A Study on Non-Orthogonal Multiple Access for Data-Centric Machine-to-Machine Wireless Networks

• Main Authors: Guan-Quan Chen, 陳冠全
• Other Authors: Hung-Yun Hsieh
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/3jm6g5

碩士 === 國立臺灣大學 === 電信工程學研究所 === 106 === Due to the rising number of machines, NOMA has recently been considered as a key technique for 5G communication in the future. In this thesis, we formulate a general modeling and analytical framework for Machine-to-Machine (M2M) wireless networks by considering the non-orthogonal multiple access (NOMA). For most of the M2M wireless networks, seldom of them focus on the correlation which is the natural data characteristic. Traditionally, most of them designs the scheduling in NOMA with consideration of the maximization sum throughput. Thus, we utilize the data-centric concept into our target scenario involved in two-tier data gathering networks. First of all, we implement NOMA in tier-1 and tier-2 is OMA. Compared to throughput maximization, we propose the “waiting time” scheduling method which considers not only minimization of transmission resource but also minimization of waiting time caused by the property of NOMA. In our proposed algorithm, once the cluster structure is given we calculate each pair of machines with the transmission resource and waiting time. Compared to both tiers with OMA, our proposed scheduling method can save 15.77% transmission resource. However, throughput maximization exceeds 7.42% compared to OMA. Secondly, we extends NOMA to tier-2. We restrict the scheduling method in tier-2 in order to investigate the influence with different scheduling methods in tier-1. Since the complexity of the target problem is NP-Hard, we refer to a meta-heuristic algorithm based on simulated annealing method in order to solve the optimization problem. Compared to both tiers with OMA, our proposed scheduling method can save 34.44% transmission resource and yet throughput maximization can only save 6.32%. Thus, considering data-centric factor in designing scheduling methods can bring in the higher performance and less transmission resource for next generation communication systems.