Video server deployment using a genetic algorithm with deterministic initialization strategy

Abstract

Video server deployment problem (VSDP) is a crucial problem in enhancing video viewing experience. It is difficult for traditional deterministic optimization algorithm to solve VSDP because it is non-polynomial hard (NP-hard). Current server deployment patterns are challenged by the growing scale of networks, and it is therefore desirable to design appropriate optimization algorithm to cope with this issue. Owing to the promising performance in solving NP-hard problems, a genetic algorithm (GA)-based optimizer is developed in this paper. We propose a novel deterministic initialization strategy to deploy the GA population in promising initial positions for performance enhancement. In addition, by embedding the maximum cost minimum cost flow (MCMF) in the fitness evaluation, the algorithm optimizes the server deployment and the bandwidth allocation concurrently. This way, the proposed GA is comprehensive, which deals with different aspects involved in VSDP. Experimental results and the comparisons with other algorithms indicate that the proposed algorithm achieves superior performance in terms of solution quality and convergence rate.