GFDM-Modulated Full-Duplex Cognitive Radio Networks in the Presence of RF Impairments

Abstract

This paper investigates the problem of sum rate maximization for a full-duplex (FD) generalized frequency division multiplexing (GFDM) based secondary user (SU) link, operating in a spectrum hole. The right and left adjacent channels of the spectrum hole have two active primary users (PUs), and thus adjacent channel interference (ACI) on them must be below a threshold. For SU link, radio frequency (RF) impairments including phase noise, in-phase (I) and quadrature (Q) imbalance, and carrier frequency offset (CFO) are considered and analog domain and digital domain self-interference (SI) cancellation techniques are applied. We consider two cases: (1) two independent oscillators for local transmitter and receiver, (2) single shared oscillator between them. We derive the powers of residual SI, desired signal and noise and signal-to-interference-plus noise ratio (SINR). Furthermore, power spectral density (PSD) of FD transmitter is calculated and ACI is formulated. By using successive convex approximation, sum rate maximization problem subject to ACI limits on adjacent PUs is defined and solved. Finally, we show that FD GFDM for the SU link can achieve twice higher sum rate than FD orthogonal frequency division multiplexing (OFDM).