Facing the SNR Wall Detection in Full Duplex Cognitive Radio Networks Using a GLRT Multipath-Based Detector

Recently, with the advances in self-interference cancellation, full-duplex (FD) techniques have become a feasible approach to improve the spectrum usage in new generation wireless systems. In particular, in cognitive radio networks, FD allows a secondary user to simultaneously sense the spectrum and transmit, improving the network efficiency. However, the residual self interference (RSI) arising from an imperfect cancellation, represents a bottleneck for the spectrum sensing performance. This paper analytically demonstrates that RSI represents a colored noise which leads to the SNR Wall phenomenon making the typical OFDM signal detectors no longer robust. To overcome this drawback, we propose a generalized likelihood ratio test detector exploiting the multipath correlation due to the primary user's (PU) channel. It is proven that the proposed method is not affected by the SNR Wall, hence it is robust w.r.t an imperfect self-interference cancellation. Specifically, best detection performance is achieved when an approximated knowledge of the maximum delay spread of the PU channel is available, otherwise it is shown that a limited performance degradation occurs. Moreover, provided results highlight that our approach outperforms the classical alternatives in different scenarios, including when PU employs a mixed numerology OFDM signal, which characterizes the 5G New Radio waveform.

Automated summary

This paper demonstrates the feasibility of full-duplex techniques in improving spectrum usage in wireless systems, proposes a generalized likelihood ratio test detector exploiting the multipath correlation due to the primary user's channel, and proves its robustness with respect to imperfect self-interference cancellation.