Signal Detection for Cognitive Radio networks using correlation of random data matrix

Cognitive Radio (CR) systems offer the opportunity to improve spectrum utilization by detecting unoccupied spectrum band and adapting the transmission to those bands while avoiding the interference to primary users. Spectrum sensing is the fundamental task in cognitive radio and it needs to detect signals presence under strict requirement such that secondary users (unlicensed users) can use the licensed spectral band without interfering primary users (licensed users). Energy Detection (ED) is a widely used spectrum sensing method for CR networks since it is very easy to implement and does not require any knowledge of the primary signal. However, ED requires accurate noise power to perform the detection and its performance is poor at low SNR and very sensitive to noise uncertainty. ED is optimal for detecting independent and identically distributed (iid) signals, but not optimal for detecting correlated signals. To overcome the limitations of ED, a novel approach is proposed based on the correlations of the sample covariance matrix computed from the random data matrix of the received signal samples in this paper. This approach does not need any information of the signal, the channel and noise power as a priori. Simulation approach is used to set the threshold for the target probability of false alarm. Since sample covariance matrix catches the correlations among the signal samples, the proposed method is better than the energy detection algorithm. Simulations based on recorded voice signals and iid signals are presented to verify the proposed method and results are compared with ED algorithm.