Bayesian modeling for dealing with uncertainty in cognitive radios.

摘要

Wireless communication systems can be affected by several factors, including propagation losses, co-channel interference, and multipath fading. Uncertainty affects all of these factors making it even more difficult to model these systems. This dissertation proposes the use of probabilistic graphical models (PGM), such as Bayesian Networks and Influence Diagrams, as the core for reasoning and decision making in adaptive radios operating under uncertainty. PGM constitute a tool to understand and model complex relations among random variables. This dissertation explains how to build effective communication models that perform its functions under uncertainty. In addition, this work also presents a spectrum sensing technique based on the autocorrelation of samples to estimate the utilization level of wireless channels.