Minimizing Cost and Delay for Video Datacenters Using Efficient Fuzzy Dominance-Based Particle Swarm Optimization

In multi-datacenter video forwarding environments, monetary cost and forwarding delay are both important concerns. This paper studies the problem of scheduling end user requests onto the media servers of multiple video datacenters (VDCs), with the objective of joint optimization of cost and delay. The studied scheduling problem is formulated as a multi-objective optimization problem (MOP) with a set of binary decision variables. We introduce a new metric of efficient fuzzy dominance (EFD) to evaluate the quality of MOP solutions with low computational complexity. With this new EFD metric, we propose a EFD-based particle swarm optimization (EFDPSO) algorithm to seek for the optimized scheduling solutions. EFDPSO employs a position-based mapping mechanism to convert each particle into a scheduling solution, represented by a request sequence, and develops a shortest path-based strategy to schedule the requests in the sequence such that cost and delay can be calculated. By using the EFD metric to determine the dominance among all the solutions mapped from the particles, EFDPSO identify a set of non-dominated MOP solutions and relies on PSO’s update rule to iteratively improve the non-dominated solution set. We perform extensive experiments to verify that the EFD metric can substantially reduce the computation time in solution evaluation and the proposed EFDPSO leads to higher quality of MOP solutions compared with baseline metaheuristic algorithms.