Signal Spectrum Sensing Robust to Noise Uncertainty

Lack of robustness against noise uncertainty is a bottleneck of current spectrum sensing strategies to detect the primary signals. Due to noise uncertainty, the performance of traditional detectors such as matched filters, energy detectors, and even cyclostationary detectors deteriorates rapidly at low Signal-to-Noise Ratio (SNR). In view of this, a new entropy-based spectrum sensing scheme is proposed in this paper. The entropy of the signal is estimated in the frequency domain with a probability space partitioned into a fixed dimension. It is also proved that the proposed scheme is strongly robust against noise uncertainty. Simulation results verify the robustness of the proposed scheme against noise uncertainty, and further show 5dB, 6dB and 4dB performance improvements compared with state-of-the-art methods such as differential entropy-based detector, energy detector and cyclostationary detectors. Experimental results also show that the sample size can be reduced of about 65% to get a traditional performance.