JOURNAL OF COMPUTER SCIENCE AND ENGINEERING, VOLUME 6, ISSUE 2, APRIL 2011 7

SQUARING BACKOFF BASED MEDIA ACCESS CONTROL FOR MOBILE ADHOC NETWORKS P. I. Basarkod, S. S. Manvi and P. N. Narendra Reddy Abstract— A good backoff technique can reduce a large number of collisions in the MAC layer. This will reduce the collision probability and hence increases the utilization of network resources. A uniform random distribution has been employed to choose the backoff value in the Binary Exponential Backoff (BEB) technique used in the IEEE 802.11 MAC protocol. This random choosing leads to unnecessary idle times and reduced throughput. This paper proposes a new backoff technique called “Squaring Backoff (SB)” in which the differences between the consecutive contention window sizes are reduced to a negligible value. Here, the value of the backoff timer is based on the size of the contention window. The size of the contention window is varied in accordance with the result of the previous transmission. A successful transmission reduces the size of the contention window whereas a failure leads to the increase in the contention window size. Simulation results indicate that the proposed technique provides better throughputs and less idle times than the logarithmic and fibonacci based techniques when used in a mobile ad-hoc environment. Index Terms—Contention window, IEEE 802.11, Medium access control, Mobile adhoc networks.

——————————
——————————

1 INTRODUCTION

W

ireless technology first came into existence in 1901, when G Marconi successfully transmitted radio signals across the Atlantic Ocean [1]. Since their emergence, wireless networks have become increasingly popular in the computer research enabling the enhancements in mobility, portability and affordability for most of the systems seen in use today. A Mobile Adhoc Network (MANET) is a collection of mobile nodes dynamically forming a temporary network without the use of any existing network infrastructure [2]. MANETs utilize multi-hop radio relaying to find the connectivity between the different mobile nodes. MANETs have received significant attention in recent years due to their easiness to set up and usage in many domains. Such networks can be very useful in situations where there is limited time and resources. MANETs find their usage in military applications where setting up a fixed infrastructure for communication among a group of soldiers in enemy territories may not be possible. The shared media used by wireless networks helps a node to transmit a packet. Access to this media is controlled by the Media Access Control (MAC) protocol. MAC protocol of IEEE 802.11 uses Distributed Coordination Function (DCF) for contention based distributed access to the channel [3].

The Backoff mechanism is a basic part of a MAC protocol. Since only one transmitting node uses the channel at any given time, the MAC protocol must suspend other nodes while the media is busy. In order to decide the length of node suspension, a backoff timer is installed in the MAC protocol. The choice of backoff mechanism should consider generating backoff timers which allow adequate time for current transmissions to finish and avoid unintended idle time that leads to redundant delay in the network. IEEE 802.11 is the most popular WLAN standard that defines the specification for the physical and MAC layers. In the IEEE 802.11 standard MAC protocol, the truncated binary exponential backoff (BEB) technique is used [4]. Here, the inital contention window (CW) size is set to a random value between (0, CWmin). Each time a collision occurs, the CW doubles its size up to a maximum of CWmax. Thus at high load, the CW size is high and therefore the resolution power of the system is high. At low loads, small CW ensures low access delay. The term ‘truncated’ has been specified since after a certain number of increases, the exponentiation stops off suddenly at that level. BEB has the disadvantage of the channel capture effect and the instability. An efficient backoff algorithm should meet at least three requirements. A backoff algorithm should maximize the total throughput of the network, minimize the delay ———————————————— of transmission, and finally, maintain a fair usage of the • P.I. Basarkod is with the Department of Electronics and Communication, network among the transmitting nodes. The Backoff algoReva Institute of Technology and Management, Bangalore, Karnataka, Inrithms have been classified into two main categories; statdia-560064. • S.S. Manvi is with the Department of Electronics and Communication, ic and dynamic backoff algorithms. Reva Institute of Technology and Management, Bangalore, Karnataka, InIn static backoff algorithms, the backoff period will be dia-560064. of the form: Backoff Timer = I, an integer. The value of I • P.N. Narendra Reddy is with the Department of Electronics and Communication, Reva Institute of Technology and Management, Bangalore, Kar- can be carefully chosen depending on many factors; such nataka, India-560064. as the number of nodes in the network, having a fixed value can work under a certain scenario for a specific © 2011 JCSE http://sites.google.com/site/jcseuk/

JOURNAL OF COMPUTER SCIENCE AND ENGINEERING, VOLUME 6, ISSUE 2, APRIL 2011 8

network topology. In the case of MANETs, the major challenges would be mobility and dynamic topology, i.e. positions of nodes within the network area. In dynamic backoff algorithms, backoff periods are changed depending on many factors. The most common factor used is the result of last attempt of transmission by the node requesting channel access. In general, dynamic backoff algorithms deploy a customized version of the general formula. The input of the formula is the current size of CW and the result of this formula is the new size of CW (CWnew). CWnew is limited between a maximum value and a minimum value referred to as CWmax and Cwmin, respectively. CWnew is used to randomly choose the value of backoff timer according to formula: CWnew =

Min (f (CW), CWmax), after successful transmission Max (g (CW), CWmin), after a collision Backoff Timer = b, where b is a random integer and CWmin
Two important aspects of the backoff algorithm that need to be taken care are as follows. Firstly, the increment behavior needs to be examined. The method used by the backoff mechanism to increase CW size directly affects the balance between reducing the number of attempts to access the channel and reducing channel idle time. Successful collision avoidance will only be possible if adequate time is allowed between any two consecutive attempts to access the channel. On the other hand, a backoff algorithm should avoid unnecessarily long backoff periods. Imposing a long backoff period on a node is directly related to network idle time since the traffic flowing over the network is often unpredictable. Secondly, the decrement behavior after successful transmissions is also a major factor that needs to be explored. The backoff algorithm has to decide the reaction of a successful transmission since this decision affects the chances of nodes winning the next contention over the network. Balance should be maintained between extremely long and extremely short new values of CW. Moreover, resetting the counters to an initial value after a successful transmission has been proved undesirable; a node that has successfully transmitted a message has a small window size afterwards. Therefore, this node generates smaller backoff values leading to a higher possibility of winning the next contention over the channel The standard BEB technique implemented in IEEE 802.11 network protocol fails to achieve the best network throughput and has caused long delays over the network [6]. Fibonacci Increment Backoff (FIB) and Logarithmic approaches achieves higher throughput than BEB but suffer from unintended idle times [7], [8]. In this paper, we present a new backoff technique, referred to as the SB technique that can overcome the limitations of the existing backoff techniques. The SB technique provides better throughput and less idle times as per the simulation results. The rest of the paper is organized as follows. Section2 presents the new SB algorithm. Section 3 presents the simulation model. Section 4 then analyses the performance results. Lastly, section 5 presents the conclusion.

2 THE PROPOSED TECHNIQUE Backoff algorithms have been used for the collision avoidance and to increase the utilization of network resources [9]. Changing the size of the CW has a greater impact on the performance of the algorithm employed. During a failure to transmit, increasing the CW rapidly increases large CW to even further larger sizes. Reaching such large window sizes decreases the expected wait time for a given node to access to the shared medium. Moreover, a large window size tends to contribute to increasing channel idle times, leading to a major waste in the shared channel bandwidth. Therefore, we have proposed a new technique to avoid the above mentioned limitations. In our proposed algorithm, we have used Square series as the new size of CW, leading to reducing the increment factor when more transmission failures take place and hence introducing smaller increment on large window sizes. The squaring can be done as follows: SQ(n) = n x n; n= 2,3,4,…….................... (2) The squaring series can be obtained by calculating the ratio between every two successive terms in the above formula as follows: SQ(n) = SQ(n)/SQ(n-1); n=2,3,4,………………. (3) The incremental behavior of the proposed approach with the other two approaches has been discussed here. The three different increment formulas are Squaring, logarithmic and Fibonacci. Fig. 1 shows the behavior of the three increment formulas. The size of CW, measured in time slots, is plotted against number of iterations. The iteration number is the number of consecutive transmission failures. As the number of iterations (Contention failures) increase, the squaring increment is the largest of all the other increments. The contention window is increased by larger sizes leading to higher throughput in squaring backoff technique.

Fig. 1. Contention Window factor Vs Iterations

Squaring backoff algorithm is as follows. Algorithm: Squaring backoff

JOURNAL OF COMPUTER SCIENCE AND ENGINEERING, VOLUME 6, ISSUE 2, APRIL 2011 9

Nomenclature: {BO = backoff time, CW = contention window, DIFS = DCF interframe space, DCF = distributed coordination function} Step 0: Set BO to initial value Step 1: While Bo≠0 do For each time slot If (channel is idle) then BO=BO-1; If (channel is idle for more than DIFS) then Send; If (Send failure) then CW = Next Square series number; BO (i + 1) = square (i); (1 < i < CW – 1) Else CW = Initial Value; BO = 0; GO to step 1; Step 2: Stop. Fig. 2 illustrates the properties of the newly proposed squaring series and the existing fibonacci series. After a certain number of terms, the fibonacci series ratio converges to a limit of 1.618 whereas the square series is formed by the ratio which almost reaches a value of approximately one which leads to reduced idle times. Contention window in case of first failure can reach up to 4 for the square series leading to increased throughput whereas it is only 2 for the fibonacci series that is only half of the value of that of square series. In square series, after a certain number of terms the ratio tends to decrease to approximately 1 whereas in fibonacci series, the ratio tends to converge into a limit of (1+ √5)/2 ≈1.6. This leads to unnecessary idle times in fibonacci backoff technique that has been avoided in the squaring backoff technique.

Fig. 2. Ratio of successive terms Vs Iterations

3 SIMULATION The proposed algorithm has been evaluated using the C++ language with the help of a practical traffic model. The standard MAC protocol has been changed to implement the modified backoff algorithm. The routing protocol used is Adhoc On-Demand Distance Vector (AODV) protocol by taking different values of probability of broadcasting the messages. Every node in the network has the same probability of broadcasting the messages. By making the variations of the parameters in the network such as the total number of nodes and the speed, results have been verified thoroughly. Simulations have been carried out for the networks having a total number of nodes varying between 20 and 100 mobile nodes and the speed values ranging from 2 to 18 m/s. Other simulation parameters include the area of 1000m*1000m, nodes transmission range of 200m, simulation time of 800seconds and the traffic generated is the constant bit rate (CBR) traffic. All the simulation parameters are given in table 1. TABLE 1 SIMULATION PARAMETERS Parameter Nodes Area Speed Packet size Simulation time Transmission range Pause time Probability of broadcast

Value 20, 40, ....., 100 1000m* 1000m 2, 4, ....., 20 m/sec 512 bytes 800sec 200m 0 sec 0.50

4 RESULTS As shown in the figures 3 to 5, the newly proposed squaring algorithm has improved the total throughput of the network. Also the channel idle times have been reduced to make the best use of the bandwidth provided. When the number of nodes is increased, the contention is higher to gain access to the channel. Due to the reduced increment in the window size, a larger size of data was successfully received by the nodes. Both the square series and the fibonacci series improve the throughput with the increase in speed, whereas the logarithmic series has reduced throughput (see fig. 3). As the speed is increased, even though there are some problems occurring, the data packets reach the destination at a faster rate allowing the next transmission to occur. As the number of transmissions is increased, the number of collisions occuring is also more. Once the collision occurs, the contention window size increases frequently. Thus, the square series has better throughput than the fibonacci series.

JOURNAL OF COMPUTER SCIENCE AND ENGINEERING, VOLUME 6, ISSUE 2, APRIL 2011 10

5 CONCLUSION The Binary Exponential Backoff (BEB) algorithm used by the IEEE 802.11 MAC protocol uses uniform random distribution to choose the backoff value. In this paper, we have proposed a new backoff approach called “Squaring Backoff” algorithm to reduce the increment factor for large contention window sizes. Results from simulations have demonstrated that the proposed algorithm has increased the total throughput and reduced the average idle times, thereby making the best use of the bandwidth and other network resources.

ACKNOWLEDGMENT Fig. 3. Throughput Vs Speed (for 20 nodes)

The square series has the least delay when compared to the other approaches. When the number of nodes is increasing, the average time taken for the data packets to reach from the source to the destination keeps on increasing (see fig. 4). The square series consumes the least time of all approaches. According to the fig. 5, the number of data packets lost is comparably less in the squaring series and thus the number of packets received is more as the speed increases.

Fig. 4. Delay Vs Number of nodes (Speed of 10m/s)

We would like to express our sincere thanks to Professor R C Biradar and D S Albur for their continued support and guidance. Their continuous feedback has always been the strongest motivation behind this work.

REFERENCES S.Gobriel, R.Melhem and D.Mosse, “BLAM: An energy aware MAC layer enhancement for wireless adhoc networks,” in Proc. WCNC04, Atlanta, USA, 2004. [2] Y. Ko and NH. Vaidya, “Location-aided routing (LAR) in mobile ad hoc networks,” Wirel. Netw. 6, 4 (Jul. 2000), 307-321, July, 2000. [3] J.Deng, PK.Varshney, and ZJ.Haas, “A new backoff algorithm for the IEEE 802.11 distributed coordination function,” in Proc. CNDS 04, San Diego, CA, USA, Jan.18-21,2004. [4] IEEE Computer Society LAN MAN Standards Committee, ”Wireless LAN medium access control (MAC) and physical layer (PHY) Specifications,” ANSI/IEEE Std. 802.11, 1999 Edition, The Institute of Electrical and Electronic Engineers, New York, 1999. [5] L.Bononi, M.Conti, L.Donatiello, “A distributed mechanism for power saving in IEEE 802.11 wireless LANs,” (ACM MONET) Journal, Volume 6 (2001), N.3, pp. 211-222, 2001. [6] S. Manaseer and M. Masadeh,”Pessimistic backoff for mobile ad hoc networks,” ICIT2009. [7] S.Manaseer, M. Ould-Khaoua and LM. Mackenzie, “Fibonacci Backoff Algorithm for mobile ad hoc Network”, PGNET 2006. [8] J.Hu, CD.Raymond, “A statistics based design of MAC protocols with distributed collision resolution for ad hoc networks,” in Proc. MobiWac05, Hawaii, June, 2005. [9] M. Bani Yassein, M. Ould-Khaoua, S. Papanastasiou, ” On the performance of probabilistic flooding in mobile ad hoc networks,” Proceedings of International Workshop on Performance Modelling in Wired, Wireless, Mobile Networking and Computing in conjunction with 11th International Conference on Parallel and Distributed Systems (ICPADS2005), IEEE Computer Society Press, Fukuoka, Japan, 20 - 22 July, 2005. [10] Bhavyesh Divecha, Ajith Abraham, Crina Grosan and Sugata Sanyal, “Impact of node mobility on MANET routing protocols models,” Journal of Digital Information Management, 2007. [11] M.Taifour, F. Nait-Abdesselam, D.Simplot-Ryl, ”Neighbourhood backoff algorithm for optimizing bandwidth in single hop wireless ad-hoc networks,” in Proc. MobiWac05, Hawaii, June, 2005. [1]

[12] S. Romaszko and C Blondia, “Enhancements of the IEEE 802.11,a MAC protocol for ad hoc network with history of power adjustment,” in Proc. MobiWac05, June, Hawaii, 2005. [13] I.Aad, Q.Ni, C.Barakat,T.Turletti, ”Enhancing IEEE 802.11 MAC in congested enviroments,” in Proc. IEEE ASWN, Boston - MA, USA, August, 2004. Fig. 5. Number of Data packets Vs Speed (for 100 nodes)

JOURNAL OF COMPUTER SCIENCE AND ENGINEERING, VOLUME 6, ISSUE 2, APRIL 2011 11

[14] G.Bianchi, L.Fatta, and M.Oliveri,”Performance evalutation and enhancement of the CSMA/CA MAC protocol for 802.11 wireless LANs,” in Proc. IEEE PIMRC, Taipei, Taiwan, pp.407-411, Oct.1996. P. I. Basarkod received B.E degree in Electronics and Communication from National Institute of Engineering, Mysore, M.E degree in Electronics from the B. M. S Engineering, Bangalore, M.Sc in Software Systems from Birla Institute of Technology and Science, Pillani India and currently doing research under the guidance of Dr. Sunilkumar S Manvi in Kuvempu University, Shankaragatta, Shimoga. He is currently working as a Professor in Electronics and Communication Department of Reva Institute of Tecnology and Management, Bangalore. He is having teaching experience of twenty four years and his areas of interest are Wireless Communication and Computer Networking. He is a member of ISTE (MISTE, India), member of IEEE (MIEEE, USA). S. S. Manvi recieved M.E degree in Electronics from the University of Visweshwariah College of Engineering, Bangalore and Ph.D degree in Electrical Communication Engineering, Indian Institute of Science, Bangalore, India. He is currently working as a professor and Head of Department of Electronics and Communication Engineering, Reva Institute of Technology and Management, Bangalore, India. He is involved in research of Agent based applications in Multimedia Communications, Grid Computing, Vehicular Adhoc Networks, E-Commerce and Mobile Computing. He has published around 150 papers in national and international conferences and journals. He has published three books. He is a fellow IETE (FIETE, India), Fellow IE (FIE, India) and member ISTE (MISTE, India), Member IEEE (MIEEE, USA). P. N. Narendra Reddy received B.E. degree in Electronics and Communication Engineering from Reva Institute of Technology and Management, Bangalore and persueing M.Tech degree from the same institution, in the field of VLSI Design and Embedded Systems. His research interests are VLSI based Agent applications, Wireless Communication and Computer Networking. He has published two papers at national conferences. He is a member IEEE (MIEEE, India).