Hybrid massive MIMO architecture with SWIPT in NOMA systems

Abstract

It is becoming apparent that radio frequency energy harvesting (EH) presents a vital potential for improving the energy efficiency of existing and future networks. EH has also emerged as a viable option, allowing for the re-use of a portion of the transmitted energy through an energy cycle. Moreover, the enormous need for large data speeds and increased traffic in wireless networks has prompted research into the millimeter wave (mm-Wave) spectrum for future (5G) communication systems. Massive antenna arrays are used for this high frequency. However, high implementation costs and signal processing complexity make completely digital (homogeneous) beamforming systems expensive and inefficient, while employing purely analog beamforming reduces system performance owing to hardware limits. To address the aforementioned issues, hybrid beamforming has been proposed. Finding an appropriate design remains a significant difficulty for the researcher. In this thesis, we first formulate and present a new method based on the long recurrence enlarged conjugate gradient (LRE-CG) algorithm for iteratively realizing the MMSE detection method while avoiding the matrix inversion complexities in massive MIMO systems. We demonstrate its superior performance to present competing methods. Then we tailor its application to massive MIMO systems and mmWave massive MIMO systems with NOMA System. It has been verified by the simulation results that the presented method outperforms the conventional methods in the literature including the Neumann series approximation-based approach and Gauss-Siedel iterative (GS) methods for massive MIMO systems. The proposed algorithm succeeds to attain the near-optimal performance of a typical MMSE algorithm with a minimal level of iterations. Secondly, we deal with the issue of power consumption in millimeter-wave (mmWave) huge MIMO systems caused by mixed-signal components like analog-to-digital converters. In order to increase the efficiency of the spectrum, we use non-orthogonal multiple access (NOMA) in mmWave large MIMO systems. Massive multi-input multi-output (MIMO) systems operating at millimeter waves will make advantage of the simultaneous wireless transmission of information and power (SWIPT). In the NOMA scenario, where inter-user interference may be utilized for energy harvesting, SWIPT is particularly helpful in extending the battery life of mobile users (MUs) and improving the system energy efficiency (EE) (EH). To begin with, we developed a user grouping algorithm based on affinity propagation clustering, which groups user equipment (UE) based on channel correlation and distance. Following that, we designed the analog RF precoder for all beams using the selected user grouping, then we designed a low-dimensional digital baseband precoder to achieve a maximum sum-rate as well as minimize inter-beam interference for the system. Afterward, we formed a joint power allocating and power splitting optimization problem. Given the existence of linked variables and inter-user interference, the studied non-convex optimization issue is challenging to handle. In order to address this issue, a decoupled approach is used, in which the power allocation and power splitting are treated as independent issues, and the Lagrangian duality technique is used to solve these sub-problems. The simulation findings validate the efficacy of the proposed technique and show that it is near-optimal and has superior spectrum and energy efficiency compared to the previous designs and the existing SWIPT-enabled mmWave MIMO-NOMA system.

Date	13 Dec 2022
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Michel Kadoch (Supervisor)