An Archive Enhanced Hybrid Genetic Algorithm for Shortest Path Routing Problem

Genetic Algorithms (GA) like any other standard metaheuristic technique suffer from a limitation that the candidate solutions are revisited several times during the course of evolution. This introduces in the standard GA (SGA), the problem of duplicity of offsprings and consequently premature convergence. Further, the success of a population-based metaheuristic such as GA is greatly influenced by the initial set of building blocks available in it's gene pool. In this paper, we propose a hybrid GA that uses greedy initialization of population and uses a Archiving_Tree_Structure based chromosomal representation, that efficiently stores all the generated solutions during the successive iterations of the algorithm, ensuring uniqueness of the individuals and consequently sufficient diversity in the population. The results of the proposed GA are encouraging when applied to the Shortest Path problem with respect to increased convergence rate and reduced number of generations required to reach the optima.