Optimal power allocation for non-orthogonal multiple access enabled full-duplex underlay cognitive relay networks under partial relay selection

This article considers non-orthogonal multiple access (NOMA) enabled full-duplex (FD) underlay cognitive relay networks (ie, NOMA-FDCRNs) with partial relay selection scheme. The secondary network, where NOMA is used, consists of a secondary base station (SBS) sending messages to two secondary users (SUs), that is, a near SU (SU1) and a far SU (SU2), by utilizing a dedicated relay selected from a set of FD decode-and-forward nodes. We obtain analytical expressions for the outage probabilities (OPs) of the SUs and then deduce the asymptotic OP expressions as well. Further, expressions are obtained for the optimal power allocation (OPA) coefficients at the SBS and at the relay that separately maximizes the throughput of the secondary network in NOMA-FDCRN. Furthermore, the jointly optimal power allocation (JOPA) coefficients that maximize the throughput are also determined. The analyses consider (i) imperfect successive interference cancelation conditions, (ii) the tolerable interference limit of the primary receiver, (iii) the secondary nodes' maximum transmit power values, (iv) interference generated by the primary transmitter, and (v) the residual self-interference (RSI) at the FD relay. It is shown that the proposed OPA coefficients at the SBS and the JOPA can mitigate the impact of RSI and significantly improve the OP and throughput performance. The numerical results show that the proposed JOPA approach provides 32% and 106% improvement of throughput compared to random power allocation (RPA) and equal power allocation (EPA) strategies, respectively. The OP of SU1 and SU2 reduce by 68% and 73%, respectively, under the proposed JOPA compared to RPA, while compared to EPA, the OP of SU1 and SU2 reduce by 96% and 97%, respectively.