Fibonacci Backoff Algorithm for Mobile Ad Hoc Networks - CiteSeerX

Fibonacci Backoff Algorithm for Mobile Ad Hoc Networks

Saher S Manaseer

Mohamed Ould-Khaoua

Lewis M Mackenzie

Department of Computing Science University of Glasgow, Glasgow G12 8RZ, UK {saher, mohamed, lewis}@dcs.gla.ac.uk Abstract- The collision probability in the MAC layer could become very high when a poor backoff algorithm is used, especially in dense networks. The Binary Exponential Backoff (BEB) used in the IEEE 802.11 MAC protocol uses a uniform random distribution to choose the backoff value. This often leads to reducing the effect of window size increment. This paper proposes a new backoff algorithm called “Fibonacci Increment Backoff (FIB)” in which the differences between consecutive contention window sizes are reduced. Results from simulation experiments reveal that the proposed algorithm achieves higher throughput than the BEB when used in a mobile ad hoc environment.

Keywords IEEE 802.11, Ad hoc networks, Medium access control, Backoff algorithm, Throughput analysis.

I.

INTRODUCTION

Since their emergence, wireless networks have become increasingly popular in the computing industry. This is particularly true within the past decade, which has seen wireless networks being widely adopted to enable mobility. Over the development process of computer networks, two main variations of mobile wireless networks have been introduced, infrastructure and ad hoc wireless networks [10]. Recently, a significant number of researchers have moved towards studying Mobile Ad Hoc Networks (MANETs). Interest in MANETs is due to many new characteristics provided only by this type of networks. Firstly, MANETs are easily deployed allowing a plug-and-communicate method of networking. Secondly, MANETs need no infrastructure [7], eliminating the need for an infrastructure reduces the cost of establishing the network. Moreover, such networks can be useful in disaster recovery where there is not enough time or resources to install and configure an infrastructure. Thirdly, MANETs also do not need central management. Hence, they are used in military operations where units are moving around the battlefield and a central unit cannot be used for synchronization [7]. Nodes forming an ad hoc network are required to have the ability to double up as a client, a server, and a router simultaneously [7]. Moreover, these nodes should also have the ability to connect to and automatically configure to start transmitting data over the network. As a result of having the characteristics mentioned so far,, protocols used for ad hoc networks generally function in a distributed manner [12]. The distributed Coordination Function (DCF) is used for synchronous, contention-based, distributed access to the channel [3]. MANETs use a shared medium to transfer data between its nodes. It is impractical to expect a MANET to be fully connected, where a node can directly communicate with every other node in the network. Typically, nodes are obliged to use a multihop path for transmission, and a packet may pass through multiple nodes before being delivered to its intended destination. The wireless medium used by MANETs has a number of problems. Bandwidth sharing, signal fading, noise, interference, etc…. with such a public medium, a well-organized and effective Medium Access Control (MAC) is indispensable to organize sharing the scarce bandwidth resource [4] [7]. Based on the features mentioned, the design of the medium access control (MAC) protocol is a significant factor affecting the performance of MANETs. Many researchers have proposed the mechanism of channel sensing, or packet sensing to avoid collision. The sensing mechanisms typically rely on the transmitter and receiver performing a handshake prior to the transmission of the data packet [2]. More specifically, The Medium Access Collision Avoidance (MACA) method proposed by Karn [11] implements the handshake via a pair of Request-To-Send (RTS) and Clear-To-Send (CTS) messages. When a node has to send data to another, it first sends a short RTS to the destination. The receiver responds with a CTS packet [2]. On receipt of the CTS, the sender

1

sends its queued data packet(s). All other nodes overhearing the CTS message will defer from sending out any packet until the predicted transmission period indicated in the CTS packet, is passed. Any node that overhears the RTS signal but not CTS is allowed to send out packets in a certain time period as either the RTS/CTS handshake is not completed or it is out of range of the receiver. As a part of an efficient MAC protocol, a backoff algorithm is used to avoid collisions when many nodes try to access the medium [12]. Only one of the nodes is granted access to the channel, while other contending nodes are suspended into a backoff state for some period (BO) [8]. Many backoff algorithms have been developed in the literature [13, 8]. One example is the Multiplicative Increase Linear Decrease (MILD) algorithm [13]. This algorithm improves the total throughput of the network, but the cost of this improvement is the need of a perfect knowledge regarding collisions happening over the network, which is high cost and hardto-acquire knowledge [7]. In a wired LAN the total number of nodes of the network is easily obtained. However, as nodes in MANETs are mobile, knowing the number of nodes may incur a high cost, since this knowledge needs to be updated due to the frequent change in the MANET topology. One approach to update and keep the knowledge consistent is by exchanging packets between neighboring nodes [12].these packets form extra load, consume a part of the network resources, and cause a longer delay. Moreover, neighbor discovery requires more control processing, and even incurs more work to the backoff algorithm itself. Other backoff algorithms have tried to find a fixed optimum backoff value to use. Even though, the distributed functionality was incomplete [1]. In the IEEE 802.11 standard MAC protocol, the truncated Binary Exponential Backoff (BEB) is used [19]. In this algorithm, when a node over the network has a packet to send, it first senses the channel using a carrier sensing technique. If the channel is found to be idle and not being used by any other node, the node is granted access to start transmitting. Otherwise, the node waits for an inter-frame space and the backoff mechanism is invoked. A random backoff time is chosen in the range [0, CW-1]. A uniform random distribution is used here, where CW is the current contention window size. The following equation is used to calculate the backoff time (BO):

BO = (Rand () MOD CW) * aSlotTime

(1)

The backoff procedure is preformed then, by putting the node on a waiting period of length BO. Using carrier sense mechanism, the activity of the medium is sensed at every time slot. If the medium is found to be idle then the backoff period is decremented by one time slot.

Backoff time (BO) new = (BO) old - aSlotTime (2)

If the medium is busy during backoff, then the backoff timer is suspended. That is, the backoff period is counted in term of idle time slots. Whenever the medium is found idle for longer than an inter-frame space, backoff is resumed. When the backoff period is finished with a BO value of zero, a transmission should take place. If the node succeeds to send a packet and receive an acknowledgment for that packet, then the CW for this node is reset to the minimum, which is equal to 31 in the case of BEB. The 'truncated' simply means that after a certain number of increases, the exponentiation stops; i.e. the retransmission timeout increases to a ceiling value, and thereafter does not increase any more. The ceiling is set at i =10 transmission failures, assuring that the maximum delay is 1023 slot times. Since these delays cause other nodes to collide as well, there is a possibility that on a busy network, many nodes may be caught in a single collision set. To overcome the possibility of starvation, the transmission process is aborted after 16 failing attempts of transmission [10] BEB has a number of disadvantages [12]. One major disadvantage is the problem of fairness. BEB tends to have a preference for most recent contention winner and new contending nodes over other nodes when allocating channel access. Determining backoff period is accomplished by choosing a random backoff value from a contention window (CW) which has smaller size for new contending nodes and contention winners. This behavior causes what is known as “Channel capture effect” in the network. Another problem of BEB is stability. BEB has been designed to be stable for large number of nodes [7]. However, a number of studies have shown that BEB could suffer from instability [7]. In this paper, we present a new backoff algorithm, referred to as Fibonacci Increment Backoff (FIB)” that can overcome the

2

limitation of the existing MAC solutions in. In the FIB algorithm the difference between consecutive contention window sizes are reduced according to a Fibonacci sequence. Results from simulation experiments reveal that the proposed algorithm achieves higher throughput than the BEB when used in a mobile ad hoc environment.

The rest of the paper is organized as follows. Section 2 presents the new FIB algorithm. Section 3 presents the simulation model. Section 4 then analyses the performance results. Finally, Section 5 concludes this study. II.

THE PROPOSED ALGORITHM

Most backoff algorithms suffer from the following shortcoming due to their inherent operations. Increasing the contention window in case of failure to transmit tends to rapidly increase large contention windows to even larger sizes. Reaching such large window sizes dangerously decrease the possibility of gaining access to the channel. Moreover, a large window size tends to contribute to increasing channel idle times, leading to a major waste the shared limited communication channel. Motivated by this above observation, we propose a new backoff algorithm to improve the performance of the backoff algorithm. One of the most famous series in math is the Fibonacci series defined by the following rule [21]:

F (n) = F (n - 1) + F (n - 2). F (0) = 0, F (1) = 1, n >= 0.

The Fibonacci series has a number of useful characteristics. One specific property is a special value, closely related to the Fibonacci series, is called the golden section. This value is obtained by taking the ratio of successive terms in the Fibonacci series. Figure 1 illustrates this property. After a certain number of terms, the ratio tends to converge into a limit of 1.6 [21]. In our proposed algorithm, we used the F (n) as the new contention window size, leading to a smaller increment on large window sizes as shown in figure 2.

Figure 1 Ratio of successive Fibonacci terms.

Figure 2 Pseudo code of FIB Algorithm

3

III.

SIMULATION

The proposed algorithm has been evaluated using the Network Simulator NS 2.29 [6]. The original standard MAC protocol has been modified to implement our proposed backoff algorithm. Modifications have mainly targeted the mac802_11.h and mac802_11.cc files. Several topologies and mobility scenarios have been created to test the algorithm as intensively as possible. Firstly, we have varied the total number of nodes in the network. Changing the number of nodes is our method of predicting the performance of our algorithm for all sizes of networks. Simulations have been carried out for networks having total number of node varying between 20 and 100 mobile nodes. Secondly, in order to address the main drawback of MANETs, we have used different values for mobility speed. Testing for speed values, ranging from 2 ms-1 to 20 ms-1 has given us useful information concerning the efficiency of our algorithm for static and highly mobile MANETs as well. Other simulation parameters are the area of 1000m×1000m, simulation time is 900 seconds, nodes transmission range is 250 m and the traffic generated is CBR traffic. Table 1 summarizes simulation parameters. Table 1 Simulation Parameters Parameter

Value

Nodes

20,40,…,100 -1

Speeds (ms )

2,4,…,20

Area

1000m X 1000m

Simulation time

900 s

Transmission range

250 m

IV.RESULTS As shown in the following figures, the new FIB has improved the total throughput of the network simulated in our work. When the number of nodes is increased, the contention is higher to gain access to the channel. Because of the reduced amount of increment on the window size, a larger size of data was successfully received by nodes over the network. The same enhancement is noticed even while increasing mobility speed. One of the major obstacles in the way of developing a MAC protocol for MANETs is mobility. Having a long backoff value allows the node to move outside the transmission range before being allowed to retry accessing the channel. With FIB, the ceiling of backoff periods is controlled to prevent extremely long backoff periods. One more factor participating in increasing throughput is reducing idle times. With smooth increments of contention window size, idle time is reduced.

Total received data for speed 10m /s

7

Data receievd MB

6 5

FIB

4

LOG 3

BEB

2 1 0 20

40

60

80

100

Num ber of nodes

Figure 3 Received Data at speed 10m/s

4

Total data received at 20 m/s 9 8 7 6

FIB 5

LOG BEB

4 3 2 1 0 20

40

60

80

100

N umbe r of node s

Figure 4 Received Data at speed 20 m/s

Total Data Received for 50 Nodes 4.5 4 3.5 3 Data MB

FIB 2.5

LOG BEB

2 1.5 1 0.5 0 1

2

3

4

5

speed m /s

Figure 5 Received Data at 50 nodes

Total Data Received for 100 Nodes 8 7 6

Data (MB)

5

FIB

LOG

4

BEB 3 2 1 0 2

6

10

14

18

Speed m /s

Figure 6 Received Data at 100 nodes

V.

CONCLUSION

The Binary Exponential Backoff (BEB) is used by the IEEE 802.11 Medium Access Control (MAC) protocol. BEB uses uniform random distribution to choose the backoff value. In this paper, we have proposed a new backoff algorithm to reduce the increment factor for large contention window sizes. Results from simulations have demonstrated that the proposed algorithm increased the total throughput of mobile ad hoc networks especially when the system size is large. The total throughput has been increased for highly mobile networks.

5

VI. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

REFERENCES

L. Bao and J. J. Garcia-Luna-Aceces, “A New Approach to Channel Access Scheduling for Ad Hoc Networks,” in ACM MOBICOM, pp. 210–221, 2001. V. Bharghavan, et al., "MACAW: a media access protocol for wireless LAN's", Proceedings of the conference on Communications architectures, protocols and applications, 1994, pp. 210 – 225. L. Bononi, et al., "A differentiated distributed coordination function MAC protocol for cluster-based wireless ad hoc networks", Proceedings of the 1st ACM international workshop on performance evaluation of wireless ad hoc, sensor, and ubiquitous networks, pp. 77 - 86, 2004. F. Cali', et al.., “IEEE 802.11 Wireless LAN: Capacity Analysis and Protocol Enhancement”, Proc. INFOCOM'98, San Francisco, CA, March 29 April 2, 1998, pp. 142-149. B. Ensaou, et al.,” Fair Media Access in 802.11 based Wireless Ad-Hoc Networks.” In IEEE/ACM MobiHOC (Boston, MA., August 2000). K. Fall and K. Varadhaa. editors. “NS notes and Documentation.” The V l N l Project UC Berkeley. LBL. USC/ISI. and Xeros PARC. 2002. Z.Fang, et al., “Performance evaluation of a fair backoff algorithm for IEEE 802.11 DFWMAC.” International Symposium on Mobile Ad Hoc Networking & Computing J. Goodman et al., “Stability of Binary Exponential Backoff”, app. In the Proc. of the 17-th Annual ACM Symp. Theory of Comp., Providence, May 1985. J. Hastad , et al., “Analysis of Backoff Protocols for Multiple Access Channels”, Siam J. Computing vol. 25, No. 4, 8/1996, pp. 740. IEEE, ANSI/IEEE std 802.11, 1999 Edition (R2003), Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications. P. Karn, "MACA - A new channel access method for packet radio", ARRL/CRRL Amateur Radio 9th computer Networking Conference, 1990, pp. 134 - 140 S. Manaseer and M. Ould-kauoa, "A New Backoff Algorithm for MAC Protocol in MANETs," 21st Annual UK Performance Engineering Workshop, pp 159-164, 2005. K. Sakakibara, et al., "Backoff Algorithm with Release Stages for Slotted ALOHA Systems." ECTI Transactions On Electrical Eng., Electronics, And Communications vol.3, no.1 pp 59-70 ,2005. K. Sundaresan and R. Sivakumar, "A unified MAC layer framework for ad-hoc networks with smart antennas", Proceedings of the 5th ACM international symposium on Mobile ad hoc networking and computing, pp. 244 - 255, 2004. C. Sauer, E. MacNair, "Simulation of Computer Communication Systems", Prentice-Hall, INC., 1983 F. Tobagi and L. Kleinrock, "Packet Switching in Radio Channels: Part I--Carrier Sense Multiple-Access Modes and Their Throughput-Delay Characteristics", IEEE Transactions on Communications, 1975, Vol. 23, No. 12, pp. 1400 – 1416. F. Tobagi and L. Kleinrock, "Packet Switching in Radio Channels: Part II--The Hidden Terminal Problem in Carrier Sense Multiple-Access and the Busy-Tone Solution", IEEE Transactions on Communications, 1975, Vol. 23, No. 12, pp. 1417 – 1433. Cheng-shong Wu and Victor O.K. Li, "Receiver-initiated busy-tone multiple access in packet radio networks", Proceedings of the ACM workshop on Frontiers in computer communications technology, 1987, pp. 336 - 342. K. Xu, et al., "How effective is the IEEE 802.11 RTS/CTS handshake in ad hoc networks", IEEE Global Telecommunications Conference, 2002, Vol. 1, pp. 72 – 76. H. Zhai and Y. Fang,”Performance of Wireless LANs Based on IEEE 802.11 protocols.” 14th IEEE International Symposium on Personal, Indoor and Mobile Radio Communication Proceedings, pp 2586-2590, 2003. Finch, S. R. "The Golden Mean." Mathematical Constants. Cambridge University Press, pp. 5-12, 2003.

6