Summary: | 碩士 === 國立臺灣大學 === 電信工程學研究所 === 107 === This thesis investigates the system-level performance of non-orthogonal multiple access (NOMA) system. The main concept of NOMA is to transmit composition of multi-user signals in a beam at the transmitter and decode with successive interference cancellation (SIC) at the receiver, which is a promising downlink multiple access scheme for further LTE enhancement. Hybrid automatic repeat request (HARQ) is an essential part of link adaptation, and it is also playing important roles in modern wireless communication system. System with HARQ has channel coding gain by re-transmitting data packets which decoded unsuccessfully. Thus, reliability of system can be improved by reducing outage probability of users. In order to reduce complexity and computing resources before transmission, we simulate the environment which implemented with iterative subband allocation scheduling for user pairs and power allocation coefficient to research the effect of scheduling algorithms. In order to maximize total throughput and fairness, the proportionally fair (PF) scheduling model is adopted, and optimization analysis methods are introduced to evaluate downlink NOMA system performance. However, implementation of HARQ is more challenging in NOMA system with subband scheduling scenario than in traditional OMA widebannd system, because of constraints for scheduling users and power within multiple subbands in the same time. We propose a heuristic algorithm based on cross entropy method to solve the optimization problem about user pair and power allocation for subband PF scheduling, and implement with HARQ in order to meet outage probability constraint and discuss about its benefits and trade-off. Compared with other sub- band allocation algorithms, Simulation results show that proposed algorithm have about 50% performance gain of baseline traditional method and sequential greedy approach.
|