Fundamental Limitations on Pilot-based Spectrum Sensing at Very Low SNR

Spectrum sensing is one of the most challenging issues of Cognitive Radio communications. The possibility of extremely low signal-to-noise ratio (SNR) of the received signal poses a fundamental challenge to spectrum sensing. In this paper, pilot-based spectrum sensing for OFDM signals is investigated. It is shown that the existing pilot-based OFDM spectrum sensing algorithms suffer from the frequency offset between the transmitter and sensing devices, as well as the noise uncertainty in the sensing threshold design. We consequently propose a robust pilot-based spectrum sensing algorithm for low SNR OFDM signals using a sliding frequency correlator. The proposed algorithm processes additional bandwidth to eliminate the impact of frequency offset. In addition, considering the unknown noise statistics and its time-varying nature, a ratio threshold which is not sensitive to the noise power level is derived for spectrum sensing. Our theoretical analysis and simulation results show that this algorithm can achieve exceptionally good sensing performance at very low SNR, while being insensitive to time and frequency offsets and requiring no information of the noise statistics.