Achieving optimal elastic traffic rewards in dynamic multichannel access

This paper proposes objective functions for dynamic multichannel access (DMA) networks that enable spectrum users (SUs) to assess, locate, and exploit available spectrum opportunities effectively, thereby maximizing the SU´s rewards measured in terms of the average received throughput. We show that the proposed objective functions are: near-optimal, as they achieve high rewards; scalable, as they perform well in small- as well as large-scale DMA networks; learnable, as they allow SUs to reach up near-optimal rewards very quickly; and distributive, as they are implementable by requiring local information sharing only.