Efficient Beamforming and Power Allocation in MillimeterWave Non Orthogonal Multiple Access

Abstract

Non-orthogonal multiple access (NOMA) has been proposed as a viable multiple access technique to meet the demanding requirements of fifth-generation (5G) and beyond wireless networks. NOMA is different from traditional orthogonal multiple access (OMA) because it sends signals to multiple users newlinein the same resource block (RB) at the same time and frequency using power-domain superposition coding (SC) at the transmitter. The receiver performs successive interference cancellation (SIC) to decode the data intended for that specific user. Therefore, NOMA has the potential capabilities to newlineserve a large number of devices while significantly improving spectrum efficiency (SE) compared to the conventional multiple access (MA) techniques, which support massive connectivity of Internet-of-Things newline(IoT) networks.NOMA has also been integrated with multiple antenna techniques with beamforming and millimeter newlinewave (mmWave) communication to increase the data rate. In this thesis, we consider two cases of multi-user NOMA: non-clustered NOMA and clustered NOMA. In the former case, users with different newlineangles of departure (AoD) are considered, where the beamforming vector is designed to generate each beam targeting each user. In clustered NOMA, users with similar AoDs are grouped into clusters, and the beamforming vector is designed to generate each beam targeting to each cluster.This thesis aims to develop algorithms for designing a low-complex beamforming vector and allocating power to users in both cases. In our first work, we consider an analog beamforming mmWave non-clustered NOMA system. The reason behind choosing the analog beamforming is that the NOMA system can even support multiple users using a single RF chain (RFC) with the analog beamforming architecture, whereas OMA can support only one user with the single RFC. In the existing research, design of the digital beamforming vector and power requirement of each user has been computed using the SDR (semi-definite relaxation) algorithm with randomization,...