Multi-criteria Fitness Function based Genetic Optimization of AODV Routing Protocol in MANETs

Abstract

Abstract The artificial intelligence search techniques are widely used to achieve performance enhancement in wireless communication systems. One such methodology of evolutionary genetic programming inspired by nature is essentially appropriate for optimized operation in the context of wireless multi-hop ad-hoc networks with several challenges to provide the necessary network services. In this paper, we develop a reliable and efficient data routing scheme employing the conventional AODV protocol based on the dynamic genetic algorithm. It is aimed at effectively allocating the scarce radio resources and improving the QoS among the wireless devices in MANETs by joint optimization of network attributes including the data transfer rate, link transmission power and round trip delay. The proposed genetic algorithm based routing scheme utilizes two different fitness functions, together with binary data coding and decoding, single-point crossover and random mutation operators to assess the fitness measure of the specific solution space and network operational characteristics. This computational learning method is trained through the sample dataset obtained via the simulation experiments of the basic AODV routing scheme. With the emerging size of the sample network data records, the deployed polynomial and logarithmic fitness functions are compared in terms of power consumption and delay metrics to design a robust and adaptive data communication scheme. Finally, our smart network data learning and genetic optimization model is compared with the previous related models to demonstrate its improved performance in terms of lower power consumption, higher throughput, and greater values of average fitness measure.