Sub-Nyquist Cyclostationary Detection of GFDM for Wideband Spectrum Sensing

Spectrum scarcity is a challenging problem in wireless communications: high data rates are needed to support 5G new technologies. However, the spectrum is underutilized. To address this problem, cognitive radio (CR) is proposed to exploit the underutilized spectrum. The main requirement for the future CR networks is wideband spectrum sensing, which provides secondary users with the available frequency bands across a wide frequency range. Secondary users should fill these bands without causing interference to licensed users. Thus, new waveforms are proposed for the 5G physical layer. Generalized frequency division multiplexing (GFDM) is considered to be a contender for the 5G new physical layer. The GFDM is a block-based waveform that is suitable for fragmented spectrum scenarios and is designed to overcome the drawbacks of orthogonal frequency-division multiplexing (OFDM) used in 4G. The GFDM is the perfect candidate for 5G and CR technologies. Considering the cyclostationarity properties of modulated signals, we propose an optimized recovery method for the GFDM signals in the wideband regime. By exploiting the signal sparsity, we can recover the spectral correlation function (SCF) of the GFDM from digital samples of the GFDM taken at a sub-Nyquist rate to reduce the sampling time. Furthermore, a generalized likelihood ratio test is applied to the recovered function to detect multiple signal sources and identify the spectrum occupancy. The numerical results show that our method achieves a high probability of detection at a low signal-to-noise ratio (SNR) with robustness in terms of rate reduction in wireless networks.