Saturation Throughput Analysis of IEEE 802.11b DCF Considering Capture Effects Haihua Han School of Information Science and Engineering Shandong University Jinan Shangdong 250100, P.R.China hanhaihua204@126.com Weihong Zhu Na Li School of Information Science and Engineering Shandong University Jinan Shangdong 250100, P.R.China zhuwh@sdu.edu.cn School of Information Science and Engineering Shandong University Jinan Shandong 250100, P.R.China nazheng708@126.com Abstract—The performance of the distributed coordination function (DCF) has been discussed in the past years. This paper provides a simple and accurate analysis by Markov chain modeling to evaluate the IEEE 802.11 DCF performance in the presence of capture effect. We extend the Bianchi’s multidimensional Markov chain analytical model of IEEE 802.11 DCF to characterize the behavior of DCF considering capture effect and packet retransmission limits. We compare the NS-2 simulation results with the derived analysis. Numerical results and simulations are provided to validate the accuracy of our model. Keywords- IEEE 802.11 DCF; NS-2; Markov chain; capture effect; packet retry limits I. INTRODUCTION The IEEE802.11 protocol is widely used in today’s wireless local area networks (WLANs) and multi-hop mesh networks. The 802.11 standard for wireless networks incorporates two kinds of medium access methods, which are the distributed coordination function (DCF) and the point coordination function (PCF). PCF is a centralized and optional protocol, which uses the pooling mechanism for gaining the access of channel. The DCF is a mandatory protocol, which is based on the carrier sense multiple access with collision avoidance (CSMA/CA) algorithm. In 802.11, The DCF defines two medium access schemes for packet transmission, the basic access mechanism and the request-to-send /clear-to -send (RTS/CTS) mechanism. The basic scheme is a default mechanism and uses DATA/ACK, a two-way handshaking technique, to confirm whether the DATA packet is successfully transmitted in the channel. The other optional RTS/CTS mechanism is a four-way handshaking access method, which uses smaller control packets (RTS and CTS) to reserve the channel before data transmission. Many researchers have been studied the performance of 802.11 DCF based on two dimensional Markov chain analytical model. As far as know, G. Bianchi[1] first proposed a two-dimensional Markov chain model to analyze the saturation throughput of the DCF, in the assumption of a limited number of terminals and ideal channel conditions (no errors). In contrast, Wu et al. [2] extended the Bianchi’s model with consideration of finite packet retry limits specified by the IEEE 802.11 standards. In Wu’s model, the station discards the packet after m back-off stages, while in Bianchi’s model, the station keeps iterating in the m-th back-off stage until the packet is successfully transmitted. The result is the throughput of 802.11 can be predicted more accurately. In [3], the authors introduced another state (idle) in which the station buffer is waiting for a new packet after a successful completion of a packet transmission to investigate the performance of DCF considering packet retry limits in the unsaturated traffic conditions. When two or more stations simultaneously transmit to an Access Point (AP), the station whose power is sufficiently stronger than the other stations captures the channel, and thus, the actual throughput increases. This phenomenon is known as capture effect. Thus, the capture effect widely exists in the IEEE 802.11 WLANs. The authors [4, 5, 6, 7] presented a derived Markov model to analyze the throughput of IEEE 802.11 taking into account transmission errors and capture effects over Rayleigh fading channel in saturated and non-saturated traffic conditions. Moreover, paper [8] analyzed the saturation throughput of the basic 802.11 protocol by including the capture effect in Rayleigh fading environment. In this paper, we present a simple analytical model to characterize the performance of IEEE802.11 DCF, which takes into account packet retry limits and capture effect in saturated traffic conditions. The key difference between our model with the others which the above papers presented is that the Markov chain models are different, where our model considers the effects of frame retransmitting limit. The rest of the paper is organized as follows. Section II briefly overviews the DCF procedure of the IEEE802.11 MAC protocol which includes both basic access and RTS/CTS mechanism. Our analysis model for the basic access mechanism under saturated traffic condition is described in Sections III, Section IV presents numerical and simulation results where typical MAC layer parameters for IEEE802.11b are used. Finally, Section V concludes the paper. II. IEEE 802.11 DISTRIBUTED COORDINATION FUNCTION(DCF) PROCEDURE The DCF, a basic MAC layer mechanism of IEEE 802.11 is based on CSMA/CA protocol to share medium between multiple stations. The DCF protocol in IEEE802.11 includes two access techniques for a packet transmission: the basic access mechanism and an optional RTS/CTS mechanism as shown in Fig.1 and Fig.2, respectively. A. The basic access method Before transmitting a new data packet, the station (STA) must gain access to the medium, which is a radio channel that all stations share. If the channel is idle for a period of duration equal to a distributed inter-frame space (DIFS), the STA will transmit its packet. If the channel is sensed to be busy, the STA will delay a DIFS and then generate a random back-off interval before transmitting the new packet. The back-off interval counter is decremented by one as long as the medium is idle, or it is frozen when the channel is sensed to be busy, and resumes after the channel is sensed to be idle again for a DIFS time. When the back-off timer reaches zero and the channel is idle, the station will send its data packet. Once the transmitting packet is received correctly, the destination STA will send an acknowledgment (ACK) frame to the source STA after a short inter-frame space (SIFS) interval. If the source STA does not receive the ACK, the data packet is assumed to be lost and a retransmission is scheduled. The random back-off interval is uniformly chosen in the interval [0, Wi -1] , where Wi is the current Contention Window (CW) and i is the station back-off stage. The value of Wi depends on the frame retransmissions and is an integer in the range W0 Wi Wm' . At the first transmission attempt of a packet, this value is set to be W0 CWmin , which is called the minimum contention window size. After each unsuccessful transmission, it is doubled, up to a maximum value Wm' CWmax 2m W0 , where m' represents the back-off stage ' Figure 2. III. RTS/CTS mechanism ANALYY ICAL MODEL AND THROUGHPUT ANALYSIS FOR DCF We use the two-dimensional random process s(t ), b(t ) to represent the sate of the two-dimensional Markov chain at a slot time t , therein, b(t ) is the back-off timer for a station, s (t ) is the back-off stage [0,......,m] for the station, where m represents the maximum back-off stage or the station short retry count (SSRC)[9], which is the maximum number of retransmissions for the RTS frame or the DATA frame when the RTS/CTS is not used. When the SSRC limits is reached, retry attempts will cease and the frame will be discarded. In this model, we suppose m is equal to 7, and each station always has a packet available for transmission. The key assumption in our model is that the collision probability Pcol and capture probability Pcap is constant and independent for each transmitted packet. For the sake of simplicity, we adopt the same notation P{i1 , k1 | i0 , k0 } P{s(t 1) i1 , b(t 1) k1 | s(t ) i0 , b(t ) k0 } as used in reference [1]. Our Markov chain is shown in fig. 3. at which the CW size reaches the maximum value. When Wi is equal to CWmax , it will be remain at the value of CWmax until it is reset. Therefore, if the station is in the i back-off stage, the CW size can be given as: i i m' Wi 2 W (1) ' m i m' Wi 2 W B. RTS/CTS mechanism The RTS/CTS mechanism tries to reserve the channel by smaller control packets (RTS and CTS) before the DATA transmission. When the back-off interval counter is zero and the medium is idle, the source STA sends a RTS frame to the destination STA and it will receive a CTS frame in response after a SISF time interval. If the source STA receives a CTS frame, the source STA starts transmitting its data. If the CTS is not received by the source STA, it is assumed that a collision occurred and a RTS retransmission is scheduled. Other procedures are same as the basic one. Figure 3. A Markov chain model for a station The state transition diagram for the Markov chain model has the following non-null one-step transition probabilities P{i, k | i, k 1} 1, k 0, wi 2, i 0, m P 0, k | i, 0 1 Pcol Pcol Pcap / W0 P{i, k | i 1, 0} Pcol (1 Pcap ) / Wi k 0,Wi 1 , k [0,Wi 1] i [0, m) i [1, m] (2) P{0, k | m,0} 1/ W0 Let bi , k lim t P{s(t ) i, b(t ) k}, i (0,m), k (0, Wi 1) Figure 1. basic access mechanism be the stationary distribution of the Markov chain. According to the one-step transition probabilities, the following relations can be established: bi ,0 bi 1,0 ( Pcol (1 Pcap )) 0 i m (3) bi ,0 b0,0 ( Pcol (1 Pcap )) i 0im (4) Because the chain is regular, for each k [0, Wi 1] , we have m 1 Wi k 1 Pcol Pcol Pcap b j ,0 bm ,0 , j 0 bi , k Wi Pcol 1 Pcap bi 1,0 , i0 (5) 1 i m Substituting (4) in (5), (5) can be rewritten as: W k bi ,k i bi ,0 , 0im (6) Wi Then, when the normalization condition for stationary distribution is applied, we have 1 Wi 1 m b i, k k 0 i 0 W 1i m b i ,0 i 0 k 0 Wi k Wi W 1 bi ,0 i 2 i 0 m From the normalization equation, the expression of b0,0 can be given as (7). , m m' , m m' where we assume Pt Pcol (1 Pcap ) 2(1 2Pcol )(1 Pcol ) , m m' m1 m1 W (1 (2Pcol ) )(1 Pcol ) 1 2Pcol (1 Pcol ) b0,0 2(1 Pt )(1 Pt ) , m m' ' ' ' ' m 1 W (1 (2Pcol ) )(1 Pcol ) 1 2Pcol 1 Pcolm1 W 2m Pcolm 1 (1 2Pcol )(1 Pcolmm ) (8) Which is similar to the one found in [2] under saturated load conditions. Now the probability τ, a randomly chosen slot time in which a station attempt transmission, can be expressed as, m i 0 m bi,0 Pt b0,0 b0,0 i 0 1 Pt m1 1 Pt (9) where b0,0 value can be attained from (7). The probability generally depends on the conditional collision probability Pcol and capture probability Pcap . In our model, we assume that capture effect is a subset of the collision events. Thus the conditional collision probability Pcol that packet collides with other packet transmission in a given time slot can be expressed as Pcol 1 (1 )n 1 Pcap (10) Our model considers power loss and multi-path fast fading of the transmitted signal. Suppose the signal is transmitted in Rayleigh fading channel, then, the received instantaneous signal power is exponentially distributed as: f x x 1 P0 e , P0 x0 i i here, the value of Pu / Pk is the ratio of the power, k 1 of i interfering frames, denoted by Pk . z0 is the capture ratio (the signal-to-interference power ratio identifying the capture threshold at the receiver), and g ( S f ) is the processing gain of inversely proportional to the spreading factor S f . For Direct Sequence Spread Spectrum (DSSS) modulation using a fixed 11-chip spreading factor ( S f 11 ), the g ( S f ) can be expressed as When Pcap 0 , then (7) can be rewritten as (8). Pcap( z0 g S f / i) 1 / 1 z0 g S f the correlation receiver. In fact, the processing gain introduces a reduction of interference power by factor g (S f ) , which is (7) n A ri p is the deterministic path loss. When two or more stations transmit simultaneously, the capture probability conditioned on i interfering frames can be defined as follows: denoted by Pu , of the detected frame to the sum of the power 2(1 2 Pt )(1 Pt ) W (1 (2 P ) m 1 )(1 P ) 1 2 P (1 P m 1 ) t t t t b0,0 2(1 2 Pt )(1 Pt ) ' ' ' m' 1 W (1 (2 Pt ) )(1 Pt ) 1 2 Pt 1 Pt m 1 W 2m Pt m 1 (1 2 Pt )(1 Pt m m ) where P0 denotes the local mean power determined by the equation: n P0 A ri p Pt n p is the path loss exponent, Pt is the transmitted power, and g (S f ) 2 3S f Probability also impacts the probability of the frame capture probability, Pcap can be obtained as following: n 1 Pcap (z0 , n) Ri Pcap( z0 g S f / i ) i 1 where, Ri is the probability of i interfering frames of n contending stations being generated in the observed time slot, according to n i 1 (1 ) n 1 i Ri i 1 Let Ptr be the probability that there is transmission occurs in a randomly chosen slot denote the successful transmission probability, sum of the conditional probability that the occurring on the channel is successful without with capture effect. Then we have Ptr 1 (1 )n n 1 n 1 Ps at least one time. Let Ps which is the transmission collision and (11) Pcap 1 (1 ) n (12) Now we are able to express the normalized system throughput S of the DCF as the ratio, Ptr Ps E[ P] S (13) (1 Ptr ) Ptr (1 Ps )Tc Ptr PsTs 0.7 Pcol ,z0=5dB Pcol ,z0=10dB 0.6 Pcol ,z0 τ,z0=5dB 0.5 Pcol ,z0 col ,τ 0.4 0.3 0.2 0.1 MODEL VALIDATION AND PERFORMANCE EVALUATION 5 10 15 20 25 30 number of stations 35 40 45 50 Figure 4. Theoretical behvior of the probability Pcol , 1 0.9 0.8 normalized throughput To validate this model, we have compared the theoretical results with that of obtained NS-2 simulation results. In this paper, we assume the NS-2 simulator considers a Basic Service Set (BSS) with an Access Point (AP) scenario in which n contending stations, which communicates only with the AP, are randomly distributed in a circular area of radius R (10 m) and the common access point is placed in the center of the transmitting area. The network parameters of IEEE 802.11b used in our simulations is summarized in table.1. Moreover, we assume C min 32 , C max 1024 and m 7 . In this section, we focus on presenting some simulation and theoretical results that show the impact of capture effect on the system capacity. τ,z0=10dB τ,z0=30dB 0 0.7 0.6 0.5 0.4 0.3 0.2 TABLE I. NETWORK PARAMETERS MAC header PHY header ACK ACK timeout τd Slot time SIFS DIFS Basic rate Data rate 24 bytes 16 bytes 14 bytes 300μs 1μs 20μs 10μs 50μs 1Mbps 1Mbps theo-no capture theo-z 0=6db 0.1 0 5 10 15 20 25 30 number of stations 35 40 45 50 Figure 5. normalized system throughput vs. numbers of contending stations 1 0.9 0.8 n1=5 no capture normalized throughput IV. Pcol ,z0=30dB P where Tc and Ts are the average time which the channel is sensed busy due to a packet collision and a successful packet transmission, respectively, and E[P] is the average packet length. For simplicity, the size of all packets is assumed to be same and fixed, and is the duration of an idle slot time. Let us now consider a system that each packet is transmitted by way of the basic access mechanism. Let H PHYhdr MAChdr indicate packet header, and d be the propagation delay. Then we can have the following expression. Tc PHYhdr MAChdr E[ P] ACK timeout Ts PHYhdr MAChdr E[ P] SIFS ACK DIFS 2 d 0.7 n2=20 no capture 0.6 n3=50 no capture 0.5 0.4 0.3 0.2 n1=5 0.1 n2=20 n3=50 0 0 5 10 15 20 capture thrshold z 0(dB) 25 30 Figure 6. normalized system throughput vs. capture thresholds The theoretical behavior of the probabilities Pcol , are depicted in Fig.4 for the basic access mechanism as a function of the number of contending stations, for four different capture thresholds z0 .The result shows that the transmission probability τ decreases as the value of capture thrshold and the contending stations increase, while the probability of collision Pcol increases. The curves of collision probability Pcol and transmission probability τ at z 0 30dB are very closed to the ones at z0 , this also means that when the capture threshold is greater than z 0 30dB , the collision probability Pcol and transmission probability τ will not change so much. In Fig.5, we can obtain the behavior of the saturation throughput for the 2-way mechnisam as a function of the number of contending stations, for the capture theshold z 0 6dB and without capture effect ( z0 ). Simulation results are marked by five-pointed star. Through comparing the curves shown in Fig.5, we easily find that the system throughput significantly increases under condition of capture effect, and with the number of contending stations increase, the throughput decreases. In order to further observe the impact of capture effect into different contending stations, we compare the throughput under three values of the number of contending stations, n1 5 , n 2 20 and n3 50 . Fig.6 shows the behavior of the saturation throughput as a function of capture thresholds z0 . From Fig.6 we can see that as the value of capture thresholds increases, the throughput for the three numbers of contending stations decreases, but it is greater than the throughput without capture effect. V. CONCLUSION In this paper, we have presented an analytical model to evaluate the performance of the IEEE802.11 DCF considering Rayleigh fading channel induced capture effect and packet retry limits under saturated conditions. Simulation and analysis results show that our analytical model can accurately predict the throughput performance of IEEE 802.11 DCF. REFERENCES [1] [2] [3] [4] [5] G.Bianchi, “Performance Analysis of the IEEE 802.11 Distributed Coordination Function ”,IEEE Journal on Selected Area in Communications, vol.18, no.3, pp. 535-547, March 2000. H. Wu, Y. Peng, K. Long, S. Cheng, J. Ma, Performance of reliable transport protocol over IEEE 802.11 wireless LAN: analysis and enhancement, in: INFOCOM 202 – 21th Annual Joint Conference of the IEEE Computer and Communications Societies, vol. 2, 2002, pp. 599–607. Y. S. Liaw, A. Dadej, A. Jayasuriya., “Performance analysis of IEEE 802.11 DCF under limited load”, In Proceedings of Asia-pacific Conference on Communications, pp.759–763, Perth, Western Australia, October 2005 F. Daneshgaran, M. Laddomada, F. Mesiti, and M. Mondin, “Saturation Throughput Analysis of IEEE 802.11 in Presence of Ideal Transmission Channel and Capture Effects”, IEEE Trans. Commun., vol. 56, no. 7, pp. 1178-1188, July 2008. F. Daneshgaran, M. Laddomada, F. Mesiti, M. Mondin.Unsaturated throughput analysis of IEEE 802.11 in presence of non ideal transmission channel and capture effects IEEE Transactions on Wireless Communication s,vol.7,no. 4, pp. 1276–1286, 2008. P. Kumar and A. Krishnan,“Throughput analysis of the IEEE 802.11 distributed coordination function considering erroneous channel and Considering Capture Effects”, Third International Conference on Emerging Trends in Engineering and Technology, pp. 836 – 841, Nov. 2010 [7] P. Kumar and A. Krishnan,“Saturation Throughput Analysis of IEEE 802.11b Wireless Local Area Networks under High Interference considering Capture Effects”, International Journal of Computer Science and Information Security(IJCSIS), Vol. 7, No. 1, pp.32-39, January. 2010. [8] P. Kumar and A. Krishnan,“Throughput Analysis of the IEEE 802.11 Distributed Coordination Function Considering Capture Effects”, Third International Conference on Emerging Trends in Engineering and Technology, pp. 836 – 841, Nov. 2010 [9] Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specification, IEEE Std. 802.11 1999. [10] The Network Simulator — NS-2, [Online], Available: http://www.isi.edu/nsnam/ns, December 18, 2010 [6]