Journal of Network and Computer Applications 36 (2013) 1001–1007

Contents lists available at SciVerse ScienceDirect

Journal of Network and Computer Applications

1084-80

http://d

n Corr

E-m

cskianis1 N

ably in

journal homepage: www.elsevier.com/locate/jnca

Network coding-based cooperative ARQ scheme for VANETs

Angelos Antonopoulos a, Charalabos Skianis b, Christos Verikoukis a,n

a Telecommunications Technological Centre of Catalonia (CTTC), Castelldefels, Barcelona, Spainb Department of Information and Communication Systems Engineering, University of the Aegean, Karlovassi, Greece

a r t i c l e i n f o

Article history:

Received 25 July 2011

Received in revised form

5 December 2011

Accepted 25 March 2012Available online 4 April 2012

Keywords:

Network coding

Cooperative networks

Automatic Repeat reQuest

802.11p

45/$ - see front matter & 2012 Elsevier Ltd. A

x.doi.org/10.1016/j.jnca.2012.03.012

esponding author. Tel.: þ34 93 645 29 00.

ail addresses: aantonopoulos@cttc.es (A. Anto

@aegean.gr (C. Skianis), cveri@cttc.es (C. Veri

ote that the words ‘‘partner’’, ‘‘relay’’ and ‘‘h

this paper.

a b s t r a c t

In this paper we introduce a novel Network Coding-based Medium Access Control (MAC) protocol for

Vehicular Ad Hoc Networks (VANETs) that use cooperative Automatic Repeat reQuest (ARQ) techniques.

Our protocol coordinates the channel access among a set of relays capable of using network coding in

order to minimize the number of the total transmissions, thus enhancing the performance of the

network in terms of Quality of Service (QoS) metrics. The proposed solution is compared to other

cooperative schemes, while analytical and simulation results are provided to evaluate our protocol.

& 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Vehicular Ad Hoc Networks (VANETs) have been deployed tofacilitate the communication between vehicles. Recently, IEEE802.11p (2010) has been introduced in order to deal with thespecial characteristics of VANETs, i.e. the high mobility of the nodesand the rapid changes in the topology. This new standard, alsoknown as Wireless Access in Vehicular Environment (WAVE), isbased on Carrier Sensing Multiple Access (CSMA), while it adoptsservice differentiation using adaptive backoff window sizes toachieve better QoS performance (Enhanced Distributed ChannelAccess—EDCA). In addition to the standardization, it would beinteresting to investigate how the recent innovative techniques,such as cooperation among nodes and network coding, could affectthe inter-vehicle communication. These new technologies createthe need of designing new MAC protocols that exploit their benefitsto efficiently use and manage the network resources.

The concept of cooperation was introduced by Cover andGamal (1979) in their fundamental paper on relay channels. Theirwork analyzed the capacity of the three-node network consistingof a transmitter, a receiver and a partner (relay).1 In their model,the spatial diversity gain is obtained by exploiting differentchannels seen by different nodes for transmitting data. On theother hand, network coding is an area that emerged in 2000(Ahlswede et al., 2000), and since then has attracted an increasing

ll rights reserved.

nopoulos),

koukis).

elper’’ are used interchange-

interest, as it promises to have a significant impact in both theoryand practice of networks. We can broadly define network codingas allowing intermediate nodes in a network to not only forwardbut also process the incoming information flows. Most of thework on this topic focuses on the physical layer aspect while onlyfew works examine these techniques considering the MAC layereffect (Katti et al., 2008; Argyriou, 2009; Zhang et al., 2008).Furthermore, the main assumption in the majority of the works isthat only one relay contributes to the communication, thusignoring the impact that the collisions and the idle slots causeto the network’s performance.

In this context, we propose a new MAC protocol (NCCARQ-MAC)that coordinates the transmissions among a set of relay nodeswhich act as helpers in a bidirectional communication that takesplace in a vehicular environment. The main contribution of ourproposed scheme lies on combining both cooperative and net-work coding techniques in order to enhance the system’s perfor-mance. To the best of our knowledge, there is no proposed MACprotocol in the literature that implements network codingin cooperative ARQ schemes with more than one relay, whilethere is a limited number of papers that apply network codingin cooperative schemes that take advantage of the multi-ratecapability of wireless standards (Tan et al., 2007).

The rest of the paper is organized as follows. Section 2 gives thebasic background on cooperative networking and outlines therelated work on MAC layer protocols for both simple and networkcoding-based cooperative schemes in the literature, especially inthe field of VANETs. In Section 3 we introduce our proposedNCCARQ-MAC protocol along with a detailed mathematical analysis.The validation of the analytical model and the numerical results areprovided in Section 4. Finally, Section 5 concludes the paper.

A. Antonopoulos et al. / Journal of Network and Computer Applications 36 (2013) 1001–10071002

2. Background and related work

2.1. Cooperative communication

In the context of cooperative communications, several schemesfocused on MAC layer have been already proposed in the literature(Lu et al., 2007; Alonso-Zarate et al., 2008; Liu et al., 2005; Guo andCarrasco, 2009; Zhu and Kuo, 2007). These works can be classifiedinto two main categories: (i) the cooperative ARQ-based protocolsand (ii) the protocols that transform one-hop transmissions tomulti-hop transmissions by exploiting the multi-rate capabilitiesof the wireless systems.

2.1.1. Cooperative ARQ-based protocols

Forward Error Correction (FEC) and Automatic Repeat reQuest(ARQ) algorithms are two basic error control methods for datacommunications (Lin and Costello, 1983). ARQ schemes havereceived considerable attention for data transmissions due totheir simplicity and higher reliability, compared to FEC schemes.

Regarding the protocols falling in this category (Lu et al., 2007;Alonso-Zarate et al., 2008), the retransmissions are initiated bythe destination after an erroneous packet reception. The helpersin a network are enabled to relay the original packets to a specificdestination, as ARQ defines, using higher data rates or betterchannel conditions in terms of Signal-to-Noise Ratio (SNR) values.

In the context of Vehicular Networks, Kaul et al. (2008) haveproposed a new protocol, GeoMAC, that exploits spatial diversityby allowing the nodes adjacent to the source to opportunisticallyforward data packets. In GeoMAC, the stations use a geographi-cally-oriented backoff mechanism which uses the geographicdistance to the destination as a heuristic to select the forwardermost likely to succeed. VC-MAC (Zhang et al., 2009) is anothercooperative ARQ protocol, designed mainly for broadcastingscenarios in VANETs. Specifically, as the nodes move fast, it ispossible that after a certain period of time they will be outsidethe range of a specific gateway that broadcasts data packets.However, other vehicles – moving relatively at the same speed –remain in their proximity, thus being potential relays.

2.1.2. Protocols that transform one-hop transmissions to multi-hop

transmissions

Using the concept of adaptive modulation (Morinaga et al., 1997),mobile stations in a multi-rate wireless network assign the modula-tion scheme and the transmission rate according to the detectedSignal-to-Noise Ratio (SNR) and the required transmission quality.Each modulation scheme could be further mapped to a range of SNRin a given transmission power. To achieve high transmission effi-ciency in wireless systems, stations select the highest available ratemodulation scheme according to the detected SNR.

The protocols of this class (Liu et al., 2005; Guo and Carrasco,2009; Zhu and Kuo, 2007) transform single one-hop transmissions tomulti-hop transmissions according to the channel conditions. Speci-fically, when the channel state between the relay and the destinationis better than the channel between the source and the destination, atwo-hop transmission is preferred instead of the direct transmission.

ADC-MAC (Zhou et al., 2011) is a novel protocol that coordi-nates the scheduled transmissions with regard to the channelconditions among the source, the destination and the relay nodesin a vehicular environment. In ADC-MAC, a three-party hand-shake takes place between the nodes in order to be decided whichis the most efficient way for the data to be transmitted.

2.2. Cooperation and network coding

Last years, there is a trend towards using network coding incooperative communications. The initial attempts for developing

network coding-based cooperative communications focused onphysical layer schemes(Wang and Giannakis, 2008; Xiao et al.,2007). These approaches refer to the coding gain and optimalpower allocation in simple cooperative topologies, usually con-sidering one relay or cooperation among the users.

However, the innovation of using network coding in coopera-tive communications is not confined only in the physical layer.Tan et al. (2007) presented one of the few works that focus onMAC layer aspect of network coding-based cooperative commu-nication. Their proposed protocol, called CODE, exploits thebenefits of both network coding and multi-rate capability of IEEE802.11 Standard. Specifically, the coding of the packets takesplace at the relay nodes, under two basic conditions: (i) the directlink between the sender and the receiver is poor and exists one ormore relay candidates that experience better link conditions and(ii) the traffic is bidirectional.

In VANETs’ domain, a content distribution protocol has beenintroduced by Lee et al. (2006). Their proposed scheme, namedCodeTorrent, adopts Random Linear Network Coding (RLNC) tech-niques to enhance the network performance. However, the imple-mentation of RLNC in VANETs has been later proven to be affectedby the resource constraints of the mobile nodes (Lee et al., 2008).

3. Proposed network coding-based cooperative ARQ MACprotocol

3.1. Protocol description

NCCARQ-MAC has been designed to coordinate the transmis-sions among a set of relays that support a bidirectional commu-nication between two nodes in a vehicular environment. The firstgoal of NCCARQ-MAC is to enable the mobile stations to requestcooperation by the neighboring nodes upon an erroneous recep-tion of a data packet. The second design goal of our proposedprotocol is to allow the helper nodes to perform network codingtechniques to the packets to be transmitted before relaying them.

The first requirement of NCCARQ-MAC is that all nodes in thenetwork should operate in promiscuous mode in order to be ableto listen to all ongoing transmissions and cooperate if required.However, this is not an energy issue in VANETs, since the wirelessinterfaces running on the vehicle’s battery power. The secondbasic requirement of NCCARQ-MAC is that the nodes should storea copy of any received data packet (regardless of its destinationaddress) until it is acknowledged by the destination station.

In NCCARQ-MAC, a cooperation phase is initiated once a packetis received erroneously by the destination. Some factors thatadversely affect the correct packet reception are the long distancebetween the nodes, the slow fading (shadowing) inside densepopulated areas, etc. Several error detection mechanisms such asCyclic Redundancy Code (CRC) can be applied in order to performerror control to the received messages. Therefore, the destinationstation initiates the cooperation phase by broadcasting a Requestfor Cooperation (RFC) message after sensing the channel idle forSIFS (Short Inter Frame Space) period of time. This message has theform of a control packet and higher priority over regular datatraffic, since data transmissions in 802.11 take place after a longerperiod of silence (DCF Inter Frame Space—DIFS). Furthermore, inthe special but not rare case of bidirectional traffic, i.e. when thedestination station has a data packet for the source station, thepacket is broadcasted piggybacked on the RFC message.

The stations that receive the RFC packet are potential candi-dates to become active relays for the communication process.Therefore, the relay set is formed upon the reception of the RFCand the participants stations get ready to forward their infor-mation. Since the partners have already stored the packets that

Fig. 1. General idea of NCCARQ-MAC operation.

A. Antonopoulos et al. / Journal of Network and Computer Applications 36 (2013) 1001–1007 1003

destined both to the destination (so called cooperative packet)and to the source (so called piggybacked packet), they create anew coded packet by combining the two existing data packets,using the XOR method. Accordingly, the active relays will try toget access to the channel in order to persistently transmit thenetwork coded (NC) packet. A simple scenario that subjects to theinitial principles of NCCARQ-MAC is depicted in Fig. 1. In thisparticular case, we consider a wireless communication between abase station (S) and a mobile destination (D). Since the destina-tion vehicle is moving at the boundaries of the source’s range, it ispossible that many packets will be received erroneously. How-ever, the vehicles that are moving inside the transmission range,between the source and the destination, constitute potentialrelays, since there is high probability of receiving the transmitteddata correctly.

In this point we have to state that NCCARQ-MAC is backwardscompatible with IEEE 802.11p Standard, as it uses the same framestructure and follows the same principles with the standard.However, some modifications have been made in order for theprotocol to efficiently exploit the advantages of both cooperativeand network coding techniques:

1.

We consider data traffic of same priority, thus not adoptingthe EDCA.

2.

There are ACK packets for the multi-cast transmission of the NCpacket in order to provide a reliable communication scheme.

3.

In the case of bidirectional traffic, the packet that is destinedback to the source is sent along with the Request for Coopera-tion packet, without taking part in the contention phase.

4.

Since the subnetwork formed by the relay set operates insaturated conditions, it is necessary to execute a backoffmechanism at the beginning of the cooperation phase in orderto minimize the probability of a certain initial collision.

Once the source and the destination receive the NC packetfrom the relay, they are able to decode it and extract therespective original data packets. Subsequently, they acknowledgethe received data packet by transmitting the respective ACK, thusterminating the cooperation phase. Receiving the acknowledg-ment packet, the relays are informed that the particular commu-nication has been completed, hence becoming able to erase thepackets of their buffers. In case that the received coded packetscannot be decoded after a certain maximum cooperation timeoutdue to transmission errors, the relays are obliged to forward againthe NC packet.

3.2. Operational example

In this subsection, we provide an example of the operation ofNCCARQ-MAC in order to clarify our proposed access protocol.A simple network topology with four stations is considered, all ofthem in the transmission range of each other. A source station (S)transmits a data packet (A) to a destination station (D) that doesalso have a packet (B) destined to the source station. Furthermore,there are two helper nodes (H1 and H2) that support this parti-cular bidirectional communication. The entire procedure is depictedin Fig. 2 and explained as follows:

1.

At instant t1, station S sends the data packet A to station D. 2. Upon reception, at instant t2, station D fails to demodulate the

data packet, thus broadcasting an RFC packet asking for coop-eration of the neighboring stations (H1 and H2 in this example)along with the data packet B, destined to the station S.

3.

The reception of the RFC (t3) triggers the stations H1 and H2

to become active relays and set up their backoff counters(CW1 and CW2, respectively) in order to participate in thecontention phase.

4.

At instant t4, the backoff counter of H1 expires and H1 transmitsthe coded packet A� B to the nodes S and D simultaneously.

5.

At instant t5, the station D retrieves the original packet A andsends back an ACK packet since it is able to decode properlythe XOR-ed packet.

6.

At instant t6, the node S acknowledges the packet B since it isable to decode properly the coded packet A� B.

3.3. Protocol analysis

3.3.1. Delay analysis

The application of network coding techniques in our proposedscheme implies the simultaneous transmission of more than onepacket in the network. Therefore, we analytically estimate theexpected time that is needed for two packets to be exchangedunder the NCCARQ-MAC protocol.

The total time that is elapsed from the initial transmissionuntil the correct reception in the destinations can be defined as

E½Ttotal� ¼ E½TD�þE½TCOOP � ð1Þ

where E½TD� represents the average time for a transmission of asingle data packet from the source to the destination and E½TCOOP �

corresponds to the average time required for a cooperative trans-mission via relays to be completed.

Fig. 3. The 2-dimensional Markov model’s state transition diagram.

Fig. 2. NCCARQ-MAC example of frame sequence.

A. Antonopoulos et al. / Journal of Network and Computer Applications 36 (2013) 1001–10071004

Since E½TD� has a value that is easy to be calculated dependingon the network’s parameters, we focus our analysis on the termE½TCOOP � in order to derive a closed-form expression for the overallpacket delay. The average time that is spent during the coopera-tion phase can be defined as

E½TCOOP � ¼ E½Tmin�þE½TCONT � ð2Þ

where E½Tmin� is the minimum average delay in case of perfectscheduling among the relays, i.e. contention-free scheme. On theother hand, the term E½TCONT � is used to denote the additionaldelay caused due to the contention phase which has been adoptedin our protocol in order for the probability of collisions to bereduced.

The expected number of retransmissions (E½r�) that arerequired in order to properly demodulate the coded packet atthe destination nodes is directly connected with the packet errorrate between the relays and the destination (PERR-D). However,in our scheme, two packets are sent at the same time via differentchannels and, as a result, the number of retransmissions can beexpressed as (Cocco et al., 2011)

E½r� ¼1þð1�PER1Þ � PER2

1�PER2þð1�PER2Þ � PER1

1�PER1

1�PER1 � PER2ð3Þ

where PER1 and PER2 represent the PERR-S and the PERR-D,respectively.

Therefore, the term E½Tmin� can be calculated as

E½Tmin� ¼ TSIFSþTRFCþTBþTONC

þE½r� � ðTDIFSþTA�BþTSIFSÞþTACKþTSIFSþTACK ð4Þ

where TRFC and TACK are the transmission times for the RFC andthe ACK packet, respectively. Furthermore, TA�B is the timerequired to retransmit a coded packet, given that the relaystations transmit at a same common transmission rate, whileTONC is the time that a relay needs in order to perform networkcoding techniques. Finally, TSIFS and TDIFS is the duration of a SIFSand a DIFS silence period, respectively.

Moreover, the term E½TCONT � can be defined as

E½TCONT � ¼ E½r� � E½Tc� ð5Þ

where E½Tc� represents the average time required to transmit asinge packet during the contention phase among all the relays. Inorder to compute this value we need to model the backoff counterof each of the relays with the Markov Chain presented by Bianchi(2000) (Fig. 3), since the formed subnetwork acts as a saturatedIEEE 802.11 ad hoc network despite the modifications in theaccess rules. According to this model, the probability t that a

station transmits in a randomly chosen slot, is given by

t¼Xm

i ¼ 1

bi,0 ¼b0;0

1�p¼

2ð1�2pÞ

ð1�2pÞðWþ1ÞþpWð1�ð2pÞmÞð6Þ

where

b0;0 ¼2ð1�2pÞð1�pÞ

ð1�2pÞðWþ1ÞþpWð1�ð2pÞmÞð7Þ

and the probability of a collision p as a function of t is given by

p¼ 1�ð1�tÞn�1ð8Þ

In formulas (6) and (7), bi,k represents the steady stateprobability of the state fi,kg, W is the size of the contentionwindow, m denotes the number of the backoff stages and n

corresponds to the number of the stations in the network.Furthermore, the probability that at least one relay attempts to

transmit can be expressed as

ptr ¼ 1�ð1�tÞn ð9Þ

and the probability of a successful transmission, i.e. one stationtransmits conditioned on the fact that at least one stationtransmits is given by

ps9tr ¼ntð1�tÞn�1

1�ð1�tÞnð10Þ

A. Antonopoulos et al. / Journal of Network and Computer Applications 36 (2013) 1001–1007 1005

Moreover, the probabilities of having an idle (pi), successful (ps) orcollided (pc) slot can be written as

pi ¼ 1�ptr ð11Þ

ps ¼ ptr � ps9tr ð12Þ

pc ¼ ptrð1�ps9trÞ ð13Þ

Considering the above probabilities, and given that the averagenumber of slots that we have to wait before having a successfultransmission can be represented as

E½N� ¼X1k ¼ 0

kð1�psÞkps ¼

1

ps

�1 ð14Þ

the total contention time can be written as

E½Tc� ¼ E½N� � E½Tslot9non_successful_slot� ð15Þ

Applying Bayes’s theorem we are able to estimate the averageduration of a slot, given that the specific slot is either idle orcollided

E½Tslot9non_successful_slot � ¼pi

1�ps

� �sþ pc

1�ps

� �Tcol ð16Þ

with s representing the duration of an idle slot, while Tcol

corresponds to the time of a collision and in our scheme isequal to

Tcol ¼ TDIFSþTA�BþTSIFS ð17Þ

Therefore, using Eqs. (14)–(16), formula (5) can be rewritten as

E½TCONT � ¼ E½r� �1

ps

�1

� �pi

1�ps

� �sþ pc

1�ps

� �Tcol

� �ð18Þ

Finally, we are able to derive a closed-form formula and computethe total delay for two packets to be exchanged in the system byexploiting the Eqs. (2), (4) and (18).

Table 1System parameters.

Parameter Value

MAC header 34 bytes

PHY header 96 ms

ACK, RFC 14 bytes

DATA packets 1500 bytes

SIFS 10 ms

DIFS 50 ms

Table 2Simulation scenarios.

SNR(S-D)

Source controlrate (Mb/s)

Source datarate (Mb/s)

Relay controlrate (Mb/s)

Relay datarate (Mb/s)

Low 6 6 6 54

Medium 6 24 6 54

High 6 54 6 54

3.3.2. Throughput analysis

The total throughput of the network can be defined as the sumof the throughput that is produced by the successful directtransmissions plus the throughput derived by the cooperationphase after erroneous packet receptions. This can be mathemati-cally expressed as

E½Stotal� ¼ E½SD�þE½SCOOP� ð19Þ

where

E½SD� ¼ ð1�PERS-DÞ �E½P�

E½TD�ð20Þ

and

E½SCOOP� ¼ 2 � PERS-D �E½P�

E½Ttotal�ð21Þ

In the above expressions, the packet error rate between thesource and the destination is given by PERS-D, while E½P� denotesthe average packet payload. Furthermore, E½TD� represents theaverage time for a transmission of a single data packet from thesource to the destination and E½Ttotal� corresponds to the totaltime required for a cooperative transmission via relays to becompleted. In this point, it must be clarified that it is mandatoryto use the coefficient 2 in formula (21), since two packets aredelivered in each particular transmission.

Thus, having obtained a closed-form expression for E½Ttotal�

in the previous subsection and since E½P�, E½TD� and PERS-D areknown parameters, we are able to compute the theoretical system’sthroughput.

4. Performance evaluation

In order to validate our analysis and further evaluate theperformance of NCCARQ-MAC we have developed a time-drivenCþþ simulator that executes the rules of the protocol. Here wepresent the simulation set up along with the results of ourexperiments.

4.1. Simulation scenario

The vehicular network under consideration consists of a pair oftransmitter-receiver (both nodes transmit and receive data) and aset of relay nodes that facilitate the communication, all of them inthe transmission range of each other throughout the simulation.Additionally, the relay nodes are capable of performing networkcoding techniques to their buffered packets before relaying them.In order to focus on the impact of both network codingand cooperative communication, we have made the followingassumptions:

1.

The traffic is bidirectional, i.e. the destination node has alwaysa packet destined back to the source node.

2.

Original transmissions from source to destination are alwaysreceived with errors (PERS-D¼1), thus initiating a cooperativephase.

3.

The channel between the source and the destination is errorsymmetric, i.e. PERS-D ¼ PERD-S.

4.

The channel between the source and the relays is error free, i.e.PERS-R¼0.

5.

The relay nodes transmit in a common transmission rate.

The configuration parameters of the network are summarizedin Table 1 considering the IEEE 802.11 physical layer. The time forapplying network coding (TONC) to the data packets is consideredto be negligible, since the coding takes place between only twopackets.

Furthermore, we consider various scenarios with differentSNR values between the original source and the destination. Thecontrol packets are transmitted always at the rate of 6 Mb/s, whilethe transmission rate for the data packets is 6, 24 and 54 Mb/s forlow, medium and high SNR values, respectively. On the other hand,given that the relays are usually placed close to the destination,we assume that the transmission rates in all scenarios are 6 and54 Mb/s for control and data packets, respectively. The threedifferent scenarios are summarized in Table 2. Furthermore, since

A. Antonopoulos et al. / Journal of Network and Computer Applications 36 (2013) 1001–10071006

the wireless nodes operate in promiscuous mode, the network isunder saturated conditions, i.e. the nodes have always packets tosend in their buffers.

In order to evaluate our approach, we compare our schemewith a simple cooperative ARQ scheme (so called CARQ-MAC),where the bidirectional communication takes place in two steps.In the first step, the source sends a packet to the destination and,upon the erroneous reception the destination broadcasts the RFCpacket, thus triggering the relays to retransmit the packet. In thesecond step, the destination transmits its own packet to thesource and the same procedure as in the first step is repeated,thus consuming valuable network resources. In both steps, therelays take part in the contention phase in order to access themedium and transmit their packets.

4.2. Performance results

Fig. 4 presents the results (both analytical and numerical) inthe three scenarios described above. The relay set consists offive nodes, each of them implementing a backoff counter starting

1 2 3 4 50.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8x 107

Number of Retransmissions (E[r])

Thro

ughp

ut (b

/s)

NCCARQ−MAC(Low SNR)an.NCCARQ−MAC(Low SNR)sim.NCCARQ−MAC(Medium SNR)an.NCCARQ−MAC(Medium SNR)sim.NCCARQ−MAC(High SNR)an.NCCARQ−MAC(High SNR)sim.CARQ−MAC(Low SNR)sim.CARQ−MAC(Medium SNR)sim.CARQ−MAC(High SNR)sim.

Fig. 4. System’s throughput (NCCARQ-MAC vs. CARQ-MAC) (n¼5, CWmin ¼ 32).

16 32 64 128 256 5123

3.5

4

4.5

5

5.5

6

6.5

7 x 106

CW minimum size, E[r]=2

Thro

ughp

ut (b

/s)

NCCARQ−MAC(Low SNR)an.NCCARQ−MAC(Low SNR)sim.CARQ−MAC(Low SNR)sim.

Fig. 5. System’s throughput (NCCARQ-MAC vs. CARQ-MAC) (n¼10, E[r]¼2).

with a contention window CWmin ¼ 32. In the figure, we can seethat the numerical and the simulation results are almost perfectlymatched, thus verifying our analysis. Comparing with simplecooperative schemes which have only the advantageof spatial diversity through relays without network coding cap-abilities, we can achieve an enhancement in the network’sthroughput up to 80%. It is also worth noticing that NCCARQ-MAC outperforms CARQ-MAC even for worse SNR scenarios.To clarify, observing the achievable throughput in NCCARQ-MACfor the medium SNR scenario we can see that it clearly outperformsCARQ-MAC under the high SNR scenario.

Fig. 5 shows the throughput’s performance with regard to theminimum contention window of the wireless nodes. In thisspecific case (low SNR scenario), there are ten relays in thenetwork (total twelve nodes including the source and the desti-nation) and we assume that two retransmissions are needed inorder for the packets to be delivered properly to the destinations(E½r� ¼ 2). As it was expected, the two curves exhibit similarbehavior, achieving maximum throughput for CWmin ¼ 64. How-ever, a useful conclusion that derives by the figure is thatNCCARQ-MAC outperforms CARQ-MAC independently of theinitial CW, i.e. the worst performance of NCCARQ-MAC is clearlybetter than the best performance of CARQ-MAC.

Fig. 6 presents the packet delay in both network coding-basedand simple cooperative ARQ MAC protocols. In this point, we mustrecall that two packets are delivered to their respective destina-tions in each transmission cycle of NCCARQ-MAC. Hence, in orderto be accurate, we compare the delay in NCCARQ-MAC with thetime required for two packets to be exchanged in CARQ-MAC.

As it can be observed, we can achieve significantly lower packetdelay by applying network coding techniques. For example, con-sidering the low SNR scenario, the average time that is required fortwo packets to be transmitted using CARQ-MAC is 5.9 ms inchannels where one retransmission is necessary, reaching up to9.5 ms when five retransmissions are required. On the other hand,the delay values in NCCARQ-MAC are 3.3 and 5 ms, for one and fiveretransmissions, respectively. This difference can be rationallyexplained considering the operation of our proposed NCCARQ-MAC scheme, where some data packets are sent to the relayattached to the RFC message, thus avoiding the erroneous channel.Furthermore, in our proposed scheme we manage to reduce thebackoff phases by sending two packets simultaneously, while in

1 2 3 4 51

2

3

4

5

6

7

8

9

10 x 10−3

Number of Retransmissions (E[r])

Tota

l Del

ay (s

ec)

NCCARQ−MAC(Low SNR)an.NCCARQ−MAC(Low SNR)sim.NCCARQ−MAC(Medium SNR)an.NCCARQ−MAC(Medium SNR)sim.NCCARQ−MAC(High SNR)an.NCCARQ−MAC(High SNR)sim.CARQ−MAC(Low SNR)sim.CARQ−MAC(Medium SNR)sim.CARQ−MAC(High SNR)sim.

Fig. 6. Packet delay (NCCARQ-MAC vs. CARQ-MAC).

A. Antonopoulos et al. / Journal of Network and Computer Applications 36 (2013) 1001–1007 1007

simple cooperative protocols the relays have to participate in thecontention phase for each packet that has to be retransmitted.Therefore, we are able to enhance the packet delay, since the timespent in idle slots and collisions is significantly reduced, especiallyas the number of required retransmissions grows.

5. Conclusive remarks

In this paper, a novel network coding-based cooperative ARQ(NCCARQ-MAC) scheme for Vehicular Networks was presented.Compared to simple cooperative ARQ protocols, the proposedsolution improves up to 80% the network’s aggregated bandwidthby minimizing the number of the total transmissions, while theaverage time to transmit data packets is significantly reduced.Our future work will be focused on mobility issues, i.e. howmobility affects the performance of our protocol. Furthermore, weplan to study the energy efficiency aspects of our scheme, whilewe intend to elaborate on relay selection and on the application ofgame theoretic techniques in VANETs.

Acknowledgments

This work has been funded by the Research ProjectsCO2GREEN (TEC2010-20823), GREENET (264759) and Green-T(CP8-006).

References

Ahlswede R, Cai N, Li SYR, Yeung RW. Network information flow. IEEE Transactionson Information Theory 2000;46(4):1204–16.

Alonso-Zarate J, Kartsakli E, Verikoukis C, Alonso L. Persistent RCSMA: a MACprotocol for a distributed cooperative ARQ scheme in wireless networks.EURASIP Journal on Advances in Signal Processing 2008;2008:13 [Article ID817401].

Argyriou A. Wireless network coding with improved opportunistic listening. IEEETransactions on Wireless Communications 2009;8(4):2014–23.

Bianchi G. Performance analysis of the IEEE 802.11 distributed coordinationfunction. IEEE Journal on Selected Areas in Communications 2000;18(3):535–547.

Cocco G, Gunduz D, Ibars C. Throughput analysis in asymmetric two-way relaychannel with random access. In: IEEE international conference on commu-nications (ICC) 2011, Kyoto, Japan, June; 2011.

Cover TM, Gamal AAE. Capacity theorems for the relay channel. IEEE Transactionson Information Theory 1979;25(5):57284.

Guo T, Carrasco R. CRBAR: cooperative relay-based auto rate MAC for multiratewireless networks. IEEE Transactions on Wireless Communications 2009;8(12):5938–5947.

IEEE Standard for Information technology—telecommunications and informationexchange between systems—local and metropolitan area networks—specificrequirements. Part 11: wireless LAN Medium Access Control (MAC) andPhysical Layer (PHY) specifications amendment 6: wireless access in vehicularenvironments, 15 July; 2010. p. 1–51.

Katti S, Rahul H, Hu W, Katabi D, Medard M, Crowcroft J. XORs in the air: practicalwireless network coding. IEEE/ACM Transactions on Networking 2008;16(3):497–510.

Kaul S, Gruteser M, Onishi R, Vuyyuru R. GeoMAC: geo-backoff based co-operativeMAC for V2V networks. In: IEEE international conference on vehicularelectronics and safety (ICVES), 22–24 September; 2008. p. 334–9.

Lee U, Park JS, Yeh J, Pau G, Gerla M. CodeTorrent: content distribution usingnetwork coding in VANET. In: Proceedings of the 1st international workshopon decentralized resource sharing in mobile computing and networking(MobiShare ’06). New York, USA: ACM; 2006.

Lee SH, Lee U, Lee KW, Gerla M. Content distribution in VANETs using networkcoding: the effect of disk I/O and processing O/H. In: Proceedings of 5th annualIEEE communications society conference on sensor, mesh and ad hoc com-munications and networks (SECON ’08), 16–20 June; 2008. p. 117–5.

Lin S, Costello DJ. Error control coding: fundamentals and applications.EnglewoodCliffs, NJ: Prentice-Hall; 1983.

Liu P, Tao Z, Panwar S. A cooperative MAC protocol for wireless local areanetworks. In: IEEE international conference on communications (ICC), vol. 5,16–20 May; 2005. p. 2962–8.

Lu K, Fu S, Qian Y. Increasing the throughput of wireless LANs via cooperativeretransmission. In: IEEE global telecommunications conference (GLOBECOM),26–30 November; 2007. p. 5231–5.

Morinaga N, Nakagawa M, Kohno R. New concepts and technologies for achievinghighly reliable and high-capacity multimedia wireless communicationssystems. In: IEEE communications magazine, January; 1997. p. 34–40.

Tan K, Wan Z, Zhu H, Andrian J. CODE: cooperative medium access for multiratewireless ad hoc network. In: Proceedings of the 4th annual IEEE communica-tions society conference on sensor, mesh, and ad hoc communications andnetworks (SECON), June; 2007.

Wang T, Giannakis GB. Complex field network coding for multiuser cooperativecommunications. IEEE Journal on Selected Areas in Communications 2008;26:561–571.

Xiao L, Fuja T, Kliewer J, Costello D. A network coding approach to cooperativediversity. IEEE Transactions on Information Theory 2007;53:3714–22.

Zhang J, Chen YP, Marsic I. BEND: MAC-layer proactive mixing protocol fornetwork coding in multihop wireless networks. In: 9th ACM MobiHoc, HongKong, May; 2008.

Zhang J, Zhang Q, Jia W. VC-MAC: a cooperative MAC protocol in vehicularnetworks. IEEE Transactions on Vehicular Technology 2009;58(3):1561–71.

Zhou T, Sharif H, Hsempel M, Mahasukhon P, Wang W, Ma T. A novel adaptivedistributed cooperative relaying MAC protocol for vehicular networks. IEEEJournal on Selected Areas in Communications 2011;29(1):72–82.

Zhu XJ, Kuo GS. Cooperative MAC scheme for multi-hop multi-channel wirelessmesh networks. In: IEEE 68th vehicular technology conference (VTC) 2008-Fall, 21–24 September; 2008. p. 1–6.