Performance Analysis of Dynamic Re-Clustering and Resource Allocation in Ultra Dense Network

Abstract

Coordinated multi-point (CoMP) is a key technology to mitigate inter-cell interference, which can significantly improve the cell edge performance and system throughput. In addition, the choice of the cells within cluster will directly affect the effect of CoMP, and especially, it may cause the increase of the number of dissatisfied users in the system. To minimize dissatisfied users in the ultra-dense network (UDN) application scenario, based on the control-data separation architecture model, a clustering algorithm is studied to improve the SINR and throughout. Then, load balancing is developed and the overall percentage of unsatisfied users are reduced obviously. Furthermore, a dynamic resource allocation scheme is proposed by defining a factor $\alpha $ and applied to optimize the performance of system, which is compared with load balancing under the conditions of a different clustered size and density of users. The experiments prove that the performance of load balancing is more sensitive to the change of clustered size than resource allocation, and the load balancing has overwhelming advantage compared with resource allocation in the dense deployment scenario. However, when the density of users becomes sparser, resource allocation performs better, which gives important meanings for future UDN.