Bio-Inspired Multiobjective Optimization for Designing Content Distribution Networks

Abstract

This article studies the effective design of content distribution networks over cloud computing platforms. This problem is relevant nowadays to provide fast and reliable access to content on the internet. A bio-inspired evolutionary multiobjective optimization approach is applied as a viable alternative to solve realistic problem instances where exact optimization methods are not applicable. Ad hoc representation and search operators are applied to optimize relevant metrics from the point of view of both system administrators and users. In the evaluation of problem instances built using real data, the evolutionary multiobjective optimization approach was able to compute more accurate solutions in terms of cost and quality of service when compared to the exact resolution method. The obtained results represent an improvement over greedy heuristics from 47.6% to 93.3% in terms of cost while maintaining competitive quality of service. In addition, the computed solutions had different tradeoffs between the problem objectives. This can provide different options for content distribution network design, allowing for a fast configuration that fulfills specific quality of service demands.