A QoS-based joint user association and resource allocation scheme in ultra-dense networks

• Main Authors: Zhiwei Si, Gang Chuai, Weidong Gao, Jinxi Zhang, Xiangyu Chen, Kaisa Zhang
• Format: Article
• Language: English
• Published: SpringerOpen 2021-01-01
• Series: EURASIP Journal on Wireless Communications and Networking
• Subjects: Ultra-dense networks (UDNs); Quality of service (QoS); Virtual cells; PRB estimation method; Graph-based; DC programing
• Online Access: https://doi.org/10.1186/s13638-020-01882-3

Abstract Ultra-dense networks (UDNs) have become an important architecture for the fifth generation (5G) networks. A large number of small base stations (SBSs) are deployed to provide high-speed and seamless connections for users in the network. However, the advantage of increasing the system capacity brought by the dense distribution of SBSs comes at the cost of severe inter-cell interference. Although the user-centric virtual cell method has been proposed to solve the interference problem, some challenges have been encountered in practical applications. For example, inter-cell interference still exists to a certain extent, and the cell load may be imbalance. Hence, under the virtual cell architecture, we propose a quality of service (QoS)-based joint user association and resource allocation scheme in UDNs. In order to mitigate the interference, balance cell load and improve the system throughput, a non-convex NP-hard problem is formulated. To effectively solve this problem, we decouple the formulated problem into three sub-problems: user association, physical resource block (PRB) allocation and power allocation. First, we consider the QoS requirements of user equipment (UE) and perform user association based on the PRB estimation method. Then, based on the overlapped virtual cells constructed, we propose a graph-based PRB allocation scheme for reducing virtual inter-cell interference. Moreover, we solve power allocation sub-problem by using the difference of concave (DC) programing method. The simulation results show that our proposed scheme is superior to other schemes in terms of user rates, cell load and system throughput.