Dynamic user clustering with hybrid beamforming in millimeter wave non-orthogonal multiple access (NOMA) and power allocation using teaching learning based optimization (TLBO)

Non-orthogonal multiple access (NOMA) is a most promising candidate in future wireless networks to improve user fairness and spectral / energy efficiency in multiuser scenarios. In this work, we investigate a clustered multi-user millimeter wave (mmWave) downlink NOMA system with a hybrid beamforming architecture. In most existing research with dynamic user clustering, the number of users in each cluster varies from one to any number. A single user in a cluster corresponds to orthogonal multiple access (OMA) case, which is not beneficial. At the same time, clusters with more than one user employ the NOMA technique. However, the main limitation of NOMA with a large number of users is the successive interference cancellation (SIC) decoding complexity which increases with the number of users. Moreover, this also causes a delay in decoding the data at the users. This problem is addressed in our proposed algorithm, where each cluster has a minimum of two users to a maximum of a fixed number of users. Together with the dynamic clustering, a hybrid beamforming vector is also designed. Then, a single-step power allocation algorithm using teaching-learning based optimization (TLBO) to maximize sum spectral efficiency is proposed, which is less complex than the existing suboptimal methods.