Information Theoretic Approach to Signal Feature Detection for Cognitive Radio

Cognitive radio (CR) systems need to be able to adjust the transceiver characteristics in response to stimuli received from the radio environment. Therefore, monitoring the wireless signal in real-time for stimuli such as unexpected changes due to sporadic interference in radio frequency band of operation is at the core of such systems. In this paper, a method of detecting anomalies in a periodic signal by means of statistical analysis of its envelope is described. The proposed scheme makes use of the Kullback-Leibler divergence between probability distributions drawn from analogous segments of the periodic signal to detect anomalous events. Experiments conducted on real wireless signals suggest that the method described is simple, robust and effective for the analysis of periodic signals.