A Bayesian Network Model of the Bit Error Rate for Cognitive Radio Networks

TL;DR

This paper introduces a Bayesian network model that predicts the bit error rate in cognitive radio networks, enabling better environmental understanding and decision-making for improved wireless communication performance.

Contribution

It presents a novel probabilistic graphical model capturing causal relationships among key variables affecting BER in cognitive radios.

Findings

The Bayesian network accurately infers environmental variables from BER evidence.

The model helps cognitive radios adapt their parameters based on environmental conditions.

The approach improves decision-making in dynamic spectrum access scenarios.

Abstract

In addition to serve as platforms for dynamic spectrum access, cognitive radios can also serve as a method for improving the performance of wireless communication systems by smartly adjusting their operating parameters according to the environment and requirements. The uncertainty always present in the environment makes the practical implementation of the latter application difficult. In this paper, we propose a probabilistic graphical model, Bayesian network that captures the causal relationships among the variables bit energy to noise spectral density ratio (EbN0), carrier to interference ratio (C/I), modulation scheme (MOD), Doppler phase shift (Dop_Phi), and bit error rate (BER). BER indicates how the communication link is performing. The goal of our proposed Bayesian network is to use the BER as evidence in order to infer the behavior of the other variables, so the cognitive radio can learn how the conditions of the environment are, and based on that knowledge make better informed decisions. This model along with the method used to build it are described in this paper.