Heuristic Methods for Resource Management in Multimedia-On-Demand Network

摘要

In this paper, we propose heuristic methods - tabu search (TS) and genetic algorithm (GA) - to design a two-level hierarchical and decentralized Multimedia-On-Demand (MOD) network architecture. To optimize the MOD network resource based on cost analysis including installation costs for servers, program storage costs, and transmission costs, both server location and storage allocation with program allocation policy were considered. In applying the proposed methods to the problem, many of the elements are elaborated in order to improve solution quality and computational efficiency: neighborhood structure and search strategy in TS and the genetic representation, evaluation function, genetic operators and procedure in GA. The results of extensive computational simulations showed that the proposed methods provided high quality solutions within reasonable time period of computations and could be used to manage and reallocate resources.