NOMA기반 5세대 통신시스템에서 보안 처리율 극대화를 위한 전력할당 기법 연구

Abstract

The rapidly-evolving fifth-generation (5G) wireless system needs to satisfy requirements in terms of low latency, throughput, high energy efficiency, network capacity, and improved quality of service (QoS). Moreover, because of the rapid advancement of the Internet of Things (IoT), 5G should be capable of supporting massive connectivity of devices. In order to meet these requirements, improved technology strategies have been proposed, such as the ultra-dense network, millimeter wave communications, massive multiple-input multiple-output (MIMO), and nonorthogonal multiple access (NOMA). In this dissertation, NOMA has been studied since it is a promising technology in 5G networks owing to benefits like higher cell-edge throughput and efficient utilization of spectrum resources so that multiple users can share the same resource block (code/time/frequency). To further improve the throughput and extend the coverage of the 5G networks, in this thesis, we use cooperative communications including relay selection, cognitive radio (CR) and power allocation (PA) frameworks.

Although cooperative communications in NOMA system have shown the likelihood of meeting 5G requirements, wireless security is still a challenging topic for 5G networks. Particularly, the risk that eavesdroppers can obtain confidential information increases in wireless transmission environments. To tackle this issue, the commonly cryptographic encryption and decryption approaches have been proposed in the industry and academia, but these techniques utilizes complex decoding/encoding algorithms and encryption keys management that heavily relying on computing ability and resource consumption. Moreover, these complex algorithms can be broken since the eavesdroppers nowadays have access to high computational power capabilities. Thus, in contrast to the conventional security techniques, in this thesis, we propose to exploit the characteristics of the wireless channels through the application of physical layer security (PLS) in our systems, which is regardless of the computing ability of the communication equipment and can provide security even if the eavesdropper has strong computing capability. For the safe transmission designs, PLS utilizes the achievable secrecy rate as a fundamental metric to evaluate transmission effectiveness. Therefore, a suitable PA design to tackle eavesdroppers’ wiretaps is a very useful method to improve the PLS of NOMA systems, in which the secrecy rate of the networks is maximized based on solving the relay selection and the PA problems of the networks.

Firstly, a cooperative NOMA system with an energy-harvesting (EH) user, simultaneous wireless information and power transfer (SWIPT) is proposed. This wireless network benefits from the energy harvesting techniques to provide self-sustainability and the possibility of sharing energy among users. For instance, the energy harvested in the strong user of the NOMA system is utilized to transmit the weak user’s message to the intended user since the strong user performs SWIPT. Moreover, the EH user is a radio frequency (RF) energy harvesting device that harness the RF power obtained by receiving the RF signals of NOMA users through an antenna. In the system, the power allocation is optimized to minimize the total transmission power subject to constraints that allow satisfying the minimum QoS requirements of the users involved in the system. In this sense, a slack variable is required to convert this non-convex optimization problem into convex one. Then, the problem is transformed into a bilevel optimization problem where the inner problem optimizes the power allocation variables, while the outer problem optimizes the slack variable and power splitting ratio of the SWIPT system. Lagrange method is used to solve the inner optimization problem, which manages the power allocation variables and particle swarm optimization (PSO) algorithm is used to find a near close-optimal solution for the slack variable and power splitting ratio values. For comparison purposes, a solution for optimizing the PA variables is developed for conventional orthogonal multiple access (OMA) based on Langrage method and PSO. The numerical results show that NOMA with SWIPT outperforms the baseline approaches namely OMA and equal power splitting (EPS).

Secondly, a cooperative transmission via a relay is considered to enhance the coverage and transmission reliability of the NOMA system, which is affected by an eavesdropper. Thus, the secrecy rate is maximized to ensure the system's reliability by applying a low-complexity PA scheme based on the PSO algorithm, which can reach near-optimal performance. To find the solution, the proposed PSO-based method requires less than 60 iterations. Therefore, by applying this method, the complexity of the optimization problem decreases and it can convergence faster, compared to the optimal exhaustive search method. The performance of the proposed network is evaluated with and without the use of the cooperative relay and it is remarkable the difference gap between these results. Although the NOMA system is still feasible without using a relay node, the secrecy rate values greatly increase by employing a relay in the network since this relay aids the distant user to easily meet the QoS requirements. Furthermore, the simulation results show that the proposed NOMA with cooperative relaying scheme outperforms the baseline OMA with cooperative relaying scheme. This comparison approach is also studied and solved by PSO algorithm.

Finally, PSO-based PA and relay selection scheme is considered to improve the PLS by maximizing the secrecy sum rate (SSR) in a cooperative relaying CR NOMA system under the presence of multiple eavesdroppers’ wiretaps. Multiple relay is an interesting approach to benefits from multiuser diversity, expand the coverage of the network, and so on. Particularly, in this scenario the relay selection scheme aids the secondary users to properly receive their messages and improve their reception reliability to prevent multiple eavesdroppers’ wiretaps. However, the conventional optimization techniques make use of derivatives and gradients are mainly focus on the yield of a single intended user to obtain closed-form expressions results. Hence, this sort of methods can hardly be applied to the proposed system where more than two users and multiple eavesdroppers are involved in the CR-NOMA network which increase the complexity of the constraints in this optimization problem. Thus, a bilevel optimization problem is formulated and solved by the proposed low computational–complexity PSO-based algorithm where the inner step optimizes the transmit power variables, while the outer step performs the relay selection scheme. Numerical results show that the proposed CR-NOMA system greatly outperforms the traditional CR-OMA scheme, in terms of SSR. Moreover, the optimal exhaustive search is used as a baseline method to validate the outputs obtained by the proposed PSO algorithm. The results acquired by PSO can reach near optimal performance since they are very close to those results obtained by the optimal exhaustive search.