Semi-blind locally optimum detection for spectrum sensing in cognitive radio

Abstract

Spectrum sensing in cognitive radio becomes a challenging task when the signals received at the secondary users' transmitters exhibit low power. Locally optimum detectors (LOD) are therefore desirable thanks to their optimality in the low SNR regime. Here, we assume that the primary user transmits a training sequence, and propose a semi-blind LOD (SBLOD). In the case of BPSK signals, the test statistic of the proposed SBLOD is shown to be a weighted sum of the matched filter output, the energy and pseudo-energy. For higher size constellations, the SBLOD reduces to a linear combination of the matched filter and the energy detector. Although combining the matched filter and energy detector is a classical approach, our study provides a systematic and (locally) optimal way of combining these detectors. Simulations results show the merits of the proposed detector.