A hybrid EDA for load balancing in multicast with network coding

Load balancing is one of the most important issues in the practical deployment of multicast with network coding. However, this issue has received little research attention. This paper studies how traffic load of network coding based multicast (NCM) is disseminated in a communications network, with load balancing considered as an important factor. To this end, a hybridized estimation of distribution algorithm (EDA) is proposed, where two novel schemes are integrated into the population based incremental learning (PBIL) framework to strike a balance between exploration and exploitation, thus enhance the efficiency of the stochastic search. The first scheme is a bi-probability-vector coevolution scheme, where two probability vectors (PVs) evolve independently with periodical individual migration. This scheme can diversify the population and improve the global exploration in the search. The second scheme is a local search heuristic. It is based on the problem-specific domain knowledge and improves the NCM transmission plan at the expense of additional computational time. The heuristic can be utilized either as a local search operator to enhance the local exploitation during the evolutionary process, or as a follow-up operator to improve the best-so-far solutions found after the evolution. Experimental results show the effectiveness of the proposed algorithms against a number of existing evolutionary algorithms. (C) 2017 Elsevier B.V. All rights reserved.