P3PGA: Multi-Population 3 Parent Genetic Algorithm and its Application to Routing in WMNs

Abstract

This paper proposes a new multi-population based global optimization algorithm called Parallel Three Parent Genetic Algorithm (P3PGA). The proposed approach is an extension of 3PGA approach. Performance of the proposed algorithm was evaluated on the eleven CEC-2014 benchmark functions. We compared the performance of proposed P3PGA algorithm with 16 other algorithms. It was observed that out of the 11 benchmarks functions P3PGA outperformed all other 16 approaches on 5 benchmark functions. Out of 5, for 2 functions the best performance of P3PGA was also equaled by few other approaches. For the other 3 functions the performance of P3PGA was unmatched by any of the other 16 algorithms. Further, this paper proposes a new P3PGA based optimal route evaluation approach for routing in Wireless Mesh Networks. The proposed approach was implemented in MATLAB and simulated for various WMN sizes and scenarios. We compared its performance with 8 other approaches namely, Ad-hoc On Demand Distance Vector (AODV) approach, Dynamic Source Routing (DSR), Genetic Algorithm (GA), Biogeography Based Optimization (BBO), Firefly Algorithm (FA), Ant Colony Optimization (ACO), BAT and Big Bang-Big Crunch (BB-BC) based optimal cost route evaluation approaches. The P3PGA based approach outperformed all other 8 approaches for the WMNs sized 1000 nodes and above.