An Efficient Hierarchical User Grouping and Successive Group Search Approximation Based Beam-Forming Method for IRS-Aided Millimeter-Wave Networks with NOMA

Abstract

The combination of non-orthogonal multiple access (NOMA) and intelligent reflective surface (IRS) is an efficient solution to significantly increase the energy efficiency of the wireless communication system. This paper considers a downlink IRS-based Millimeter wave (mmWave) massive Multiple-Input Multiple-Output (MIMO)-NOMA system. In this work, a two-layer hierarchical AGglomerative NESting- DIvisie ANAlysis Clustering Algorithm (AGNES-DIANA) user grouping (2L-HAD-UG) is proposed to group users. In the proposed 2L-HAD-UG, the first layer will use the AGNES algorithm to efficiently group the users into different clusters. After the first level of clustering, some user groups are larger than others, and perhaps some users with weakly correlated channels are assigned to the same groups. To address these concerns, the large user groups are divided into several smaller groups, and each user whose channels are weakly correlated is isolated as a separate group. In the second layer, the DIANA Hierarchical Clustering is used to divide the larger clusters based on the channel correlation value. After user grouping, a new joint active and passive beam-forming design problem is formulated to maximize the achievable rate of each user in each cluster under the Quality of Service (QoS) requirements of other users, the conditions of Successive Interference Cancellation (SIC) decoding rate, the constraints of IRS reflection components and the transmission power restrictions. The formulated optimization problem is solved by proposing a new Successive Chaotic Group Search Approximation (SCGSA) algorithm. With the proposed massive mmWave MIMO-NOMA system, the spectral efficiency, energy efficiency, and sum rate are achieved to 18.56 bits per second (bps)/Hertz (Hz), 57.56 bps/Hz/Watt (W) and 21.58 bps/Hz, respectively.