Link-Layer Abstractions for Utility-Based Optimization in Cognitive Wireless Networks

Abstract

We argue that utility functions are very natural tools for formulating optimization problems in cognitive wireless networks. However, their use mandates that the platform used supports well-defined APIs and abstractions for retrieving information necessary to calculate the application utilities. We present the design of a unified link-layer API that offers precisely this functionality in the context of link-layer information. The API developed offers powerful query and notification mechanisms, which considerably simplify the implementation of technology-independent reasoning engines for cognitive networks. Prototype implementations of the API show that despite the rich functionality offered, the implementation can be made very lightweight and fast, enabling optimization decisions even on per-frame basis with standard user terminals