save mode)). But a totally

new approach is needed.

Nowadays mobile phones with evolutionary technologies

(such as 3G, 4G LTE) are dominating the market of cellularcommunication systems [1], [3], [4].

Figure 1: Power consumption for different phone’s services

normalized to the power consumption needed for downloadingdata using HSDPA.

0

20

40

60

80

100

120

N o r

m a l i s e d P o w e r C o n s u m p t i o n ( % )

[Services]

978-1-4673-4463-0/13/$31.00 ©2013 IEEE

The 8th International Conference onComputer Science & Education (ICCSE 2013)April 26-28, 2013. Colombo, Sri Lanka

234

SaB2.8


2/6

These phones have been provided with better hardware and

are becoming more powerful day by day. Power consumption

as the biggest consequence with these features does not match

well with the evolution of mobile terminals which tend to have

less room available for the battery in order to accommodate

additional components and technologies. Power consumptions

by various services are shown in figure (1) [3].

A. Energy Consevation

Our objective is to study various technologies to reduce power consumption available in other networks and

implement them more effectively in the VoIP network. Higher

data rate needs more power [3], [4] which makes energy

consumption as a major issue both in wireless sensor network

as well as in mobile networks. VoIP service consumes the

considerable amount of battery as compared to other services.

TABLE I. POWER CONSUMPTION FOR A NOKIA N95 IN DIFFERENTSCENARIOS OF VOICE SERVICE

Scenario GSM UMTS

Receiving a voice call 612.7 mW 1224.3 mW

Making a voice call 683.6 mW 1265.7 mW

Idle mode 15.1 mW 25.3 mW

TABLE II. EXAMPLE OF ENERGY SAVING BY USING GSM INSTEAD OFUMTS FOR DIFFERENT SCENARIOS

Scenario Time [hour Energy saved [J]

Idle mode 8 220

Making voice calls 1 2095

TABLE III. E NERGY COMPARISON USING 2G ALONE, 3G ALONE AND THEINTELLIGENT SWITCHING BETWEEN THE NETWORKS

Service 2G 3G Switching

50 SMS of 100 bytes 90 J 110 J 90 J

100 Mbytes downloading 10006.2 J 3512.1 J 3512.1 J

5 hours of voice calls 12304.8 J 22782.6 J 12304.8 J

50 handoffs 245 J

TOTAL 22401.0 J 26404.7 J 16151.9 J

B. VoIP Call Rauting

Although routing algorithms (such as Location basedRouting & data Centric Routing Algorithms) [6] works more

effectively in wireless sensor networks (WSN) to achieve primary goal of Energy Conservation along with data routing.Clustering approach or Tree based approach is followed formore energy efficient routing in WSN. VoIP networks are verymuch similar to wireless networks in various aspects as follows

Wireless Infrastructure

Mobile nodes

All nodes reports to the centralized entity (BaseStation or ISP)

Major concern: Limited Battery life

The only difference is that VoIP networks are based on theIP network. Hence the prior concern and the goal will be to

implement the WSN Energy efficient routing approach into theVoIP network infrastructure to conserve considerable amountof battery life.

II. PROPOSED METHODOLOGY

A. Proposed Model

The existing energy efficient model for the VoIP networkshows the considerable improvement in one or more objective,to suite the specific application. Still a lot of work is needed to

be done on energy efficient model in terms of low latency, realtime transmission, quality of voice, clustering overhead,continuous packet delivery and reduced data fusion cost.

This paper is to consider all the factors for energy efficientVoIP communication model. The proposed model involvesfollowing steps:

Register all the mobile devices (mobile phones andlaptops) in the network with their respective

parameters (such as IP address, Port number, battery

status remaining, Receiving signal status)

Establishing successful VoIP communication between

mobile devices

Clustering based on the algorithm

Much improved and reliable cluster head selectionthrough RSS (Received Signal Strength) value.

Registering mobile devices to their respective clusterHeads.

Alternate CH (Cluster Head) selection for continuous packet delivery.

Figure 1. VoIP server application module.

235

SaB2.8


3/6

Figure 2. VoIP server application module showing the battery status as

Charging.

Figure 3. VoIP server application module showing the remaining batterystatus (in seconds).

Figure 4. VoIP server application module with Clustering feature (ON)

III. BACKGROUND A ND R ELATED WORK

A. VoIP network

Voice over IP network [6] is the service to provide voicecommunication between two hosts in the network. In the

wireless network, the VoIP device converts the dialed number

into the network data packets which are transmitted over theradio waves to the wireless access point or other such wireless

receivers as shown in Figure (1).

Figure 1: Voice over Internet Protocol

VoIP network has various features along withrequirements as well.

VoIP applications do not require high throughput but

cannot allow jitter as such applications are vulnerable

to delays which may directly affect the voice quality.

Voice requires quality of service in terms of low

latency, low packet loss, low jitter and high

availability. Conversely, most of the power saving

techniques & mechanisms do transactions with the

latency and availability.

VoIP wireless phones require support for seamless

roaming capabilities to enable user mobility.

Security in terms of prevention of denial of service

attacks and eavesdropping is a must for the wireless

media. Power consumption relies on the complexityof the algorithms.

VoIP is an isochronous traffic stream with a

packetization rate of 20ms (G.711 codec).

B. VoIP Devices

VoIP application can be implemented over WiFi

(IEEE802.11) and WiMax (IEEE 802.16) networks in

infrastructure mode or multihop mesh mode. A WiFi VoIP

(VoWiFi) device is similar in structure to a cell phone and

consists of the following basic components: processor,

speaker, microphone, numeric keypad and function keys,

236

SaB2.8


4/6

lithium ion battery, screen (LCD), antenna and memory. The

WiFi VoIP device uses a different radio frequency than the

cell phone.

The chief advantages of VoIP devices are:

Low Cost: The network is less expensive to installand maintain. Calls can be placed across the world

practically free as compared to usual landline or cell

phone services.

At the enterprise level, the person need not be tied to

the desk to receive calls. Allows the flexibility to

roam while maintaining the low cost. This is

especially useful in the enterprise environments such

as the healthcare industry where the doctors and

caregivers are constantly on the move in the hospital.

VoIP offers more convergence with existing datacentric technologies.

Provides coverage where there may be poor or

unavailable cell coverage.

C. Energy conservation issues in VoIP network

Comparative research on VoIP network & PSTN network

[7] shows that when the power consumption of VoIP network

is compared with an equivalent PBX, VoIP consumes large

amount of power unless an efficient power saving scheme is

provided.

Mostly preferred Android devices have very pleasing

services, functions, and features but it unfortunately drains a lot

of energy from the battery [1] as shown in figure (2). It results

in limited battery life. These devices run lots of services in the

background along with VoIP service. Study shows higher data

rate (3G) is responsible for more energy consumption [3].

a) Elements of power consumption

The power consumed by a communicating device can be

factored into following elements [8], [9], [10]:

Transmission: This accounts for the energy spent in data

packet transmission.

Reception: This accounts for the energy spent by a node in

data reception.

Idle listening : Refers to the power consumed when the radio

of the node is waiting to receive potential packets but the

media is idle.

Overhearing: Refers to the power used by a node when it is

receiving packets on the media meant for another destination.

Control Overhead : This factor accounts for the power used to

send and receive control packets.

Reliability [8]: This element pertains to energy consumed in

meeting the protocol reliability requirements, which is the data

retransmissions caused from the lossy media, collisions and

mobility.

Turnaround time [10]: This is the time required to switch

modes from transmit to receive and vice versa.

Different power saving techniques attempts to minimize

the energy consumed by the combinations of these factors.

Various schemes were developed and implemented

successfully such as Opportunistic scheduler [4] and Greencall

algorithm [2] based on the Power Saving Mode (PSM).

Though the quality of the performance is maintained, it does

not satisfy the requirement of saving sufficient battery power

[3].

D. Energy efficient routing in WSN

As energy efficiency is the major concern in wireless

sensor networks, related research was already taken place in

WSN. Larger size data transmission consumes more battery

power of the sensor node. Hence the basic principle of smallersize data (using data fusion) transmission reduces the battery

consumption is followed.

There are two types of data aggregation. The first type

fuses (to reduce the data size) the collected data from all the

nodes to forward to the based station. But it may results in less

accuracy and precision of the collected data from various

nodes. Whereas the second type of approach combines thecollected data and forward it as a single data packet with single

header. It maintains the data redundancy to achieve accuracy.

Two approaches are followed for energy efficient routing in

WSN, which are,

Clustering approach

Tree based approach

E. Clustering Techniques in WSN:

Clustering is the process of division of larger sensor

network into smaller manageable group to improve thescalability of the network. Other merits apart from scalability

such as bandwidth conservation within the cluster, avoidtransmission of redundant message within the network, energy

efficient route setup within the cluster are achieved. Further

research on the clustering based approach helped to design

various energy efficient routing protocols such as LEACH,HEED, and DECA.

a) LEACH

Low energy adaptive clustering hierarchy [5] follows the

clustering approach to distribute the energy consumption to all

along its network. Network is divided into Clusters based on

data collection and Cluster heads are elected randomly. The

CH then gathers the information from the nodes which belong

237

SaB2.8


5/6

to the respective cluster. Following are the steps for the

LEACH protocol.

Advertisement phase: The first step in LEACH protocol is theeligible cluster head nodes to issue a notification to the nodes

in its range to become a cluster member of the cluster. All the

nodes will be accepting the offer based upon the ReceivedSignal Strength (RSS).

Cluster set-up phase: Nodes respond to their selected cluster

heads in this phase.

Schedule creation: the cluster head have to make a TDMA

scheme after receiving response from the nodes. CH then

sends this TDMA scheme back to its cluster members to

intimate them when they have to pass their information to it.

Data transmission: The CH is provided with the data collected

by the individual sensors during its time interval and on all

other time, the cluster members’ radio will be off to reduce itenergy consumption.

Here in the LEACH protocol, multi cluster interference

problem was solved by using unique CDMA codes for each

cluster. This method helps to prevent the energy-drain for the

same sensor nodes which has been elected as the cluster leaderusing randomization of parameters (energy-life remaining) for

each time, CH would be replaced. The CH is in charge for

gathering data from its cluster members and multiplexes it to

form a single packet to be forwarded to the base station.

LEACH has shown a considerable improvement as comparedwith its previous protocols.

b) HEED:

Though the LEACH protocol is much more energy

efficient, but when compared with its predecessors, the maindrawbacks in this approach is the random selection of cluster

head. Sometimes the CH nodes may not be distributed

uniformly in the region and it will have its effect on the data

gathering. To prevent the random selection of CHs, a new

algorithm called Hybrid Energy Efficient Distributed

clustering: HEED [11] was designed and developed, whichselects the CHs based not only on the residual energy level but

the communication cost. The HEED protocol follows three

subsequent phases,

Initialization phase: During this phase, the initial CHs nodes’

percentage, represented by the variable C prob which will be

provided to the nodes. Every sensor node computes its

respective probability to become a CH using the formula,

CH prob=C prob * Eresidual/Emax

Where, Eresidual to residual energy level of the respective node,

Emax represents maximum battery energy. HEED supports

heterogeneous sensor nodes, where Emax may be different for

different nodes according to its functions and capacity.

Repetition phase: Until the CH node was found with the least

transmission cost, this phase was repeated. The concerned

node itself was selected as the CH unless the node finds the

appropriate CH.

Finalization phase: In this phase, the selection of CH is

finalized and it becomes the final CH node.

c) DECA:

DECA is an improved Distributed Efficient Clustering

A pproach [12], [13]. The basic difference between the HEED

and DECA is the decision making process and the score

computation. The phases in DECA operations are,

Start Clustering : In this primary phase, all the nodes will

compute its score using the function

Score=w1E+w2C+w3I

Where, E corresponds to the residual energy, C is the node

connectivity, and I is the node identifier. After some interval

time, the score value is provided to the neighboring nodes

along with the node ID and the cluster ID, if the computed

score is of higher value.

Receive Clustering Message: When the node receives the

score value higher than it and if it is not attached to any cluster

it accepts the sender node as its CH.

Actual announcement : After finishing the second phase, when

new nodes and already exciting nodes from some other cluster

forming a cluster with a new head, the CHs ID, cluster ID andscore value should be broadcasted.

Finalize Clustering: As same as in HEED protocol, the new

cluster with its head is decided for all other nodes.

d) EDECA:

EDECA is an Extended Distributed Efficient Clustering

A pproach [14] which is designed for making the network

more reliable, dependable and efficient. A minimum cost

spanning tree is designed for the transmission between

wireless nodes.

The basic difference between the HEED and DECA is thedecision making process and the score computation. In DECA,

each node periodically transmits a Hello message to identify

itself, and based on which, each node maintains a neighbour

list. EDECA is designed for more efficient results than that of

HEED and DECA. EDECA is based on the following scorefunction,

Score = w1B + w2C + w3P + w4M

Where B is battery power left, C is the node connectivity

and p is considered to be a probability of failure and P=1-p,

238

SaB2.8


6/6

which depends on the unfriendly & hostile environments of

the network nodes.

EDECA has three phases. Cluster head selection, Cluster

formation, and Cluster Conversion.

IV.

EXPECTED OUTCOME

The initial work shows the possibility of the positive result.

Our main goal is to implement clustering approach and analyze

the results to find energy efficient approach in VoIP

communication. The implantation of this approach within an

intra-network will definitely help to study on various other

parameters such as latency, mobility support, cluster stabilityand more over that, a good quality of service as well.

V. FUTURE WORK

As soon as all the mobile devices are registered in thenetwork with their details, the server application cancontinuously monitor the battery status and RSS values forselection of the new CH after the predefined interval. Futurework involves following goals of the model.

Alternate CH (Cluster Head) selection for continuous packet delivery using EDECA approach [14].

Compression Techniques for reduced data fusion cost.

The proposed model is initially designed to be implantedfor the campus or smaller network. But keeping future scope inmind, the model is designed flexible enough for the furthermodification to implement it on the larger network. Previouswork [2], [15], [16], [17], [18] could also be useful for theeffective and reliable module implementation.

VI.

CONCLUSION

The limited battery life is rising up as a major concern in

Mobile handheld devices. Thus the energy efficient clustering

based routing approaches would certainly be taken and to be

implemented in VoIP network for energy conservation and

longer battery life. Precaution should be taken as the quality ofthe service, security, voice quality, and jitter are some

important factors to consider along with elements responsible

for power consumption while working with VoIP. WSN

algorithms with relaxed constraints could be useful for an

effective and energy efficient VoIP communication.

R EFERENCES

[1] Goran Kalic, Iva Bojic and Mario Kusek “ Energy Consumption in

Android Phones when using Wireless Communication Technologies”,

The Ministry of Science, Education and Sports of the Republic ofCroatia., 2012.

[2] Vinod Namboodiri, Lixin Gao, “ Energy Efficient VoIP over Wireless

LANs”, ACM Press, Apr 2010.

[3] Gian Paolo Perrucciy and Frank H.P. Fitzeky and Giovanni Sassoy and

Wolfgang Kellererx and J¨org Widmerx, “On the Impact of 2G and 3G Network Usage for Mobile Phones’ Battery Life”, European Wireless,

2009.

[4] Kwan-Wu Chin,“ On Maximizing VoIP Capacity and EnergyConservation in Multi-Rate WLANs ”, IEEE Communication letters,

Vol. 14, No. 7, July 2010

[5] W.B. Heinzelman, A.P. Chandrakasan, H.Balakrishnan , ”Application

specific protocol architecture for wireless microsensornetworks”, IEEETransactions on Wireless Networking (2002), 2002.

[6] Zeadally, F. Siddiqui and P. Kubher, “Voice over IP in Intranet andInternet environments”, IEEE Proc.-Commun., Vol. 151, No. 3, June

2004.

[7] Florin Bota, Faheem Khuhawar, Marco Mellia, Michela Meo (Italy),

“Comparison of Energy Efficiency in PSTN and VoIP Systems”, ACMPress, May 2012.

[8] J.Zhu, C. Qiao and X. Wang, “ A Comprehensive Minimum Energy

Routing Scheme for Wireless Ad hoc Networks”, IEEE INFOCOM,

2004.

[9] M.L.Sichitiu, “Cross Layer Scheduling for Power Efficiency in

Wireless Sensor Networks”, IEEE INFOCOM , 2004.

[10] J. C. Chen, K. M. Sivalingam, P. Agrawal and S. Kishore, " A

Comparison of MAC Protocols for Wireless Local Networks Based on

Battery Power Consumption", In Proc. of IEEE INFOCOM, 1998.

[11] O. Younis, S. Fahmy, “ HEED: A Hybrid,Energy- Efficient, Distributed

clustering approach for Ad Hoc sensor networks”, IEEE Transactions

on Mobile Computing 3 (4) (2004) 366 – 379, 2004.

[12] Miau Yu, Jason H.Li and renato Levy, “ Mobility Resistant Clustering

in Multi- Hop Wireless Networks”, Journal of Networks, Vol.1, No.1,

May 2006.

[13] J. H. Li, M. Yu, and R. Levy, “ A Distributed Efficient Clustering

Approach for Ad Hoc and Sensor Networks”, Intelligent AutomationInc., Rockville, MD 20855, US, 2005.

[14] K. Chakraborty, A. Sengupta, “Extended Distributed EfficientClustering Approach : Energy Efficient Data Gathering in Wireless

Sensor Network”, Jadavpur University, kolkata, India.

[15] J. Surana, Rajeev G. Vishwakarma., “ Energy-Saving VoIP over

Wireless LANs”, ICAESM 2012, Mar 2012.

[16] U Shrawankar, V Thakare, “ Feature Extraction for a Speech Recognition System in Noisy Environment: A Study”, ICCEA,2010.

[17] U. Shrawankar, V. Thakare “Voice Activity Detector and NoiseTrackers for Speech Recognition System in Noisy Environment ”

IJACT, 2010.

[18] N. Chavhan, S. Chhabria, “ Multiple design patterns for voiceover IP security”, International Conference on Advances inComputing, Communication and Control, ACM, 2009.

239

SaB2.8

Documents

Energy Efficient VoIP Communication Using WSN Clustering Approach