Deep embedded based energy efficient user grouping and Kookaburra Goshawk optimization for optimal power allocation in Terahertz MIMO-NOMA systems

Abstract

In recent years, terahertz (THz) communications have gained significant attention due to their potential for supporting ultra-high data rates in future wireless systems, particularly in multiple-input multiple-output non-orthogonal multiple access (MIMO-NOMA). However, efficient power allocation and interference management remain critical challenges. In this work, a power allocation approach is devised for THz MIMO-NOMA systems based on fuzzy integrated user pairing. First, the system model is considered and the devised work is based on grouping the users and hybrid precoding. By considering three constraints including position, signal to interference plus noise ratio (SINR), and initial power the user grouping procedure is done. Additionally, the grouping of users is based on the deep-embedded fuzzy clustering, where the measure of distance is modified by applying the above-mentioned three parameters. Next, the hybrid precoding of the THz MIMO-NOMA scheme is done. Later, the power allocation is done based on the user grouping, which is performed concerning the established Kookaburra goshawk optimization (KGO) by employing energy efficiency (EE), a utility function. Here, the KGO is established by unifying the Kookaburra optimization algorithm (KOA) and northern goshawk optimization (NGO). The KGO for power allocation reached the maximal achievable rate, EE, sum rate, utility, and simulation time of 0.958, 0.909, 0.858, 0.899, and 3.579, respectively, for 400 users.