Non-linear space-time Kalman filter for cooperative spectrum sensing in cognitive radios

A cooperative spectrum-sensing problem for a cognitive radio (CR) system is investigated, where CR users collaborate to sense and track the primary users´ (PUs) activities in frequency-selective fading channels. To sense PUs activities, channel gain estimation is performed by CRs through space-time extended Kalman filtering (STEKF). The STEKF method captures the channel gain from any point in space to each CR at each frame for a specific range of frequencies. The proposed channel gain tracking enables CRs to detect the transmit power, location and the number of active subcarriers of each PU via a time spatial weighted non-negative Lasso (TSWNL) algorithm. The TSWNL exploits the sparsity of the PUs activities in a geographical area to track PUs activities in frequency-selective fading channels. Numerical results indicate that the proposed spectrum sensing based on STEKF significantly improves the performance of CRs in tracking of PUs activities.