Taguchi's Quality Loss Function Approach for Optimizing the Performance of AODV MANET Routing Protocol

Mobile ad hoc network (MANET) is composed of mobile nodes that can be formed without the need of any pre-existing infrastructure or centralized administration. Mobile nodes independently form a network over radio links in an ad-hoc fashion. Routing is done arbitrarily and randomly, due to frequent and unpredictably changes in network wireless topology. To provide and maintain better connectivity for this type of network, a number of routing protocols have been proposed. Simulator models are used for the performance evaluation of the routing protocols before their implementation in real life. CBR traffic is the commonly used traffic type to evaluate MANET routing protocols performance. But TCP traffic is connection oriented and more reliable than CBR traffic. More than 80% of the Internet traffic is based on TCP. Therefore it is necessary to analyze the impact of TCP traffic types on MANET routing protocols before selecting a particular protocol for the network. In this paper, the performance of a most widely used reactive MANET routing protocol "Ad hoc On-demand Distance Vector Routing" (AODV) is optimized on TCP traffic by using Taguchi design of experiment (DOE) approach. We inferred the main effect of five factors (Performance Parameters) i.e., No. of nodes, Pause time, Maximum node movement speed, Maximum no. of concurrent connections and Network area. For the evaluation purpose five performance metrics i.e., Routing overhead, Average end-to-end delay, Packet delivery ratio, Throughput and Energy consumed have been considered. The study also uses Taguchi quality loss function for tackling multiple performance metrics and various network parameters simultaneously. Taguchi experimental design comprising of L8 orthogonal array of two levels for each parameter is used to organize the parameters and their range. A statistical analysis of means (ANOM) and analysis of variance (ANOVA) is performed to find the statistically significant optimal values of these parameters in order to achieve the best performance of AODV MANET routing protocol.