A Fluid model based Heuristic for Optimal Speed-scaling in Bandwidth-sharing Networks

International audienceWe investigate the energy-delay tradeoff in multi-class queues in which the server can regulate its speed ac- cording to the load of the system. Assuming that the queue is initially congested, we investigate the rate al- location to the classes that drains out the queue with minimum total energy and delay cost. We propose to solve this stochastic problem using a deterministic fluid approximation. We show that the optimal-fluid solution follows the well-known cμ rule and obtain an explicit expression for the optimal speed. Numerical results show the utility and the applicability of the fluid-optimal policy