22
INTERNATIONAL JOURNAL OF COMMUNICATION SYSTEMS Int. J. Commun. Syst. 2009; 22:937–958 Published online 20 April 2009 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/dac.1007 The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement Yun Li 1, 2, , , Chonggang Wang 3 , Weiliang Zhao 1 and Xiaohu You 2 1 Key Laboratory of Network Control and Intelligent Instrument, Chongqing University of Posts and Telecommunications, Ministry of Education, Chongqing, People’s Republic of China 2 National Mobile Communications Research Laboratory, Southeast University of China, Nanjing, People’s Republic of China 3 NEC Laboratories America, Princeton, U.S.A. SUMMARY The hidden-terminal problem significantly degrades the performance of IEEE 802.11 DCF. Many previous works have investigated its influence on the throughput of CSMA-based medium access control (MAC) protocols, especially IEEE 802.11 DCF. In this paper, we introduce a new Jamming problem for IEEE 802.11-based mobile ad hoc networks, which is caused by hidden terminals. An analytical model is established for this problem. Based on this model, an adaptive DCF (ADCF), is designed to solve the jamming problem through adaptively adjusting the minimum contention window of hidden terminals. Simulation results effectively demonstrate that the proposed A-DCF can avoid the jamming and in turn greatly improve channel utilization and throughput. Copyright 2009 John Wiley & Sons, Ltd. Received 20 January 2008; Revised 20 January 2009; Accepted 29 January 2009 KEY WORDS: IEEE 802.11; mobile ad hoc networks; hidden terminals; jamming problem 1. INTRODUCTION Carrier sense multiple access (CSMA) protocols have been used in a number of packet-radio networks in the past. The major feature of CSMA protocols is listen-before-talk, which basically is Correspondence to: Yun Li, Key Laboratory of Network Control and Intelligent Instrument, Chongqing University of Posts and Telecommunications, Ministry of Education, Chongqing, People’s Republic of China. E-mail: [email protected] Contract/grant sponsor: National Science Foundation of China; contract/grant number: 60702055 Contract/grant sponsor: New Century Excellent Talents in University; contract/grant number: NCET-07-0914 Contract/grant sponsor: Science and Technology Research Project of Chongqing Municipal Education Commission of China; contract/grant number: KJ070521 Copyright 2009 John Wiley & Sons, Ltd.

The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

  • Upload
    yun-li

  • View
    214

  • Download
    1

Embed Size (px)

Citation preview

Page 1: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

INTERNATIONAL JOURNAL OF COMMUNICATION SYSTEMSInt. J. Commun. Syst. 2009; 22:937–958Published online 20 April 2009 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/dac.1007

The Jamming problem in IEEE 802.11-based mobile ad hocnetworks with hidden terminals: Performance analysis

and enhancement

Yun Li1,2,∗,†, Chonggang Wang3, Weiliang Zhao1 and Xiaohu You2

1Key Laboratory of Network Control and Intelligent Instrument, Chongqing University of Posts andTelecommunications, Ministry of Education, Chongqing, People’s Republic of China

2National Mobile Communications Research Laboratory, Southeast University of China, Nanjing,People’s Republic of China

3NEC Laboratories America, Princeton, U.S.A.

SUMMARY

The hidden-terminal problem significantly degrades the performance of IEEE 802.11 DCF. Many previousworks have investigated its influence on the throughput of CSMA-based medium access control (MAC)protocols, especially IEEE 802.11 DCF. In this paper, we introduce a new Jamming problem for IEEE802.11-based mobile ad hoc networks, which is caused by hidden terminals. An analytical model isestablished for this problem. Based on this model, an adaptive DCF (ADCF), is designed to solve thejamming problem through adaptively adjusting the minimum contention window of hidden terminals.Simulation results effectively demonstrate that the proposed A-DCF can avoid the jamming and in turngreatly improve channel utilization and throughput. Copyright q 2009 John Wiley & Sons, Ltd.

Received 20 January 2008; Revised 20 January 2009; Accepted 29 January 2009

KEY WORDS: IEEE 802.11; mobile ad hoc networks; hidden terminals; jamming problem

1. INTRODUCTION

Carrier sense multiple access (CSMA) protocols have been used in a number of packet-radionetworks in the past. The major feature of CSMA protocols is listen-before-talk, which basically is

∗Correspondence to: Yun Li, Key Laboratory of Network Control and Intelligent Instrument, Chongqing Universityof Posts and Telecommunications, Ministry of Education, Chongqing, People’s Republic of China.

†E-mail: [email protected]

Contract/grant sponsor: National Science Foundation of China; contract/grant number: 60702055Contract/grant sponsor: New Century Excellent Talents in University; contract/grant number: NCET-07-0914Contract/grant sponsor: Science and Technology Research Project of Chongqing Municipal Education Commissionof China; contract/grant number: KJ070521

Copyright q 2009 John Wiley & Sons, Ltd.

Page 2: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

938 Y. LI ET AL.

used to guarantee that neighboring stations will not transmit simultaneously and in turn avoid colli-sions. However, the existence of hidden-terminal problem, which cannot be completely preventedby carrier sensing, could substantially impact the performance of CSMA protocols. Several workshave been conducted to evaluate the performance of CSMA-based packet-radio networks consid-ering hidden terminals. For example, it has been shown in [1] that hidden terminals greatly degradethe performance of CSMA. The earliest solution is busy tone multiple access (BTMA) proposedin [1] as a natural extension of CSMA to eliminate the hidden-terminal problem. In BTMA, a basestation transmits a busy-tone signal on the busy-tone channel as long as it successfully senses theoccurrence of carrier on the data channel, so as to prevent hidden terminals from contending thedata channel. But BTMA is only effective for base stations. To overcome this limitation, Deng andHass [2] proposed dual busy tone multiple access (DBTMA) that does not require base stations.In fact, busy-tone-based CSMA protocols need two physical channels, one for data transmissionand another for busy-tone signal transmission. For single-channel networks, we need to designdifferent schemes. In [3], Karn proposed a protocol called multiple access collision avoidance(MACA). MACA attempts to detect collisions at the receiver by establishing a request–responsedialogue between the sender and the intended receiver. Before a station starts to transmit data, itfirst sends a request-to-send (RTS) to a receiver, and the receiver responds with a clear-to-send(CTS) if it successfully receives the RTS. Several similar protocols were designed in [4–6] basedon such RTS–CTS exchanges.

The most deployed CSMA-based medium access control (MAC) protocol is IEEE 802.11distributed coordinated function (DCF) [7]—a popular standard for wireless local area networks(WLANs). Many works have been conducted on the DCF’s performance analysis and enhancement.The theoretical throughput limit of IEEE 802.11 DCF with p-persistent was given in [8]. Worksin [9, 10] present a Markov chain to capture its binary slotted exponential backoff procedure andcalculate the saturated throughput of DCF. Those works assume that each station is located withinthe carrier sense range of each other and no hidden terminals. A modified backoff algorithm toguarantee fairness for scenarios with variable packet length is proposed in [11]. In [12], a generalmechanism is presented to provide proportional fairness through translating a given fairness modelinto a corresponding contention resolution or a backoff algorithm. In [13], the authors analyzed thefairness of traditional transmission control protocol (TCP) over 802.11 DCF. Besides throughputand fairness, there is a growing concern on the capability of WLAN to support quality of service(QoS). In IEEE 802.11e [14], enhanced distributed channel access (EDCA) access rule is specifiedto support differentiated QoS with the introduction of traffic categories (TCs), which enables MACservice data units (MSDUs) to be delivered through multiple different backoff instances withinone station. Each backoff instance is parameterized with TC-specific parameters, which representQoS requirements of different TCs. In addition, a distributed MAC weighted round Robin (WRR)packet scheduler based on the differentiated backoff is proposed in [15] to emulate an idealscheduler and achieve proportional bandwidth allocation. The performance of EDCA includingits system saturation throughput has been analyzed in [16, 17]. Researchers have also analyzedthe delay properties of IEEE 802.11. In [18], a Markov chain model is established to derive theaverage packet delay of IEEE 802.11 DCF. Zanella and Pellegrini [19] obtained a closed-formprobability generating function for the average packet service time of a cluster of IEEE 802.11terminals. Another analytical model is proposed in [20] to calculate the average saturation delaysof different prioritized traffic classes. The authors of [21] derived both throughput and averageMAC delay of IEEE 802.11. Reference [22] studies the throughput and delay performance ofaccess control in both IEEE 802.11 DCF-based WLANs and ETSI RES 10 HiperLan through

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 3: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 939

simulations. The average access contention delay of 802.11 DCF is analyzed in [23] by assumingthat all stations share the channel fairly. The medium access delay in multi-hop ad hoc networks iscalculated in [24]. Recent works such as [13, 25] studied the performance degradation of DCF dueto hidden terminals. Paper [26] improved the performance of TCP over MANET by using a gentleand more conservative approach to increase the sending rate during the congestion avoidance phaseof TCP. Tsertou and Laurenson proposed a Markov model to analyze the saturation throughput ofDCF with hidden terminal in [27] and extended this model for binary exponential backoff (BEB)in [28]. Wu et al. [29] also proposed a Markov chain for saturated throughput of 802.11 WLANswith hidden terminals when access point (AP) exists. Another relatively simple analytical model isproposed in [30] to study the channel utilization of an IEEE 802.11 ad hoc network when hiddenterminals exist. Paper [31] focused on the short-term fairness of IEEE802.11 and showed that theRTS/CTS access is unfair compared with basic access. Kim et al. [32] gave a simple but effectivemechanism for hidden station detection by exploiting RTS/CTS exchange. In [33], the authorsproposed two methods to detect hidden terminals in 802.11-based ad hoc networks.

In the literature, many works have been conducted on the performance of 802.11 DCF, especiallyin terms of throughput, delay and QoS. Some of them focus on the hidden-terminal problem.However, most existing works on the hidden-terminal problem only concern its influence on thesaturation throughput. Although there exist some jamming-related works [34–37] for IEEE 802.11DCF-based WLANs, yet most of them study only the jamming—which is caused by misbehavingstations such as manipulating access control and contention window (CW) parameters. The authorsof [34, 35, 37] discussed several jamming techniques and their effectiveness in IEEE802.11b interms of network throughput and power expended by the jammer. They show that it is possible todesign intelligent jamming schemes that use little energy and can hardly be detected are effectivein decreasing throughput. The impact of jamming techniques on IEEE 802.11e is studied in [36].In this paper, we analyze a different jamming problem in 802.11-based mobile ad hoc networks,which is caused by hidden terminals whose behavior is normal and strictly according to the standardspecifications. A model is built to analyze this problem and an adaptive DCF (A-DCF) is proposedto resolve this problem.

The remainder of this paper is organized as follows. Section 2 briefly reviews IEEE 802.11 DCFbackoff procedure and the jamming problem is described and analyzed in Section 3. Section 4presents A-DCF to solve the jamming problem and numerical results are given in Section 5. Finally,Section 6 concludes the whole paper.

2. OVERVIEW OF 802.11 DCF BACKOFF PROCEDURE

IEEE 802.11 standard [7] specifies the PHY and MAC operations for a WLAN. Two definedMAC mechanisms are point coordination function (PCF) for contention free services and DCFfor contention-based services. PCF is actually based on DCF. DCF relies on two techniques totransmit packets: basic access and RTS/CTS-based access. The basic access employs a two-wayhandshaking: (1) First, a data frame is sent out at the source station once a channel is sensedidle; (2) Second, a positive acknowledgement (ACK) is sent back from the destination station tothe source station if the data frame is correctly received. In contrast, RTS/CTS uses a four-wayhandshaking technique, where the source station and the destination station first exchange shortRTS and CTS frames to reserve the channel before starting to transmit the data frame.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 4: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

940 Y. LI ET AL.

A DCF-enabled station performs CSMA with the Collision Avoidance (CSMA/CA) procedureto grab the channel as follows. A station with a pending packet waits until the medium becomesidle if the medium is currently sensed (physically and virtually) busy. After the channel becomesidle and keeps idle for the duration of distributed interframe space (DIFS) period, the stationsets its backoff timer to uniform[0,CWmin−1]×aSlotTime. aSlotTime is set as the required timeby any station to detect the transmission of packet from any other station. The default value ofaSlotTime is 20�s as defined in IEEE 802.11. If there is no channel activity during each backoffslot, backoff timer is decreased by aSlotTime. If the channel is sensed busy during a backoffslot, the backoff timer is suspended until the channel is idle for the duration of DIFS periodagain. After that, the backoff timer is resumed. When the backoff timer reaches zero, transmissioncommences. After a transmission, an ACK, indicating if the transmission is successful, shallbe received from the destination. If the transmission is successful, the station sets its backofftimer to uniform[0,2iCWmin−1]×aSlotTime again before transmitting the next packet. If thetransmission has failed, the station sets its backoff timer to uniform[0,2i ×CWmin−1]×aSlotTimeto retransmit the failed packet. Here, i is the retransmission count for a given packet. DCF adoptsan exponential backoff scheme. At each packet transmission, the backoff time is uniformly chosenin the range of [0,CW−1]. The value of CW depends on the number of failed transmissions.At the first transmission attempt, CW is set to CWmin, the minimum contention window. Aftereach unsuccessful transmission, CW is doubled, up to the maximum value CWmax=2m ∗CWmin,where m is the maximum backoff stage.

3. ANALYSIS OF THE JAMMING PROBLEM

This section studies the impact of the jamming on the performance of IEEE 802.11 DCF-basedmobile ad hoc networks. First, we conduct simulations to coarsely discover the influence ofthe jamming problem. Then an analytical model is established to quantitatively investigate theperformance degradation that resulted from the jamming.

3.1. Experiments

All experiments were conducted using NS-2 [38]. A simple link topology with four nodes as shownin Figure 1 was used. Suppose that each node can only directly communicate with its one-hopneighbor and the receiving range and interference range are identical. We configure two constantbit rate (CBR) flows: (1) one is from Node 0 to Node 1, denoted as F1; (2) another is from Node 2to Node 3, denoted as F2. F1 and F2 start to send packet at time t=1 and 60 s, respectively. Thewireless channel is saturated and the simulation time is 120 s. The PHY layer uses direct sequencespread spectrum (DSSS) and the MAC layer is DCF in RTS/CTS-based access mode, as definedin IEEE 802.11 [7]. The routing protocol is ad hoc on-demand distance vector (AODV) [39].Simulation parameters are listed in Table I.

The simulation results are shown in Figure 2, where the average throughput of F1 and F2 iscalculated in every 1.2 s time interval. It can be seen from Figure 2 that the throughput of F1sharply decreases to zero when F2 starts at the time t=60s, and F2 occupies all wireless channelsafter that. This means that F2 completely stops F1 from transmitting packets. We refer to thisphenomenon as the Jamming Problem, which actually is caused by hidden terminals. We can findthe detailed reason from the simulation trace file.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 5: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 941

250m

200mNode0 Node1 Node2 Node3

F2F1

Figure 1. A linear topology.

Table I. Simulation parameters.

Channel bit rate 2Mbit/sSlot time 20�sSIFS 10�sDIFS 50�sPHY header 192 bitsMAC header 144 bitsRTS length 160 bitsCTS length 112 bitsCWmin 32CWmax 1024CBR packet size 920 BytesRouting protocol AODV

0 20 40 60 80 100 1200

500

1000

1500

2000

Simulation Time(s)

Thr

ough

put(

Kbp

s)

F1F2

Figure 2. Throughput of F1 and F2.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 6: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

942 Y. LI ET AL.

59.99 60 60.01 60.02 60.03 60.04 60.05

0

1

2

Simulation Time(s)

Nod

e ID

60.1 60.2 60.3 60.4 60.5 60.6 60.7 60.8

0

0

2

Simulation Time(s)

Nod

e ID

60.9 60.91 60.92 60.93 60.94 60.95 60.96

0

1

2

Simulation Time(s)

Nod

e ID

90.55 90.555 90.56 90.565 90.57 90.575 90.58 90.585 90.59 90.595 90.6

0

1

2

Simulation Time(s)

Nod

e ID

sendRTS RTS-COL sendCTS recvCTS sendDATA

recvACK RET send-route-req recv-route-rep route-req-col

(a)

(b)

(c)

(d)

Figure 3. Events extracted from the simulation trace file: (a) RTS frames sent by Node 0 collide in Node 1with transmission of Node 2; (b) route request packets sent by Node 0 collide in Node 1 with transmissionof Node 2; (c) at t=60.915s, Node 0 successfully establishes the route to Node 1. However, all of theimmediate RTS frames sent by Node 0 again collide with transmission of Node 2; and (d) at t=90.561s,Node 0 successfully establishes the route to Node 1. However, all of the immediate RTS frames sent by

Node 0 again collide with transmission of Node 2, which is similar to Figure 3(c).

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 7: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 943

Figure 3 shows a part of the events exacted from the trace file. Since Node 2 is hidden fromNode 0, Node 0 cannot sense the transmission of Node 2. Therefore, Node 0 still sends RTSframes even if Node 2 is in transmission, which results in all of the RTS frames sent by Node0 collided at Node 1 (as shown in Figure 3(a)). When the number of RTS transmissions reachesthe maximum retry limit (Rmax), Node 0 deduces that the wireless link with Node 1 is ‘broken’and in turn it starts to send route request packets to discover a path to Node 1. However, becauseof hidden-terminal problem, once again most of the route packets get collided at Node 1 with thetransmission from Node 2 (as shown in Figure 3(b)), hence Node 0 can hardly find the path toNode 1 in the simulation time from 60 to 120 s. Though Node 0 does find the path to Node 1only twice, respectively, at 60.914727667 and 90.560147000 s, but all of the immediate RTSframes following the successful path establishment again get collided (as shown in Figures 3(c)and (d). Therefore, Node 0 cannot successfully send data packets to Node 1 after Node 2 starts totransmit because of the hidden-terminal problem, and this causes the throughput of F1 to be 0 inthe simulation time from 60 to 120 s. In the next sub-section, we will establish a model to analyzethis problem.

3.2. Analytical model

The above sub-section has shown that the jamming problem is caused by collisions betweentransmissions from Node 0 to Node 1 and the transmissions from Node 2 to Node 3. In this section,we will first build an analytical model for the above simulation topology (shown in Figure 1) tocalculate the collision probability of transmission from Node 0, and the probability that the numberof transmission attempts of Node 0 exceeds the maximum retry count (Rmax) or the probabilitythat Node 0 triggers route discovery procedure. Then, we will analyze the factors that determinethose probabilities and study how to reduce those probabilities.

Let Node 2 start to transmit at time t=0, and Node 0 start to transmit at time t0. For simplicity,we assume that 0<t0<T . Here, T denotes the time when Node 2 completes transmitting the firstdata frame. Figure 4 shows the DCF procedure defined in IEEE 802.11 according to Figure 1. In

DNode2

Node3

Node0

W 0

W i: Contention Window R: RTS Frame

DR

SIFS SIFSSIFS

W1

W 2 W 3

C: CTSFrame

A: ACKFrame

t

t0

0

D: Data Frame

CAC

R

0

0

t

t

tout: RTS Timer

tout

DIFS

R R R R

Node1t

R R R R

0

R : RTS Collision

T

Figure 4. IEEE 802.11 DCF procedure according to Figure 1.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 8: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

944 Y. LI ET AL.

Figure 4, the four RTS frames sent by Node 0 are corrupted at Node1 because they collide withthe transmission from Node 2.

Let Y [n] be the stochastic process representing the time at which Node 0 starts transmitting thenth RTS frame and T [n] be the stochastic process representing the time at which Node 2 startstransmitting the nth RTS frame. According to Figure 4, we have:

Y [n] =

⎧⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎩

t0, n=1

t0+(n−1)×tout+n−1∑j=1

Wj , 1<n�m

t0+(n−1)×tout+m∑j=1

Wj +(n−1−m)×Wm, m<n�Rmax

(1)

T [n] =

⎧⎪⎨⎪⎩0, n=1

(n−1)×tb+n−1∑j=1

W0, n>1(2)

Let w=CWmin. In Equations (1) and (2), Wi =uniform[0,2iw−1] is the selected value of thebackoff timer when a node retransmits a given RTS frame for the i th times. It is a random variablethat uniformly distributes between 0 and 2i ×w−1 if i<m, or uniformly distributes between 0and 2mw−1 if i�m. Here, m is the maximum backoff stage tRTS is the time that an RTS frameis sent, tout is the value of RTS timer and tout= tRTS+ tCTS+2×tSIFS. tSIFS is the SIFS time,tD and tA correspond to the data frame transmitting time and ACK frame transmitting time,respectively. tb is the time spent on successful transmission of a data packet in RTS/CTS mode,and tb= tRTS+ tCTS+ tDIFS+3 ·tSIFS+ tD+ tA.

From Figure 4, we can infer that the condition that Node 0 successively transmits an RTS frameat nth times is that Node 0 starts transmitting this RTS frame after T [i]− tRTS− tDIFS−W0 andfinishes transmitting it before T [i]− tSIFS, and vice versa. Figure 5 gives an example for Node 0to successively transmit an RTS frame. Thus, Node 0 can successively transmit an RTS frame atnth times if and only if:

T [i]−W0− tDIFS− tRTS<Y [n]<T [i]− tRTS− tSIFS ∀i ∈[N1,N2] (3)

In expression (3), N1 is the possible minimum value of i for a given n, which is achieved whenY [n] is minimum and W0 is exactly set to (CWmin−1) for each backoff of Node 2. N2 is thepossible maximum value of i for a given n, which is achieved when Y [n] is maximized and W0is exactly set to 0 for each backoff of Node 2. The expression of N1 and N2 are as follows:

N1 =⌈

(n−1)×tout+ t0tb+(w−1)

N2 =

⎧⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎩

⌊(n−1)×tout+∑n−1

k=1(2kw−1)+ t0

tb

⌋, n�m

⌊(n−1)×tout+∑m

k=1(2kw−1)+∑n−1

k=m+1(2mw−1)+ t0

tb

⌋, m<n�Rmax

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 9: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 945

DNode2

Node3

W0 Random Value of Backoff Timer R RTS Frame

R

SIFS SIFSSIFS

C emarFSTC ACKFrameD Data Frame

AC

R

t

t

DIFS

tACK

tRTStRTS

The time window to start to successfully send RTS frame for Node0.

T[i]T[i-1] W0

A

Figure 5. The condition that Node0 successively transmits a RTS frame.

where �x� denotes the minimum positive integer not less than x and �x� denotes the maximumpositive integer not greater than x . ∀i ∈[N1,N2] if (3) is satisfied, Node 0 can successively transmitan RTS at nth time. Let pn be the probability that Node 0 successively transmitting an RTS at nthtime. pn is calculated as follows:

pn =N2∑

i=N1

P{T [i]−W0− tDIFS− tRTS<Y [n]<T [i]− tRTS− tSIFS} (4)

Let Rmax be the maximum transmission time of a given RTS frame, and pRET be the probability thatthe number of Node 0’s transmission attempts for a given packet reaches to Rmax, then we have:

pRET=Rmax∏n=1

(1− pn) (5)

In the following, we calculate pn and pRET. According to the assumption made for t0(0<t0<T ),p1=0. Let tn = t0+(n−1)×tout. Combining Equations (1), (2) and (4), we obtain:

pn =

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

0, n=1

N2∑i=N1

{P

{n−1∑j=1

Wj<(i−1)×tb+i−1∑j=1

W0− tRTS− tSIFS− tn

}

−P

{n−1∑j=1

Wj +W0<(i−1)×tb+i−1∑j=1

W0− tDIFS− tRTS− tn

}}, 2�n�m

N2∑i=N1

{P

{m∑j=1

Wj +n−1∑

j=m+1Wm<(i−1)tb+

i−1∑j=1

W0− tRTS− tSIFS− tn

}

−P

{m∑j=1

Wj +n−1∑

j=m+1Wm+W0<(i−1)tb

+i−1∑j=1

W0− tDIFS− tRTS− tn

}}, m<n�Rmax

(6)

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 10: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

946 Y. LI ET AL.

For a given pair of values for i and n, we define the following stochastic variables, Xn =∑n−1j=1Wj , Yn =∑n−1

j=1Wj +W0, Zi =∑i−1j=1W0, Un =∑m

j=1Wj +∑n−1j=m+1Wm , Vn =∑m

j=1Wj +∑n−1j=m+1Wm+W0. Let �Xn , �Yn , �Zi , �Un and �Vn be the sample spaces of Xn , Yn , Zi , Un and

Vn , respectively, then �Xn =[0, (2n−2)w−n+1], �Yn =[0,2nw−w−n], �Zi =[0, (i−1)(w−1)],�Un =[0, (2m(n−m+1)−2)w−n+1], �Vn =[0, (2m(n−m+1)−1)w−n]. Here, Xn , Yn , Zi , Unand Vn are introduced to facilitate the performance analysis in terms of pRET and are actually thefunctions of CW Wi .

Moreover, let pkx , pky , prz , pku and pkv be the probability distributions of Xn , Yn , Zi , Un andVn , respectively, then

pn =

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

0, n=1

N2∑i=N1

( ∑r∈�Zn

prz · ∑k∈�Xn ,k<(i−1)·tb+r−tRTS−tSIFS−tn

pkx

)

−N2∑

i=N1

( ∑r∈�Zn

prz · ∑k∈�Yn ,k<(i−1)·tb+r−tDIFS−tRTS−tn

pky

), 2�n�m

N2∑i=N1

( ∑r∈�Zn

prz · ∑k∈�Un ,k<(i−1)·tb+r−tRTS−tSIFS−tn

pku

)

−N2∑

i=N1

( ∑r∈�Zn

prz · ∑k∈�Vn ,k<(i−1)·tb+r−tDIFS−tRTS−tn

pkv

), m<n�Rmax

(7)

Using the probability distributions of Wi , and with the method of probability generating function,we obtain:

pkx = 1

2n(n−1)/2 ·wn−1·

l∑s=0

cs ·(n−1+k−1−2s ·w

n−2

)(8a)

where k∈[2l ·w,2l ·w+2w−1], l∈[0,2n−1−2]; c0=1,c1=−1 and cs =−1 ·cs−2[log2 s] when s�2.

pky = 1

2n(n−1)/2 ·wn·

l∑s=0

cs

(n+k−1−s ·w

n−1

)(8b)

where k∈[l ·w, l ·w+w−1], l∈[0,2n−2]; c0=1,c1=−1 and cs =−1 ·cs−2[log2 s] when s�2.

prz = 1

wi−1·l−1∑s=0

(−1)s ·(i−1

s

)·(i−1+r−1−s ·w

i−2

)(8c)

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 11: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 947

where r ∈[(l−1)w, l ·w−1], l∈[1, i−1].

pku = 1

2m(2n−m−1)/2 ·wn−1·

l∑s=0

cs

(n−1+k−1−2sw

n−2

)(8d)

where k∈[2l ·w,2l ·w+2w−1], l∈[0, (n−m+1)2m−2]; c0=1, c1=−1, cs =−1 ·cs−2[log2 s]when s∈[0,2m−1−1], and cs =(−1) j ·( n−m

j ) ·cs− j ·2m−1 when s∈[ j ·2m−1, ( j+1) ·2m−1−1], j ∈[1,n−m].

pkv = 1

2m(2n−m−1)/2 ·wn·

l∑s=0

cs

(n+k−1−s ·w

n−1

)(8e)

where k∈[l ·w, l ·w+w−1], l∈[0, (n−m+1)2m−2]; c0=1, c1=−1, cs =−1 ·cs−2[log2 s] whens∈[0,2m−1], and cs =(−1) j ·( n−m

j ) ·cs− j ·2m when s∈[ j ·2m, ( j+1) ·2m−1], j ∈[1,n−m].Given the parameters of IEEE 802.11 DCF, i.e. minimum contention window (CWmin),

maximum backoff stage (m), the length of RTS, CTS and ACK, and the channel bandwidth, for agiven number of contention stations (n) and the length of data packet, we can get pkx , pky , prz ,pku and pkv according to Equations (8a), (8b), (8c), (8d) and (8e), respectively. With the value ofpkx , pky , prz , pku and pkv , we can get pn using Equation (7). Finally, we can obtain pRET usingEquation (5).

3.3. Model validation

In this sub-section, we validate the analytical model by comparing the theoretical pRET calculatedaccording to Equation (5) with the simulated pRET collected from simulations. We also evaluatethe relationship between pRET and the length of data frame, CWmin, and Rmax.

As shown in Figures 6–8, theoretical and simulation results are very close to each other. Therelationship between pRET and the length of data frame, CWmin and Rmax, is shown in Figures 6–8.We conclude that pRET can be decreased through transmitting shorter MAC data frame, increasingthe minimum contention windows or increasing maximum retransmission count. However, theeffects of these schemes on decreasing pRET are limited. In other words, they cannot lead to a verysmall pRET. When the length of an MAC data frame is 240 bytes, which means that the lengthof data packet encapsulated in the data frame is 160 bytes, pRET decreases only to 0.34, whichis still a large value and will result in the frequent route discovery process. At the same time,reducing the length of MAC data frame also reduces the payload in the frame, which will increasethe relative overhead of MAC protocols and decrease the utilization ratio of the wireless channel.When the minimum contention window increases to 128, pRET is decreased to 0.23, which also islarge. Even if the maximum retransmission count of a given RTS frame increases to 16, the valueof pRET is still large (0.46), which will prolong the access delay. Hence, we have to design othermethods to reduce pRET.

In the analysis, we find that by increasing the random delay between two consecutive trans-missions of Node 2, the value of pRET can be largely reduced. Note that Node 2 is the hiddenterminal of Node 0. In Figure 4 and Equation (3), W0 is the random delay between two consecutivetransmissions of Node 2. As W0 is uniformly set in range of [0, CWmin−1], we can statisticallyincrease W0 by increasing CWmin of Node 2. In order to show this, Figure 9 depicts the relationshipbetween pRET and W0. We can see that pRET reduces exponentially with the increase of Node 2’s

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 12: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

948 Y. LI ET AL.

Figure 6. pRET vs data-frame-length; CWmin=32, CWmax=256, Rmax=7.

Figure 7. pRET vs CWmin; CWmax=256, Rmax=7, length-of-data-frame=1024 Byte.

CWmin. When the value of Node 2’s CWmin equals to 256, pRET becomes 0.08. Figure 9 enlightensus to decrease pRET through increasing the value of CWmin of hidden terminals, e.g. Node 2 inFigure 1, and accordingly solve the jamming problem. In the next section, we will provide A-DCFfor adaptively adjusting the value of CWmin of hidden terminal in 802.11-based mobile ad hocnetworks.

4. ADAPTIVE DCF (A-DCF)

The basic idea of A-DCF is to let hidden terminals increase their minimum contention windowin order to statistically increase the backoff time between two consecutive transmissions, which

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 13: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 949

Figure 8. pRET vs Rmax; CWmin=32, CWmax=256, length-of-data-frame=1024 Byte.

Figure 9. The relationship between pRET and the value of CWmin of Node 2, CWmin=32, CWmax=256,Rmax=7, length-of-data-frame=1024 Bytes.

in turn increases the successive access probability of jammed terminals. But first of all, we needto identify the station that belongs to a hidden terminal. We divide A-DCF into two parts, hiddenterminal judgment and adaptive adjustment of CWmin.

4.1. Hidden terminal judgment

In the literature, hidden-terminal problem is described as that for a given transmission from Nodea to Node b, if Node c can sense the transmission of Node b, but cannot sense the transmission of

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 14: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

950 Y. LI ET AL.

Node a, Node c is hidden from Node a. According to this definition of hidden terminal, we givea simple hidden terminal judgment algorithm as follows. If node n senses a CTS frame from nodey to node x , but does not sense an RTS frame from node x to node y, then node n determinesthat it is hidden from the transmission from x to node y.

4.2. Adaptive adjustment of CWmin

Once a station n identifies itself as a hidden terminal, it adaptively adjusts its minimum contentionwindow according to the algorithm shown in Figure 10. In Figure 10, the CWmin of hidden terminal

Figure 10. A-DCF algorithm.

0 20 40 60 80 100 1200

500

1000

1500

Simulation Time(s)

Thr

ough

put(

Kbp

s)

F1F2

Figure 11. Throughput of F1 and F2 for A-DCF.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 15: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 951

60 60.02 60.04 60.06 60.08 60.1 60.12 60.14 60.16

0

1

2

Simulation Time(s)

Nod

e ID

66.22 66.24 66.26 66.28 66.3 66.32 66.34 66.36 66.38

0

1

2

Simulation Time(s)

Nod

e ID

sendRTS RTS-COL sendCTS recvCTS sendDATA recvACK RET send-route-req

recv-route-rep

(a)

(b)

Figure 12. Events extracted from the simulation trace file for A-DCF: (a) both Node 0 and Node 2 cansuccessfully transmit packets after F2 starts at 60 s and (b) RET event occurring at Node 0.

Figure 13. The throughputs of F1 and F2 on the conditions that channel are saturated, thedata rates of both F1 and F2 are 4.6Mbps.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 16: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

952 Y. LI ET AL.

Figure 14. The throughputs of F1 and F2 on the condition that channel is saturated, the data rates of F1and F2 are 460 kbps and 4.6Mbps, respectively.

is dynamically adjusted. When node n is jamming other nodes and arbitrates that it is a hiddenterminal, it increases its CWmin. If node n has successively transmitted � data frames and alwayssenses an RTS frame for each corresponding sensed CTS during this period, node n concludesthat it is not jamming any other node to transmission and starts to decrease its CWmin until thevalue of CWmin equals its default value defined in IEEE 802.11 DCF [7]. � in Figure 10 is calledas adjustment parameter. On the one hand, the bigger the value of �, the higher the successiveaccess probability of jammed stations will be achieved. On the other hand, too big value of � canmake the throughput of hidden terminal less than that of jammed stations. As a result, we give theupper bound of �, �max.

5. PERFORMANCE EVALUATION

This section evaluates the performance of A-DCF through simulation. In order to compare withthe results in Section 3, the simulation configurations of Section 3 are used here unless explicitlystated.

5.1. The fairness of A-DCF

In the first experiment, the simulation topology is the same as that in Figure 1. The simulationtime is still 120 s. F1 and F2 start at 1 and 60 s, respectively. We evaluate the throughput of F1 andF2, which is shown in Figure 11. We collected all events in the trace file and depict partial eventsafter F2 starts in Figure 12. Compared with Figure 2, Figure 11 shows that A-DCF makes F1 andF2 fairly share channel bandwidth after F2 starts. From the trace file, we find that RET event atnode 0 happens only three times after time t=60s, which are 66.3548, 83.5015 and 111.5853 s,respectively. However, after RET events, node 0 can always quickly find the route to node 1 andsuccessfully send data packets to node 1. Thus, A-DCF not only decreases the probability of RETevents but also makes jammed nodes recover from RET events quickly. Figure 12 shows the events

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 17: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 953

Figure 15. The throughputs of F1 and F2 on the condition that channel is saturated, the data rates of F1and F2 are 4.6Mbps and 460 kbps, respectively.

occurring around 66.3548 s. The events that occur around 83.5015 and 111.5853 s are similar toFigure 12 and are not presented in this paper.

In another experiment, both F1 and F2 start at t=1s. Then F1 stops from 40 to 80 s. Thisexperiment is used to evaluate the throughput of A-DCF under saturated and/or unsaturatedsituation, respectively. The simulation results are shown in Figures 13–16. Figures 13–15 showthat F1 and F2 can fairly share the channel when the wireless channel is saturated. Figures 13–15also show that when the data sending rate of a node is less than its bandwidth share, its throughputis nearly equal to the sending rate and other nodes can fully utilize the remaining bandwidth, whichilluminates that A-DCF can fully utilize channel bandwidth. Figure 16 shows that A-DCF is alsoeffective when channel is unsaturated. In simulation, we also collected CWmin of Node 2 duringthe time interval when F1 stops from 40 to 80 s. As shown in Figure 17, Node 2 can adaptivelyincrease its CWmin after F1 starts, and decrease its CWmin after F1 stops.

The above simulation results illustrate that A-DCF can effectively solve the jamming problemin 802.11-based mobile ad hoc networks with improved channel utilization.

5.2. Comparison with traditional IEEE 802.11 DCF

In this experiment, a linear topology with seven nodes is used (see Figure 18). A CBR flowis configured from Node 0 to Node 6 and the channel is saturated. The PHY layer uses DSSSand the routing protocol is AODV. Table I gives the values for some related parameters. Thesimulation time is set to 80 s and the CBR flow starts at 1 s. We compute the instantaneousthroughput of CBR, which is defined as all the data successfully received by Node 6 every 1 s.Figures 19 and 20 show the instantaneous throughput of CBR for conventional DCF and A-DCF, respectively. From Figures 19 and 20, we can conclude that the performance of A-DCF isobviously better than the conventional DCF. Furthermore, Table II shows their average throughputover the whole simulation time. On average, A-DCF achieves 30% higher throughput than theconventional DCF.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 18: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

954 Y. LI ET AL.

Figure 16. The throughputs of F1 and F2 on the condition that channel is unsaturated, thedata rates of both F1 and F2 are 460 kbps.

Figure 17. CWmin of Node 2: (a) the CWmin of Node 2 from 0 to 120 s;(b) the CWmin of Node 3.0 from 0 to 3.2 s; (c) the CWmin of Node 2 from

40.8 to 41.1 s; and (d) CWmin of Node 2 from 80.05 to 80.2 s.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 19: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 955

Figure 18. A linear topology consisting of seven nodes.

0 10 20 30 40 50 60 70 800

50

100

150

200

250

300

350

400

Simulation Time(s)

Thr

ough

put(

Kbp

s)

Conventional DCF,CWmin = 16Conventional DCF,CWmin = 32Conventional DCF,CWmin = 64

Figure 19. The instantaneous throughput of CBR for conventional DCF.

0 10 20 30 40 50 60 70 800

50

100

150

200

250

300

350

400

Simulation Time(s)

Thr

ough

put(

Kbp

s)

Adaptive DCF, = 1Adaptive DCF, = 2Adaptive DCF, = 4Adaptive DCF, = 6

Figure 20. The instantaneous throughput of CBR for A-DCF.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 20: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

956 Y. LI ET AL.

Table II. The average throughput over simulation time of A-DCF and conventional DCF.

Conventional DCF A-DCF

CWmin=16 CWmin=32 CWmin=64 �=1 �=2 �=3 �=4 �=6

Average throughput (kbps) 201 207 203 280 273 298 293 281

6. CONCLUSIONS

In this paper, we introduce the jamming problem in 802.11-based mobile ad hoc networks. It isshown that if a node n is hidden from the communication from node x to node y, the transmissionof n will restrain the communication from x to y. An analytical model is set up to facilitate thequantitative study of the influence of the jamming on system performance and find essential reasonsthat the jamming problem exists and how it affects the performance. Then a novel algorithm namedA-DCF is designed as the solution for the jamming problem, which adaptively adjusts CWmin ofhidden terminals. As verified by the extensive simulations, A-DCF effectively defeats the jammingproblem and achieves higher throughput while keeping good fairness as well.

REFERENCES

1. Tobagi FA, Kleinrock L. Packet switching in radio channels: part II—the hidden terminal problem in carriersense multiple-access modes and the busy-tone solution. IEEE Transactions on Communications 1975; COM-23(12):1417–1433.

2. Deng J, Hass ZJ. Dual busy tone multiple access (DBTMA): a new medium access control for packet radionetworks. Proceedings of The International Conference on Universal Personal Communication, Florence, Italy,vol. 2, 1998; 973–977.

3. Karn P. MACA—a new channel access method for packet radio. Proceedings of the ARRL/CRRL’90, New York,U.S.A., 1990; 134–140.

4. Fullmer CL, Garcia-Luna-Aceves JJ. Floor acquisition multiple access (FAMA) for packet-radio networks.Proceedings of The ACM SIGCOMM’95, Cambridge, MA, U.K., 1995; 265–293.

5. Fullmer CL, Garcia-Luna-Aceves JJ. Solutions to hidden terminal problems in wireless networks. ComputerCommunication Review 1997; 27(4):39–49.

6. Bharghavan V, Demers A, Shenker S, Zhang L. MACAW: a media access protocol for wireless LAN’s. Proceedingsof the ACM SIGCOMM ’94, 1994; 212–225.

7. IEEE standard for wireless LAN medium access control (MAC) and physical layer (PHY) specifications, 1999Edition.

8. Cali F, Conti M, Gregori E. Dynamic tuning of the IEEE 802.11 protocol to achieve a theoretical throughputlimit. IEEE/ACM Transactions on Networking 2000; 8(6):785–799.

9. Bianchi G. Performance analysis of the IEEE 802.11 distributed coordination function. IEEE JSAC 2000;1(3):535–547.

10. Wu H, Peng Y, Long K, Chen S. Performance of reliable transport protocol over IEEE 802.11 wireless LAN:analysis and enhancement. Proceedings of the IEEE INFOCOM’02, New York, U.S.A., vol. 2, 2002; 599–607.

11. Wang Y, Bensaou B. Achieving fairness in IEEE 802.11 DFWMAC with variable packet lengths. Proceedingsof the IEEE GLOBECOM’01, Texas, U.S.A., vol. 6, 2001; 3588–3593.

12. Nandagopal T, Kim TE, Gao X, Bharghavan V. Achieving MAC layer fairness in wireless packet networks.Proceedings of the ACM MobiCom’00, Boston, MA, U.S.A., 2000; 87–98.

13. Xu S, Saadawi T. Does the IEEE802.11 MAC protocol works well in multihop wireless ad hoc networks? IEEECommunications Magazine 2001; 39(6):130–137.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 21: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

PERFORMANCE ANALYSIS AND ENHANCEMENT 957

14. Mangold S, Choi S, May P, Klein O, Hiertz G, Stibor L. IEEE 802.11e wireless LAN for quality of service(invited paper). Proceedings of the European Wireless, Florence, Italy, vol. 1, 2002; 32–39.

15. Jie H, Devetsikiotis M. Designing improved MAC packet schedulers for 802.11e WLAN. Proceedings of theIEEE GLOBECOM’03, San Francisco, U.S.A., vol. 1, 2003; 184–189.

16. Kong Z, Tsang DHK, Bensaou B, Gao D. Performance analysis of IEEE 802.11e contention-based channelaccess. IEEE JSAC 2004; 22(12):2095–2106.

17. Hui J, Devetsikiotis M. A unified model for the performance analysis of IEEE 802.11e EDCA. IEEE Transactionson Communications 2005; 53(9):1498–1510.

18. Chatzimisios P, Boucouvalas AC, Vitsas V. IEEE 802.11 packet delay: a finite retry limit analysis. Proceedingsof the IEEE Globecom’03, San Francisco, U.S.A., vol. 2, 2003; 950–954.

19. Zanella A, Pellegrini F. Statistical characterization of the service time in saturated IEEE 802.11 networks. IEEECommunications Letters 2005; 9(3):225–227.

20. Xiao Y. Performance analysis of priority schemes for IEEE 802.11 and IEEE 802.11e wireless LANs. IEEETransactions on Wireless Communications 2005; 4(4):1506–1515.

21. Ho T, Chen K. Performance analysis of IEEE 802.11 CSMA/CA medium access control protocol. Proceedingsof the PIMRC’96, Taipei, Taiwan, vol. 2, 1996; 407–411.

22. Weinmiller J, Schlager M, Festag A, Wolisz A. Performance study of access control in wireless LANs: IEEE802.11 DFWMAC and ETSI RES 10 HIPERLAN. Mobile Networks and Applications 1997; 2(1):55–67.

23. Wang G, Shu Y, Zhang L, Yang O. Delay analysis of the IEEE 802.11 DCF. Proceedings of the IEEE PIMRC’03,Beijing, China, vol. 2, 2003; 7–10.

24. Ahn GS, Andrewl TC, Veres A, Sun LH. Supporting service differentiation for real-time and best-effort trafficin stateless wireless ad hoc networks (SWAN). IEEE Transactions on Mobile Computing 2002; 1(3):192–207.

25. Khurana S, Kahol A, Jayasumana AP. Effect of hidden terminals on the performance of IEEE 802.11 MACprotocol. Proceedings of the LCN’98, Boston, MA, U.S.A., 1998; 12–20.

26. Papanastasiou S, Lewis MM, Mohamed OK. Reducing the degrading effect of hidden terminal interference inMANETs. Proceedings of the MSWiM’04, Venice, Italy, 2004; 311–314.

27. Tsertou A, Laurenson DI. Insights into the hidden node problem. Proceedings of the ACM IWCMC’06, Vancouver,Canada, 2006; 767–772.

28. Tsertou A, Laurenson DI. Modeling the effect of BEB for a hidden terminal topology from a new perspective.Proceedings of the 3rd Annual IEEE Communications Society on Sensor and Ad Hoc Communications andNetworks, New York, U.S.A., vol. 2, 2006; 607–614.

29. Wu H, Zhu F, Zhang Q, Niu Z. Analysis of IEEE 802.11 DCF with hidden terminals. Proceedings of The IEEEGlobecom’06, Catalunya, Spain, 2006; 1–5.

30. Vassis D, Kormentzas G. Throughput analysis for IEEE 802.11 ad hoc networks under the hidden terminalproblem. Proceedings of the 3rd IEEE Consumer Communications and Networking Conference, San Francisco,U.S.A., vol. 2, 2006; 1273–1276.

31. Azafindralambo T, Valois F. Stochastic behavior study of backoff algorithms in case of hidden terminals.Proceedings of the 17th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications,Las Vegas, NV, U.S.A., 2006; 1–6.

32. Kim Y, Yu J, Choi S, Jang K. A novel hidden station detection mechanism in IEEE 802.11 WLAN. IEEECommunications Letters 2006; 10(8):608–610.

33. Li FY, Kristensen A, Engelstad PE. Passive, active hidden terminal detection in 802. 11-based ad hoc networks.Proceedings of the IEEE INFOCOM’06, Poster and Demo Session, 2006.

34. Thuente D, Acharya M. Intelligent jamming in wireless networks with applications to 802.1 lb and other networks.Proceedings of the MILCOM’06, Helsinki, Finland, 2006.

35. Acharya M, Thuente D. Intelligent jamming attacks, counterattacks and (counter)2 attacks in 802.11b wirelessnetworks. Proceedings of the OPNETWORK’05, Washington, DC, U.S.A., 2005.

36. David JT, Newlin B, Acharya M. Jamming vulnerabilities of IEEE 802.lle. Proceedings of the MILCOM’07,Washington, DC, U.S.A., 2007.

37. Acharya M, Sharma T, Thuente D, Sizemore D. Intelligent jamming in 802.1 lb wireless networks.OPNETWORK’04, Washington, DC, U.S.A., 2004.

38. http://www.isi.edu/nsnam/.39. Perkins CE, Belding-Royer EM, Das S. Ad hoc on demand distance vVector (AODV) Routing. IETF RFC 3561.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac

Page 22: The Jamming problem in IEEE 802.11-based mobile ad hoc networks with hidden terminals: Performance analysis and enhancement

958 Y. LI ET AL.

AUTHORS’ BIOGRAPHIES

Yun Li was born in Sichuan China in 1974. He received his BA and MA degrees from theSouthwest University and the Chongqing University of Posts and Telecommunications,China, in 1997 and 2000, respectively, and the PhD degree from the University ofElectronic Science and Technology, China, in 2004. Dr Li works as a full professor inthe Chongqing University of Posts and Telecommunications. He is also a post doctorfellow in Southeast University of China. He has published about 100 refereed journaland conference papers.

Chonggang Wang ([email protected]) received his PhD degree in Computer Sciencefrom Beijing University of Posts and Telecommunications, China, in 2002. His currentresearch interests include wireless and mobile networks, multimedia communicationsprotocols, and Internet protocols. He is currently an associate technical editor of IEEECommunications Magazine and a member of IEEE.

Weiliang Zhao was born in 1962 and received his PhD degree from the University ofElectronic Science and Technology of China in 2001. He is a professor and a post-doctoral fellow in Beijing University of Posts & Telecommunications of China. Hiscurrent interest lies in wireless communications.

Professor Xiao-Hu You is now with the National Mobile Communications ResearchLaboratory at Southeast University, Nanjing, P. R. China. From 1999 to 2006, he wasthe Principal Expert of the C3G Project and B3G FuTURE Project, responsible fororganizing Chinas 3G and beyond Mobile Communications R&D Activities.

Copyright q 2009 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2009; 22:937–958DOI: 10.1002/dac