A fully distributed approach for joint user association and RRH clustering in cloud radio access networks

• Published in: Computer networks (Amsterdam, Netherlands : 1999) Netherlands : 1999), 2020-12, Vol.182, p.107445, Article 107445
• Main Authors: Taleb, Hussein, Khawam, Kinda, Lahoud, Samer, Helou, Melhem El, Martin, Steven
• Format: Article
• Language: English
• Subjects: Algorithms; Call Detail Records (CDR); Cellular communication; Cloud radio access networks (C-RAN); Clustering; Computer Science; Linear programming; Load fluctuation; Mixed integer; Nonlinear programming; Optimization; Power consumption; Radio; User association (UA); Wireless networks
• ISSN: 1389-1286; 1872-7069
• DOI: 10.1016/j.comnet.2020.107445

Cloud radio access network (C-RAN) is a new centralized architecture to meet the exponential growing of demand of mobile traffic in 5G cellular wireless networks. However, C-RAN requires an efficient mechanism for the joint user association and the Remote Radio Head (RRH) clustering to improve network performance. In this paper, we investigate the problem of joint user association (UA) and RRH clustering (RC) in C-RAN. Our objective is to maximize the network utility function incurred by both network power consumption and total user throughput for both streaming and elastic traffic. The formulation of the joint optimization problem is a mixed-integer non-linear programming problem (MINLP), which is NP-hard and usually has no feasible solution. To solve it, we propose to decouple the joint problem into two sub-optimization problems: the user association (UA) sub-problem and the RRH clustering (RC) sub-problem. These two sub-problems are sequentially and iteratively solved until convergence is reached. Leveraging on the information delivered by the Call Detail Records (CDR), simulation results reveal the effectiveness of our heuristic solution for the RC sub-problem in enhancing network utility and adapting to the traffic load variation for both elastic and streaming traffic. It outperforms the performance of the state-of-the-art algorithms for RRH clustering solutions, including no-clustering and grand coalition methods. Moreover, the results show that our approach for the UA sub-problem provides close performance to the optimal UA sub-problem.