Sensing OFDM Systems Under Frequency-Selective Fading Channels

In this paper, we consider the detection of orthogonal frequency-division multiplexing (OFDM) signals under frequency-selective fading channels for cognitive radio. First, we propose a new decision statistic for the signal detection based on the special feature-cyclic prefix embedded in OFDM signals. The probability distribution function (pdf) of the statistic is derived under the hypothesis of signal presence or absence. Then, the generalized likelihood ratio test (LRT) and the pdf's under the two hypotheses are derived to obtain the theoretical false-alarm rate (FAR) and the probability of detection (PD). After that, a suboptimal test is proposed to resolve the interference/noise uncertainty problems met for many detection algorithms such as the energy detection in practice. The theoretical FAR and PD are derived, and the theoretical threshold for any given FAR according to Neyman-Pearson criterion is accordingly given. Finally, extensive simulations are performed to confirm the derived theoretical results and demonstrate the promising detection.