Kalman Filter Enhanced Parametric Classifiers for Spectrum Sensing Under Flat Fading Channels

In this paper we propose and investigate a novel technique to enhance the performance of parametric classifiers for cognitive radio spectrum sensing application under slowly fading Rayleigh channel conditions. While trained conventional parametric classifiers such as the one based on -means are capable of generating excellent decision boundary for data classification, their performance could degrade severely when deployed under time varying channel conditions due to mobility of secondary users in the presence of scatterers. To address this problem we consider the use of Kalman filter based channel estimation technique for tracking the temporally correlated slow fading channel and aiding the classifiers to update the decision boundary in real time. The performance of the enhanced classifiers is quantified in terms of average probabilities of detection and false alarm. Under this operating condition and with the use of a few collaborating secondary devices, the proposed scheme is found to exhibit significant performance improvement with minimal cooperation overhead.