Enhancing revenue maximization with adaptive WRR

In the future converged wireless and wired networks will be required to provide support to a number of different types of traffic, each with its own particular characteristics and quality of service parameters including e.g. guaranteed bandwidth, jitter, and latency. The customers of different classes pay different prices to the service provider, who must share resources in a plausible way. Differentiation can be implemented by using a multiqueue system, where each queue corresponds to one service class. In this paper, an adaptive weighted round robin (WRR) based algorithm for traffic allocation is presented and studied in the single node case. The weights in the adaptive gradient type WRR algorithm are updated using revenue as a target function. Due to the adaptive nature of the algorithm, it can operate in the nonstationary environments. In addition, it is nonparametric and deterministic in the sense that any assumptions about call density functions or duration distributions are not made.