Performance analysis and optimization of frequency division duplex based virtual full-duplex communication systems

Virtual full-duplex (VFD) communication systems have become smart alternative over residual self-interference (RSI) affected full-duplex (FD) systems and thus has drawn significant research attention in the recent years. However, in comparison to FD, the VFD system models studied in the literature are based on successive relaying (SR) that require an extra time phase and a double number of relays for each destination user. Thus, we propose a novel frequency division duplex (FDD) based VFD (FDD-VFD) system model that can not only match FD in terms of resource utilization but also outperform both SR-VFD and FD. We consider two practical scenarios of absence of inter relay interference (A-IRI) and presence of IRI (P-IRI). We analyze the outage probability and ergodic rate performance metrics and present their analytical expressions in closed-form over Nakagami-m fading channels. Next, to minimize the effect of inter-relay interference in P-IRI scenario, we investigate optimization problems of minimizing the system outage probability and maximizing the ergodic sum rate for jointly allocating transmit power of the relays. We utilize particle swarm optimization algorithm and deep learning architecture to solve the formulated problems. Extensive evaluations are carried out over different system parameters to prove the efficacy of proposed FDD scheme and obtain key inferences.