Sparse Frequency Domain Spectrum Sensing and Sharing Based on Cyclic Prefix Autocorrelation

Cognitive radio (CR) is considered an important solution to the current spectral scarcity, which is expected to be a significant issue in the next generation of wireless communication systems, namely 5G. Wideband spectrum sharing and sensing constitute highly desirable features of CR systems as they aim to increase the probability of identifying available spectral bands, which ensures a more efficient resource utilization. The present work proposes an efficient frequency-domain cyclic prefix (CP) autocorrelation based wideband spectrum sensing and sharing method that can provide accurate detection of orthogonal frequency-division multiplexing (OFDM) based primaries in wideband CR systems. Novel analytic expressions are derived for the corresponding threshold, probability of false alarm and probability of detection in the presence of noise uncertainty (NU) and frequency selectivity. The derived models are validated by extensive comparisons with respective results from computer simulations. It is demonstrated that the introduced autocorrelation based sensing method is able to counteract NU and the frequency-selective multipath channel effects in realistic wideband communication scenarios. Furthermore, the method facilitates partial band sensing, allowing the sensing of weak OFDM-type primary user (PU) signals in channels which are partly overlapped by other strong PU or CR transmissions. This is considered a crucial element in practical spectrum sharing scenarios. Since, the proposed sensing method makes use of sparsity in the spectral domain, it can be technically considered as compressed sensing method. The flexibility of this approach supports robust wideband multi-mode, multi-channel sensing with low complexity. Finally, it is shown that the offered results are particularly useful in the context of spectrum sharing as their high performance and reduced complexity can enable the co-existence of non-exhaustive yet highly efficient algorithms.