An efficient anycast routing algorithm for load-balancing based on evolutionary algorithm

Anycast is a new “one-to-one-of-many” communication method in IPv6. By taking advantages of anycast paradigm and evolutionary algorithm, an efficient anycast routing algorithm for load-balancing based on evolutionary algorithm is proposed in this paper. A new approach of initializing population is presented that the initial population is related to the load of anycast group, thus the diversity of initial population is improved and the load of anycast group can be balanced effectively. Meanwhile, penalty functions of QoS parameter with higher distinction degree are presented, so the route with better QoS can be achieved. In simulation experiments, a new approach of generating a link's delay is introduced. At last, simulation experiments are carried out in networks which randomly generated by the model of Waxman, the simulation results show that our algorithm can achieve the optimal anycast route satisfied multi-QoS constraints in a few iterations, and the load of network is balanced effectively.