Power and Discrete Rate Adaptation in Wideband NOMA in Frequency-Selective Channels

Power-domain non-orthogonal multiple access (NOMA) superimposes signals of multiple users and transmits them simultaneously. To be implemented in 5G and beyond orthogonal frequency division multiplexing systems, it must adhere to the constraint imposed by the standard that the same modulation and coding scheme (MCS) and power must be used across all physical resource blocks (PRBs) assigned to each user. However, the PRBs have different gains in wideband channels and the MCSs must belong to a discrete, pre-specified set. We propose a method that uses the exponential effective signal-to-noise ratio mapping (EESM) to systematically determine whether a feasible power allocation exists for a given choice of MCSs, and to find the MCSs that maximize the weighted sum rate for multiple user NOMA. We then propose a novel power-normalized EESM with backtracking (PB) method. It develops and exploits explicit analytical criteria to check for feasibility. We prove that it is a relaxation of the original problem under various conditions and is exact for narrowband channels. The average weighted sum rate of PB is indistinguishable from that of the EESM-used method despite its lower complexity. It is higher than that of wideband orthogonal multiple access, which is currently employed by 5G.