Autocorrelation Based Spectrum Sensing of FBMC Signal

The focus of this paper is on a feature detector for filter bank multicarrier (FBMC) signal in cognitive radio. In this paper, we first prove that the FBMC signal samples are uncorrelated with each other. However, if the FBMC signal is processed by our proposed method, then the autocorrelation function (ACF) of the FBMC signal becomes non-zero at the lag equal to a number of subcarriers. On the other hand, additive white Gaussian noise (AWGN) samples after the same proposed processing remain uncorrelated. Using this feature, an autocorrelation-based feature detector is proposed to detect FBMC signals in the noise.

The main advantage of the proposed detector is that, unlike blind detectors, this detector can distinguish between FBMC signal and noise (or interference). Next, the distribution of the test statistic of the proposed detector is derived under a noise-only scenario so that the threshold of the Neyman-Pearson detector can be designed to maintain a constant false alarm rate while maximizing the probability of detection. Simulation results demonstrate the efficacy of the proposed detector.

KEYWORDS—Autocorrelation, cognitive radio, FBMC signal, feature detector.