Adaptive blind wideband spectrum sensing for cognitive radio based on sub-Nyquist sampling technique

Spectrum sensing is the core process in cognitive radio networks. Spectrum sensing of wideband signal results in enhancement of system reliability and effectiveness. In this paper, a modified system of wideband spectrum sensing based on non-uniform Multi-coset samplers is proposed. The proposed system is designed to work on totally blind input wideband signal. Lomb-Scargle periodogram is used to detect the number of active bands (N). System parameters are selected to decrease the average sub-Nyquist sampling frequency that results in activating lower number of parallel branches of non-uniform samplers. Then, a lower power consumption and complexity is achieved. The proposed system is able to adapt its operating parameters, time offsets and number of active branches continuously. Simulation results show accurate reconstructed signal, high probability of detection and low errors even at a small number of sub-Nyquist samples and low SNR.