Network Performance Improvement Using Fuzzy ...

International Journal of Applied Engineering Research ISSN 0973-4562 Volume 9, Number 23 (2014) pp. 22373-22383 © Research India Publications http://www.ripublication.com

Network Performance Improvement Using Fuzzy Genetic Based Aodv 1

1

N.K. Kuppuchamy 2R.Manimegalai

VidyaaVikas College of Engineering and Technology, Tiruchengode, India. 2 Park College of Engineering and Technology, Coimbatore, India. Email: [email protected], [email protected]

Abstract An ad hoc network is an autonomous system of nodes moving in an arbitrary fashion forming a network without any centralized support. In a Mobile Ad hoc Network (MANET), communication between the nodes takes place by multihop links. As each node acts as a router, data packets are forwarded through the nodes in the network to the destination node. As part of this research work, Genetic Algorithm (GA) is used to improve the initial rule-set as well as the membership functions of the proposed fuzzy routing. Success depends on the ability of the selected route to keep up with the required QoS demand, at low cost. GA offer an optimization method in which the stochastic search algorithm is based on biological principles of evolution like selection, cross-over and mutation. GA finds the shortest path in less time and is quicker than other routing protocols. Keywords: Ad Algorthm(GA)

hoc

ondemand

Distance

Vector(AODV),

Genetic

Introduction Ad-Hoc On-Demand Distance Vector (AODV) protocol has been extensively studied in terms of the route recovery mechanism, security, anomaly detection with different mobility patterns and speeds. A common problem faced by nodes in ad hoc network located in the edge of the network are unreliable links, poor signal quality, higher latency and larger packet loss. Mobility of nodes affects the performance of routing protocols. For efficient communication in the network, dynamic routing protocols were developed. Mobility of the nodes, large number of nodes, and limited bandwidth are some of the challenges faced by the routing protocols [1]. Many numbers of routing protocols were developed for MANETS. Dynamic optimization in routing is used to find paths that satisfy some optimality criteria and some constraints. Usually,

22374

N.K. Kuppuchamy and R.Manimegalai

shortest distance, minimal bandwidth usage and minimum delay can be used as optimality criteria. Constraints may be limited power and limited capability of wireless links. Ad-Hoc On-Demand Distance Vector (AODV) A common problems faced by nodes in ad hoc network located in the edge of the network are unreliable links, poor signal quality, higher latency and larger packet losses. Mobility of the nodes affects the performance of the routing protocols. During high mobility, link failures between the nodes occur frequently [2]. AODV starts new route discoveries on link failures as alternate routes are not stored in the routing tables. Thus, during high mobility, with an increase in link failures, the route discovery also increases proportionally leading to increase in routing load. During low mobility, nodes tend to cluster, and it leads to congestion during high traffic. This leads to degradation of the network performance by increasing routing loads and decreasing throughput and packet delivery ratio. In this paper, a fuzzy genetic algorithm based AODV is proposed to avoid abnormal behavior of routing during high mobility. The proposed method improves performance by applying local congestion methods in AODV routing protocols. Fuzzy Based Routing in MANETs The strength of fuzzy controllers lies in the ability to model and to express decisions in a near-natural language, thus efficiently exploiting expert knowledge for building automated systems. Fuzzy control relies on fuzzy set theory which is a generalization of classic set theory in that set membership is not determined by a Boolean expression: it is allowed to be expressed in degrees of membership in the range [0,1], imitating the human, linguistic, non-precise approach to describing conditions (cold, warm etc) [3]. By using such an approach, a fuzzy controller can be expressed using rules of the form IF {condition} THEN {action}. The membership values control the degree to which each rule “fires”, illustrating the interdependent relationship between the rule set and the membership functions. Evolutionary Algorithm based Routing in MANETs The routing function is a basic control function in wireless networks. In datagram networks, such as the Internet, the routing function decides upon the next-hop interface, when a packet is received on one of the node’s interfaces. Furthermore, information about network conditions may be incomplete or out dated. The routing function must find a path that can satisfy the users’ requirements. In addition to QoS parameters, in a commercial multi-provider environment, policy constraints must also be seriously considered. Therefore, policy parameters must also be included in route computation which further complicates the problem making the problem of routing given an extended parameter NP-complete. Therefore, heuristics such as intelligent techniques should be employed to find a near-optimal solution. Genetic Algorithm (GA) is a computing search technique to find true or approximate solutions to optimization and search problems. GA is a programming technique that mimics the biological evolution as a problem-solving strategy [4].

Network Performance Improvement Using Fuzzy Genetic Based Aodv

22375

When solving a specific problem, the input to GA is a potential solutions domain, encoded with a metric titled fitness function allowing quantitatively evaluation of each candidate solution. Each candidate’s evaluation is based on the fitness function. Fitness function evaluates a set of chromosomes from a population. To start with, a population of individuals, P is created and new population P’. The algorithm routinely chooses individuals for the creation of new ones. The parents produce two offspring through the use of a crossover technique. Generation of mutants is also possible. Crossover technique includes random exchange of bits between an intermediate population’s strings. Mutation operators are new string bits. This algorithm provides an optimal acceptable solution. Genetic algorithms work for the topological design of networks. A chromosome is chosen when it is contained within network parameters. Each chromosome’s fitness function is chosen based on the design problem objective which is to reduce the source-destination route.

Related Works This section discuses some of the work available in the Related works in the following areas, namely, enhancing routing using fuzzy logic and optimization methods such as evolutionary algorithm, swarm intelligence. Fuzzy based routing Hiremath and Joshi presented energy efficient routing protocol with adaptive fuzzy threshold energy for MANET [5]. This approach used adaptive threshold energy policy using fuzzy logic approach. Remaining residual energy at each node and energy level of neighborhood nodes were used for selection of routes. Jing Zuo et al presented Fuzzy logic DSR for ad hoc networks [6]. Mobile nodes created inaccurate information for routing. Proposed method used fuzzy logic that took expected life time and number of hops as input parameters and selected highest stability route for transmission from a source node to a destination node. Fuzzy logic DSR was compared with existing DSR for ad hoc networks. Results showed that fuzzy logic approach performed better than DSR approach. Vamsee Mohan et al presented A Reliable Routing Algorithm in MANET Using Fuzzy [7]. Network topology was represented using peri nets, and fuzzy reasoning was used find a routing tree that was useful to select a reliable route from the source node to the destination node. Degree of reliability was compared with existing routing methods. Result showed that more than 80% reliability was achieved than AODV protocol. An adaptive routing Fuzzy based Balanced Power Aware Routing Algorithm (FBPRA) incorporating path maintenance mechanism and a stable route using fuzzy logic was proposed by Abirami, et al., [8]. Path maintenance mechanism reduced path breakages by establishing new path through neighboring nodes, before packet transmitted path broke due to node mobility. Performance evaluation was through simulation to compare the proposed fuzzy logic approach and classical methods. The results revealed that the algorithm improved network performance effectively.

22376

N.K. Kuppuchamy and R.Manimegalai

Genetic Algorithm based routing Anjum et al presented optimal routing in ad-hoc network using genetic algorithm [9]. Since the nodes of MANET are mobile finding shortest path between a source and a destination node that satisfies QoS was a challenging task. This GA algorithm used varied length of chromosomes and cross over partial chromosomes were used to find partial shortest routes. GA approach was compared with DSR routing. End to end delay, packet loss ratio, time to find route failure were less in proposed approach. VikasSiwach et al presented an approach to optimize QoS routing protocol using genetic algorithm in MANET [10]. To implement QoS routing, optimization was the best approach. In this approach, all the possible paths from the source node to the destination node were found and congestion free efficient routing was achieved by using genetic intelligent approach. For the selected path ACK was sent in the reverse order of path selection. A reliable fuzzy-logic based routing algorithm to find a reliable reactive protocol in MANETs was proposed by Ghalavand, et al., [11]. Nodes with more trust value and maximum energy are selected as routers based on a parameter called “Reliability Value” during route discovery. This formed a reliable source to destination route, increased network life and decreased packet loss during transmission. A technique to analyze exposure to attacks in AODV, specifically Black Hole attack and to develop specification based Intrusion Detection System (IDS) using GA was proposed by Sujatha, et al., [12]. The new process was based on GA which analyzed every node’s behaviors and provided details of the attack. Genetic Algorithm Control (GAC) was sets of rules based on AODV’s vital features like Request Forwarding Rate, Reply Receive Rate and others. MANET performance analysis was GAC based.

Methodology This section highlights the essential details of the work done in the following areas: i) Fuzzy logic routing, ii) Fuzzy logic and GA approach based routing. Fuzzy Logic Approach for MANET Routing This work proposes fuzzy logic and genetic approach to select optimal routes to satisfy QoS required by applications using MANETs. Fuzzy rule based system is formed by the linguistic variables actual end to end delay (Ae), number of times the node leaves the network (Nl), number of packets dropped (Nd) and number of Rerr generated (Nr).Fuzzy logic is a classical logic set with degree of membership. Fuzzy Logic was designed to represent fuzziness and vagueness mathematically and provide fundamental concept to handle imprecision intrinsic to issues of subjective evaluation and measurement[13]. Fuzzified input data trigger one/many rules in fuzzy model to calculate results. IF - THEN rules map input values to output space regarding implication relation between fuzzy sets in “IF” and “THEN” parts. A key Fuzzy logic feature is handling uncertainties and non-linearity in physical systems, similar to reasoning by human beings, making it attractive for decision making systems [14]. A fuzzy logic system has 3 elements: (i) Fuzzification, (ii) Knowledge base (rule and function), and (iii) Defuzzification. Fuzzification determines the membership degree to a crisp input in fuzzy sets. Fuzzy rule base

Network Performance Improvement Using Fuzzy Genetic Based Aodv

22377

presents input-output fuzzy variables fuzzy relationship. The fuzzy rule base output is based on membership degree specified by fuzzifier. Defuzzification converts outputs to fuzzy rule base into crisp values. Fuzzy Logic Controller (FLC) has two inputs: message precedence and network status, and an output: routing decision. Rules are expressed in Mamdani form [15]: : where linguistic variables a, b and c represent two process state variables and a control variable (two inputs and one output); Ai, Bj, and Ck are linguistic values (fuzzy sets specify meaning) of linguistic variables a, b, and c in universes of discourse U, V, and W, respectively.Linguistic variables used for fuzzy systems are actual end to end delay (Ae), when a node leaves the network (Nl), number of packets dropped (Nd) and number of RRER generated (Nr). Table 1 explains the definition points of MemBership Function (MBF) of Ae, Nl, Nd , Nr and route preferences for different term names (Low, Medium, High) along with its corresponding shape. IF part of the rules defines the situation and 'then' part defines the response of the fuzzy system. The degree of support (DoS) is used to specify the importance of the rule. The processing of the rules starts with calculating the 'if' part. The operator type of the rule block determines which method is used. The operator types MIN-MAX is used in this work. 675 rules are generated and some of the rules are given below. Some of the rules obtained are tabulated in Table 1. Table 1: Min-Max Rules Ae

Nl

Nd

Nr

DoS

RoutePreference

low low low low low medium

decrease decrease decrease decrease decrease decrease

very_small very_small very_small very_small medium very_small

low medium high high low low

0.99 0.98 0.97 0.9 0.94 0.96

very_high very_low low high high very_low

medium medium medium medium

decrease decrease decrease decrease

very_small very_small small small

medium high medium high

0.93 0.97 0.98 0.99

high very_high low low

high high high high high

decrease decrease steady steady steady

large very_large small small small

low high low low medium

0.92 0.97 0.96 0.91 0.95

medium low low very_high very_low

22378

N.K. Kuppuchamy and R.Manimegalai

Simulations are conducted with the proposed method. Throughput, number of retransmission attempts and number of packets dropped are taken as performance evaluation parameters. Numeric results show that fuzzy approach for routing improves the performance than existing AODV routing protocol. Fuzzy Genetic Approach Incorporating GAs into fuzzy systems overcomes the difficulties of manually choosing appropriate rules and membership functions for a given problem. The GA takes responsibility for the selection of a rule-set that is high-performing and for the tuning of the membership functions in respect to the rule set. This results in highperforming, robust routing algorithms, which can work over a wide parameter range and which are almost completely computer-designed. GA is used to improve the initial rule-set as well as the membership functions of our routing Unit. Success depends on the ability of the selected route to keep up with the required QoS demand, at low cost and without violating policy. For the nodes in the edges, the following QoS parameters have been defined: Throughput, end to end delay and data loss rate. The fuzzy part of the Routing Unit calculates the potentiality of each available route. The output of the Routing algorithm will result after all possible routes are evaluated , or when a route is evaluated as good enough, in terms of the evaluation function becoming greater than a threshold. The next hops that are subject for evaluation are chosen among the neighboring nodes based on its performance evaluation.

Experimental Results The fitness function is a computation that evaluates the quality of the chromosome as a solution to a particular problem A novel fitness function to improve the QoS. The derived fitness function is given by V F (t ) 

art

d

 m

e



2 g

2  in

d

m

 in P D R x Ld

Each chromosome is represented by a fuzzy rules with ‘1’ indicating the rule as selected and ‘0’ indicating the rules is not used. Mutation operators act on an individual chromosome to flip one or more allele values. In the case of bit-string chromosomes, the normal mutation operator is applied to each position in the chromosome. A random number in the interval [0,1] is generated with uniform probability and compared to a pre-determined mutation rate. GA operation is shown in Table 2 GA parameters are

Network Performance Improvement Using Fuzzy Genetic Based Aodv

22379

L d is the link quality V art is the route time out PDR is the packet delivery ratio d m is the max jitter

in is the input packet to the node g is the packet generated by the node Table 2: A Parameters used in fitness function GA Parameters Population size

50

Number of generations Cross over type Cross over rate Mutation rate

500 Two point 0.7 0.05

Comparison based on various mobility nodes Genetic Algorithm is used to improve the initial rule-set as well as the membership functions of our Routing Unit. Success depends on the ability of the selected route to keep up with the required QoS demand, at low cost and without violating policy. Figure shows the Comparison based packet delivery ratio and end to end delay respectively.

Average Packet Delivery Ratio

1 0.95 0.9

0.85 0.8

0.75 0.7

0.65 0.6 15

30 AODV

45

60

Node mobility in KMPH Fuzzy AODV

75 Fuzzy GA

Figure 1: Comparison Based Packet Delivery Ratio

90

22380

N.K. Kuppuchamy and R.Manimegalai

Figure 1. shows the average packet delivery ratio for Fuzzy GA achieves better results than AODV and Fuzzy AODV in the range of 2.1% to 15.31% for varying mobility speed.

Average End to End Delay in second

0.08

0.06

0.04

0.02

0 15

30

45

60

75

Node mobility in KMPH Fuzzy AODV

AODV

90

Fuzzy GA

Figure 2: Comparison based on end-to-end delays Figure 2. shows the average End to End Delay for Fuzzy GA achieves better results by lowering delay than AODV and Fuzzy AODV in the range of 3.54% to 146.77% for varying mobility speed.

AODV

Simulation time (sec) Fuzzy Fuzzy-GA

Figure 3: Throughput

353

338

324

310

295

281

266

252

238

223

209

194

180

166

151

137

1100000 1050000 1000000 950000 900000 850000 800000 750000 700000 650000 600000 122

Throughput (bits/sec)

Simulation Result

Network Performance Improvement Using Fuzzy Genetic Based Aodv

22381

Figure 3. shows that the throughput is around 9.11% more in the proposed FuzzyGA method when to Fuzzy approach and more by 19.08% when compared to AODV.

Conclusions In this works obtained using the proposed fuzzy based AODV are compared to with the conventional. Performance analysis reveals that the proposed fuzzy approach for routing improves most of the QoS parameters when compared to the conventional AODV. However further improvements in the algorithm need to be carried out to decrease the overall network delay and Packet delivery ratio. To implement QoS routing, optimization is the best approach. GA is used to improve the initial rule-set as well as the membership functions of the routing. Success depends on the ability of the selected route to keep up with the required QoS demand, at low cost and without violating policy. GAs offers an optimization method in which the stochastic search algorithm is based on biological principles of evolution like fitness and mutation. GA finds the shortest path in less time and is quicker than routing protocols. Simulation results demonstrated that the Throughput around 9.11% more in the proposed FuzzyGA method when comparing to the AODV protocol.

22382

N.K. Kuppuchamy and R.Manimegalai

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

Mewara, H, S., and Porwal, S., 2012, "Throughput Performance Comparison of Reactive and Proactive Routing Protocols in Mobile Ad Hoc Networks using OPNETv14.5", International Conference in Recent Trends in Information Technology and Computer Science (ICRTITCS 2012) Proceedings published in International Journal of Computer Applications® (IJCA) (0975 – 8887). Uma, M., and Padmavathi, G., 2009, "A Comparative Study And Performance Evaluation Of Reactive Quality Of Service Routing Protocols In Mobile Adhoc Networks", Journal of Theoretical and Applied Information Technology, 6(2), pp. 223-239. Fadil, Y, A., 2010, "Routing Using Genetic Algorithm for Large Networks", Diyala Journal of Engineering Sciences. Vol. 03, No. 02, pp. 53 -70. Priyadharshini, T., and Arunachalam A, R., 2013, "Efficient Genetic Algorithm for Optimal Routing In Ad Hoc Networks”, International Journal of Advanced Research in Computer Science and Software Engg. 3(2), pp. 364-367. Hiremath, P, S., and Joshi, S, M., 2012, "Energy Efficient Routing Protocol with Adaptive Fuzzy Threshold Energy for MANETs", International Journal of Computer Networks and Wireless Communications (IJCNWC), ISSN: 2250-3501 Vol, 2. Jing Zuo, Soon Xin Ng., and LajosHanzo., 2010, "Fuzzy Logic Aided Dynamic Source Routing in Cross-Layer Operation Assisted Ad Hoc Networks". Page 1-10. Vamsee Mohan, B., Sunil Kumar, V., and VamsiNath, J., 2012, "A Reliable Routing Algorithm in MANET Using Fuzzy". International Journal of Advanced Research in Computer Science and Software Engineering, Volume 2, Issue 10. Abirami, S., Dhanasekaran, R., and Bhanumathi, V., 2012, "A Balanced Approach for Power Aware Routing in MANET using Fuzzy Logic". International Conference on Recent Trends in Computational Methods, Communication and Controls, (ICON3C) ICON3C (5): 15-20. Anjum, A., Mohammed and GihanNagib., 2012, "Optimal Routing In AdHoc Network Using Genetic Algorithm", International Journal Advanced Networking and Applications 1323 Volume: 03, Issue: 05, Pages: 13231328. VikasSiwach., Dr. Yudhvir Singh, Seema, DheerDhwaj Barak., 2012, "An Approach to Optimize QOS Routing Protocol Using Genetic Algorithm in MANET", IJCSMS International Journal of Computer Science and Management Studies, Vol. 12, Issue 03. Ghalavand, G., Dana, A, Ghalavand, A., and Rezahosieni, M., IEEE 2010, August, "Reliable routing algorithm based on fuzzy logic for Mobile Ad

Network Performance Improvement Using Fuzzy Genetic Based Aodv

[12]

[13] [14]

[15]

22383

hoc Network", Advanced Computer Theory and Engineering (ICACTE), 2010 3rd International Conference on Vol. 5, pp. V5-606. Sujatha, K, S., Dharmar, V., and Bhuvaneswaran, R, S., 2012, "Design of genetic algorithm based IDS for MANET", Recent Trends in Information Technology (ICRTIT), International Conference on pp. 28-33. Alandjani, G., Johnson, E., 2003, "Fuzzy routing in ad hoc networks", In Performance, Computing, and Communications Conference, pp. 525-530. Gorbanzadeh, F., and Kazem, A, P., 2012, "Hybrid Genetic Algorithms for Solving Winner Determination Problem in Combinatorial Double Auction in Grid", IAES International Journal of Artificial Intelligence (IJ-AI), 1(2), 53-62. Mueller, S., and Tsang, R, P., and Ghosal, D., 2004, "Multipath routing in mobile ad hoc networks: Issues and challenges", In Performance Tools and Applications to Networked Systems (pp. 209-234).

22384

N.K. Kuppuchamy and R.Manimegalai