Spectrum Sensing of OFDM Signals Over Multipath Fading Channels and Practical Considerations for Cognitive Radios

Despite the promising role of orthogonal frequency-division multiplexing (OFDM) in communication systems, the spectrum sensing of OFDM signals and its practical considerations for cognitive radios (CRs) remain vital and challenging topics. This paper presents a new scheme for detecting OFDM signals based on the Neyman–Pearson (NP) principle. In contrast to conventional approaches in which additive white Gaussian noise (AWGN) channels are considered or empirical second-order statistics based on correlation coefficients are employed, to improve the detection performance, the proposed approach involves considering multipath fading channels and the classical NP detector. The log-likelihood ratio (LLR) test is formulated without requiring additional pilot symbols by using the redundancy of the cyclic prefix. Analytical results indicate that the LLR of received samples is the sum of the log-likelihood function (LLF) of the samples, which is typically used for estimating unknown parameters, and the LLR of an energy detector (ED). These results provide insight into the NP detector and the relationship between the NP detector, a detector based on the LLF, and the ED. 1 Because many unknown parameters must be estimated in the NP detector, two practical generalized log-likelihood ratio test (GLRT) detectors are designed. To develop a channel-independent GLRT, which is crucial for achieving favorable performance over multipath fading channels, the complementary property of the correlation coefficient is employed to derive an estimate independent of multipath channel profiles. Simulation results confirm the advantages of the proposed detector compared with the state-of-the-art detectors. 1 EDs are optimal only for independent and identically distributed random samples.