Decoupling or Learning: Joint Power Splitting and Allocation in MC-NOMA With SWIPT

Abstract

Non-orthogonal multiple access (NOMA) is one of the most significant technologies to meet the demand of high spectral efficiency (SE) in the fifth generation (5G) cellular networks. The utilization of simultaneous wireless information and power transfer (SWIPT) contributes to prolonging the battery life of the mobile users (MUs) and enhancing the system energy efficiency (EE), especially in the NOMA scenario where the multi-user interference can be reused for energy harvesting (EH). In this paper, we study the achievable data rate maximization problem for the downlink multi-carrier NOMA (MC-NOMA) network with power splitting (PS)-based SWIPT, in which power allocation and PS control are jointly optimized with the limitation of available power budget as well as the requirement for EH. The considered non-convex optimization problem is arduous to tackle, resulting from the presence of the coupled variables and the multi-user interference. To cope with the problem, a decoupled approach is developed, in which the power allocation and PS control are separated and the corresponding sub-problems are respectively solved through Lagrangian duality method. Furthermore, an alternative approach based on deep learning is proposed, which is capable of effectively obtaining the approximate optimal solution according to the empirical data. Simulation results confirm the effectiveness of the proposed schemes, and demonstrate the superiority of the combination of PS-based SWIPT with MC-NOMA over SWIPT-aided single-carrier NOMA (SC-NOMA) and SWIPT-aided orthogonal multiple access (OMA).Index Terms-Multi-carrier non-orthogonal multiple access (MC-NOMA), simultaneous wireless information and power transfer (SWIPT), deep learning.

Outage and throughput analysis of power-beacon assisted nonlinear energy harvesting NOMA multi-user relay system over Nakagami-m fading channels

“…As observed from Fig. 5, the OP remarkably reduces as gets higher, then sharply increases to 1 These values are based on the IEEE 802.11a/g standards. 1 when is larger than a certain value, e.g., = 0.8 for th = 15 dB.…”

“…The average channel gains are set as Ω 1 = Ω 2 = 1, Ω 3 = Ω ,1 = 1, Ω ,2 = 2, and Ω ,3 = 4. 1 The power allocation coefficient for D is = ( − + 1)∕ where is selected so that ∑ =1 = 1. Fig.…”

“…Electronic devices may not be powered by electricity mesh in some circumstances. Therefore, using radio frequency (RF) energy harvesting (EH) to support the operations of low-power communication devices in small-cell and sensor networks is considered as a promising research direction [1,2]. On the other hand, due to the increase in the amount of data traffic, the number of wireless devices and communications in wireless networks, there is increasing spectrum demand.…”

Antenna selection in nonorthogonal multiple access multiple‐input multiple‐output systems aided by machine learning

“…Applied ML techniques in wireless communications has grown exponentially since their recent inception. ML is a powerful tool to deal with complex not fully and well-modeled problems, specifically in telecommunications systems, due the achieved accuracy-complexity trade-off, including problems of resource management (RM) in wireless networks, [17][18][19] selective channel estimation, 20 information detection, [21][22][23] and so on.…”

User selection and dynamic power allocation in the SWIPT-NOMA relay system

“…AI techniques can extract valuable information from data to learn and support different functions for optimization, prediction, and decision-making in mobile edge computing, mobility prediction, optimal handover solutions, and spectrum management [17]. Deep reinforcement learning (DRL) can solve real-time and dynamic decision-making problems for power allocation [18][19][20]. Reference [18] proposed a deep Q network (DQN) for each MT to obtain the optimal power allocation scheme.…”