The Performance of IEEE 802.11 RTS/CTS with Random Waypoint Mobility Nasser M. Sabah and Aykut Hocanin Department of Electrical and Electronic Engineering Eastern Mediterranean University Gazimagusa, TRNC, via Mersin 10, Turkey
[email protected] and
[email protected]
Abstract— Mobility model and other various factors such as the network size, the routing scheme and the traffic intensity may result in unpredictable variations in the overall network performance. However, nodes’ mobility is a major factor that contributes to topology changes in ad-hoc networks. In this paper, the IEEE 802.11 RTS/CTS handshake mechanism is implemented under fixed and mobile ad-hoc network scenarios. The random waypoint mobility model is deployed in the mobile ad-hoc network scenarios. We examine the system performance in terms of throughput, delay and retransmission rate for various numbers of nodes in the simulated area. In the case of the fixed ad-hoc network scenario, 1-hop transmission is adopted. In the mobile scenarios, 1-hop routing and multi-hopping routing schemes are employed. The disadvantage of the multi-hopping routing scheme is that it involves large delays. The simulation results show that mobility leads to a throughput enhancement and a higher path availability for the mobile ad-hoc networks with multi-hopping routing scheme compared to the mobile ad-hoc networks with 1-hop route. However, this comes at an expense of increased transmission delay. Keywords— IEEE 802.11, MANETs, RWP, mobility models.
1. I NTRODUCTION Mobile Ad hoc Networks (MANETs) is a collection of wireless mobile nodes forming a temporary network without the need for base stations or any other preexisting network infrastructure. Ad-hoc networking received a great interest due to its low cost, high flexibility, fast network establishment, selfreconfiguration, and high speed for data services. However, a wireless network without a fixed infrastructure and with nodes’ mobility enabled, the network topology keeps on changing. This causes frequent path changes and leads to increase the network congestion and transmission delay over the network. Most researches study the IEEE 802.11 model, assume a single-hop network. Usually they provide a bound for the packet transmission probability and the network throughput in the condition of traffic saturation [1], [2]. In [3], investigates the effectiveness of RTS/CTS handshake mechanism to reduce the interference between nodes. It shows that when the interference range is larger than the transmission range, the RTS/CTS cannot function properly. The authors in [4], show that the 802.11 RTS/CTS is not effective with long distance link in outdoor scenarios and propose an adaptive setting mechanism of the carrier sensing threshold to increase the performance. In [5], proposed a model to compute the network throughput of fixed ad hoc networks, where nodes are randomly located
in the network and each source node chooses a random destination. Moreover, every node acts simultaneously as a source, a destination and a relay for other source-destination (SD) pairs. The √ result shows that the throughput per (SD) throughput pair is Θ(1/ N log N ). Also it shows that the √ per (SD) pair decreases to approximately like 1/ N as the number of nodes increases per unit area. This is the best performance can be achieved, even if nodes are optimally placed in a unit area, traffic patterns are optimally assigned, and each transmission’s range is optimally chosen. In the fixed ad-hoc network model, the performance limitation is because of long direct transmission range communication between nodes pairs is infeasible due to the excessive interference caused. Therefore, most communication between√ neighbor nodes occurs at transmission range of order 1/ N . As a result, each packet is transmitted through many other relay nodes before reaching the destination. The number of hops in √ a typical route is of order N . In [6], proposed a model to compute the throughput of mobile wireless ad-hoc networks, a 2-hop relaying scheme is proposed. It shows that the proposed scheme can achieve a throughput capacity of Θ(1) per (SD) pair. This throughput remains constant as the number of nodes grows arbitrarily large. However, the strategy of choosing a short transmission range communication in mobility network scenario is inefficient. This is because the time fraction of two nodes to be in range is too small, of the order of 1/N . Instead, the efficient strategy is for each source node to broadcast its packet stream to its neighbor nodes if the destination node is out of range. These nodes keep in moving and serve as node relays and whenever they are in range with the destination node, they transmit the packet to the destination node. 2. D ISTRIBUTED C OORDINATION F UNCTION (DCF) The 802.11 IEEE standard is well-established as the medium access control protocol (MAC) for wireless local area networks (WLANs) and has been extensively studied in ad-hoc settings either through simulations or real hardware deployments [7], [8]. MAC protocol offers a contention service with stochastic bandwidth sharing used by the DCF. It is based on carrier sense multiple access with collision avoidance (CSMA/CA) scheme with rotating backoff for the distributed medium sharing. DCF defines two access mechanisms for packet transmission. The basic access mechanism (2-ways handshaking) and the RTS/CTS virtual carrier sensing mechanism (4-
ways handshaking). The IEEE 802.11 group integrated the RTS/CTS virtual carrier sensing mechanism developed in [9] and evaluated in detail in [10], [11]. The RTS/CTS is typically used in MANETs, this is because it increases the bandwidth efficiency by reducing the collision probability, and expends more bandwidth by transmitting two additional control packets per data packet transmission as shown in figure 1. DIFS
Source Station
Destin. Station
SIFS RTS
FRAME
CTS
ACK
SIFS
SIFS
Backoff Time
Neighbor Station NAV RTS NAV CTS
DIFS
Defer Access
Fig. 1. Carrier sense multiple access with collision avoidance (CSMA/CA) scheme with the RTS/CTS handshake mechanism [1].
Network connectivity is measured in terms of the connectivity ratio, which is defined as the ratio of pairs of connected nodes to the total number of pairs. In [12] algorithm, the connectivity ratio drops rapidly as the average speed increases. When the transmission range is 20m, the delivery ratio improves significantly under mobility of 1 − 40m/sec. The algorithm has almost %100 connectivity ratio with a transmission range of 100m and under high mobility of 160m/sec. Each node maintains a list of the reachable neighbors that is updated periodically. This is to successfully receive an interference-free transmission. The density is defined as the average number of neighbors in the simulated area, this depends on the fraction of the covered transmission area by the network size to the entire simulated area: πR2 N. (1) A where N is the number of nodes in the simulated area, R is the transmission range and A is the entire simulated area. Furthermore, the neighbors’ list is used for route discovery to enhance packet delivery. It is the best method for a successful transmission and for broadcasting a frame to a subset of an updated neighbor list. Additionally, the network topology will remain effectively constant for appropriately chosen time intervals. The update interval of neighbors’ list is specified as a function of nodes’ speed v. Moreover, the gossip scheme with flooding is used to determine and updates the list of neighbors. Because the network topology of MANETs changes significantly over time due to nodes’ mobility, broadcasting is a fundamental communication primitive, essential to adhoc routing algorithms for route discovery [13]. Flooding Nnbr =
is a suited approach for MANETs for broadcasting, as it requires no topological knowledge. Each node rebroadcasts the received frame to its neighbors upon receiving it for the first time. 3. R ANDOM WAY P OINT M OBILITY M ODEL Mobility has a dramatic effect on the performance of MANETs due to nodes mobility, which in return affects the whole performance of MANETs in terms of efficiency, throughput and delay. In RWP mobility model, each node of the network is assigned an initial location (X0 , Y0 ), and destination waypoint (X1 , Y1 ) independently from a random uniform distribution. A node moves from the initial location to the destination point with a constant speed v and straight line. The speed is chosen randomly from a uniform distribution of a minimum speed and a maximum speed [Vmin , Vmax ]. Once the node reaches its destination, it pauses for a certain time and then selects a new random destination and speed, it repeats the whole procedure independently through the simulation period. RWP mobility model is simple and widely-used in many simulation studies of ad-hoc routing protocols [14]. The authors in [15] investigate and quantify the effects of various factors (node speed, node pause time, network size, number of traffic sources, routing protocol) and their two-way interactions on the overall performance of ad hoc networks using the factorial experiment design. The study uses a RWP mobility model, and shows that the nodes’ speed has an impact effects on control overhead and throughput, while pause time has no effect. In [16], a speed decay is noted in the RWP model as the simulation progresses, and may fail to reach the steady state in terms of instantaneous average speed of node. This is because the convergence time may exceed the simulation period. Furthermore, the study proposed setting minimum speed to a positive value to reach stability as fast as possible. 4. S IMULATION R ESULTS In this paper, Matlab package is used in simulating the RWP mobility model and the CSMA/CA scheme. The results for the RWP mobility model are obtained by taking the average of 30 distinct scenarios of the simulated time. Also the figures of CSMA/CA scheme are obtained by generating 10 sample runs for each data point. This is to have accurate results. Each node chooses an initial location and destination point independently from a random uniform distribution, where nodes are uniformly distributed and move independently in a square region of 1000m × 1000m simulation area. The speed is chosen randomly in the interval of [Vmin , Vmax ] and sampled from the designated distribution. The initial data observed at the first 500sec in the RWP mobility model (U ∼ [1, 19]) are disregarded in simulating the CSMA/CA scheme, as well the first 100sec in the Gamma RWP mobility model (Γ ∼ [1, 19]). This is to insure that the system enter the steady state. Pause time is set to zero to keep continues mobility. The Gamma RWP mobility model propose in [17]
outperforms the typical and the modified [16] RWP mobility models as shown in figure 2. It provides a significant performance improvement in terms of having a higher steady state speed and achieving faster convergence to the steady state. The average speed in the typical (U ∼ [0, 20]), the modified (U ∼ [1, 19]) and Gamma RWP mobility models (Γ ∼ [1, 19]) are about 4m/sec, 6m/sec and 9.67m/sec, respectively. 12
8
6 1 4
0.8
Speed = U ∼ [0−20] Speed = U ∼ [1−19] Speed = Γ ∼ [1−19]
2
0
1−hop, Stationary 1−hop, Γ ∼ [1−19] 1−hop, U ∼ [1−19] multi−hoping, Γ ∼ [1−19] multi−hoping, U ∼ [1−19]
0.9
0
500
Fig. 2.
0.7
1000 1500 2000 Simulation Time (sec)
2500
3000
Average node speed in RWP mobility model.
In the CSMA/CA scheme, a node generates a request at an arrival rate λ = 100 packets/sec to access the medium. The node transmits a unicast or a broadcast frame using the RTS/CTS (4-way handshake), this is in accordance with the ad-hoc network scenarios. The traffic is assumed to spread across the network if the destination node is out of range with the source node. Furthermore, the traffic spreads to as many relay nodes as possible, and delivered to the destination node as soon as any of the relaying nodes is in range with the destination node. Each node generates 512 byte of data packet size and transmits at a channel bit rate (CBR) of 11 Mbps to a random chosen destination node. During simulation, nodes’ movement is updated every 4sec in accordance to RWP mobility. Other parameters are listed in table 1. Table 1.
Validation parameter used
Basic bit rate (BBR) PHY header (PH) MAC header (MH) RTS CTS ACK SIFS DIFS Slot Time CWmin Backoff stage (m) Propagation Delay (τ ) Simulated Time Transmission Range
1Mb/s 192 bits 272 bits 160 bits 112 bits 112 bits 10µs 50µs 20µs 32 5 1µs 100s 250m
Throughout
Average node speed (m/s)
10
Figure 3 shows the average throughput of the RTS/CTS mechanism of variable number of nodes in the network. The minimum throughout is 0.24 and 0.027 at a network size of 5 nodes for the fixed and mobility network scenarios with multi-hopping routing scheme, respectively. The maximum throughput of about 0.7 is obtained at a network size of 20 nodes, this is for the fixed and mobility network scenarios with multi-hopping routing scheme. However, the mobility network scenarios with 1-hop route can achieve at most a maximum throughput of 0.038 at a network size of 50 nodes. Also, the results show that under mobility, the multi-hopping routing scheme can take advantage of nodes’ movement to achieve higher path availability and end-to-end throughput compared to the 1-hop routing scheme.
0.6 0.5 0.4 0.3 0.2 0.1 0
0
5
Fig. 3.
10
15
20 25 30 Number of Nodes
35
40
45
50
Average Throughout of RTS/CTS mechanism.
Figure 4 shows the average delay of the RTS/CTS mechanism of variable number of nodes in the network. As the number of nodes increases, the average delay keeps in increasing, especially in the case of mobility network scenarios with multi-hopping routing scheme. Although the Gamma RWP model has a higher average speed compared to the uniform RWP model, the average delay induced by both RWP mobility patterns to the mobility network scenarios are similar. The maximum average delay is almost 17sec and 145sec at a network size of 50 nodes for the fixed and mobility network scenarios with multi-hopping routing scheme, respectively. While the maximum average delay of the mobility network scenarios with 1-hop route is about 5sec at a network size of 50 nodes. Mobile ad-hoc network scenario with multi-hopping routing scheme involves large delays, this is because of packet routing and buffers. Figure 5 shows the packet retransmission rate of the RTS/CTS mechanism of variable number of nodes in the network. As the number of nodes increases, the packet retransmission rate increases up to a maximum value of 48 and 139 at a network size of 50 nodes. This is for the fixed and the
of nodes per unit area increases. This is in contrast to the mobile ad-hoc networks scenario with 1-hop route. The result implies that the throughput performance improvement is obtained through the exploitation of nodes’ mobility and multihopping routing scheme. However, the disadvantage of the mobile ad-hoc networks scenario with multi-hopping routing scheme is that it involves large delays. This is in tradeoff with the end-to-end average throughput. The results show that the effects of both mobility patterns on the IEEE 802.11 RTS/CTS are similar. This is in terms of throughput, delay and packet retransmission rates. However, the Gamma RWP model has a higher steady state average speed compared to the uniform RWP model.
Average Packet Delay (ms)
150 1−hop, Stationary 1−hop, Γ ∼ [1−19] 1−hop, U ∼ [1−19] multi−hoping, Γ ∼ [1−19] multi−hoping, U ∼ [1−19] 100
50
0
R EFERENCES 0
5
Fig. 4.
10
15
20 25 30 Number of Nodes
35
40
45
50
Average System Delay of RTS/CTS mechanism.
mobile ad-hoc network scenarios with multi-hopping routing scheme, respectively. While the packet retransmission rate for the mobile ad-hoc networks scenario with 1-hop route is too small. This is because the time fraction of two nodes to be in range is too small, thus most of nodes are out of range most of the time and cannot access the medium. Also, it should be noted that the failure of packet delivery and/or packet collision, in turn increases the packet retransmission rate.
Packet Retransmission Rate
150 1−hop, Stationary 1−hop, Γ ∼ [1−19] 1−hop, U ∼ [1−19] multi−hoping, Γ ∼ [1−19] multi−hoping, U ∼ [1−19] 100
50
0
0
Fig. 5.
5
10
15
20 25 30 Number of Nodes
35
40
45
50
Packet Retransmission Rate of RTS/CTS mechanism.
5. C ONCLUSION Mobility and multi-hopping routing scheme provide performance improvement in terms of the average throughput. Additionally, the average throughput of the IEEE 802.11 RTS/CTS can be kept constant under the mobility constraint provided that direct or indirect transmission link exists between nodes. The average throughput remains constant, even as the number
[1] G. Bianchi, “Performance analysis of the IEEE 802.11 distributed coordination function”, IEEE Journal on Selected Areas in Communications, vol. 18, no. 3, pp. 535-547, Mar. 2000. [2] K. Medepalli and F. A. Tobagi, “Throughput analysis of IEEE 802.11 wireless LANs using an average cycle time approach”, IEEE Global Communications Conference (GLOBECOM), Saint Louis, USA, Nov. 2005. [3] K. Xu, M. Gerla and S. Bae, “How effective is the IEEE 802.11 RTS/CTS handshake in ad-hoc networks”, IEEE Global Communications Conference (GLOBECOM), Taipei, China, Nov. 2002. [4] D. Valerio, L. De Cicco, S. Mascolo, F. Vacirca and T. Ziegler, “Optimization of IEEE 802.11 parameters for wide area coverage”, IEEE Global Communications Conference (GLOBECOM), Taipei, China, Nov. 2002. [5] P. Gupta and P. R. Kumar, “The capacity of wireless networks”, IEEE Trans. Inform. Theory, vol. 46, pp. 388-404, Mar. 2000. [6] M. Grossglauser and D. N. C. Tse, “Mobility increases the capacity of ad-hoc wireless networks”, IEEE International Conference on Computer Communications (INFOCOM), pp. 1360-1369, Alaska, USA, Mar. 2001. [7] IEEE computer society LAN MAN standards committee, “Wireless LAN medium access protocol (MAC) and physical layer specification. IEEE std 802.11 1997”, Institute of Electrical and Electronic Engineers, NY, pp. 784-792, Mar. 1999. [8] J. Weinmiller, M. Schlager, A. Festag and A. Wolisz, “Performance study of access control in wireless LANs IEEE 802.11 DFWMAC and ETSI RES 10 HIPERLAN”, Mobile Networks and Applications, vol. 2, pp. 55-67, 1997. [9] P. Karn, “A new channel access method for packet radio”, ARRL/CRRL Amateur Radio 9th Computer Networking Conference, Sep. 1990. [10] V. Bharghavan, A. Demers, S. Shenker, L. Zhang, “MACAW: A media access protocol for wireless LAN’s”, in Proceedings of SIGCOM, Nov. 1994. [11] N. Sabah and A. Hocanin, “The use of negative acknowledgement control packets (NACKs) to improve throughput and delay in IEEE 802.11 networks”, International Conference on Computer Technology and Development (ICCTD), Cairo, Egypt, Nov. 2010. [12] F. Dai and J. Wu, “Distributed dominant pruning in ad hoc wireless netwoks”, in Proceedings of the IEEE International Conference on Communications (ICC), pp. 353, Alaska, USA, May 2003. [13] Charles E. Perkins, Elizabeth M. Royer and Samir R. Das, “IP flooding in ad hoc networks”, Internet draft (draft-ietf-manet-bcast-00.txt), 2001. [14] M. Zonoozi, P. Dassanayake, “User mobility modeling and characterization of mobility patterns”, IEEE Journal on Selected Areas in Communications, 15(7):1239-1252. Sep. 1997. [15] D. Perkins, H. Hughes, C. Owen, “Factors affecting the performance of ad hoc networks”, in Proceedings of the IEEE International Conference on Communications (ICC), New York, USA, April 2002. [16] J. Yoon, M. Liu, B. Noble, “Random waypoint considered harmful”, IEEE International Conference on Computer Communications (INFOCOM), Vol. 2, pp. 1312-1321, San Francisco, USA, April 2003. [17] N. Sabah and A. Hocanin, “An improved random waypoint mobility model for wireless ad-hoc networks”, International Conference on Information and Multimedia Technology (ICIMT), Hong Kong, accepted.
