Bipartite network based multi-cell clustering scheme in randomly located CoMP systems

Coordinated multiple point transmission (CoMP) technology becomes popular due to its potential for improving the performance of cell-edge UEs who undergo strong inter-cell interference (ICI). To fully exploit the potential CoMP gain, the appropriate method to converge cells into cluster is a prerequisite. However, the majority of existing research are based on theoretical hexagonal cell grid model solely. In this paper, a CoMP multi-cell clustering scheme which considers randomly located cellular network is proposed. In randomly located cells model, the layout of cells and UEs are generated via a Poisson Point Process (PPP). The proposed clustering scheme takes full account of both ICI suffered by UEs and cells distribution, by mapping the CoMP network onto a bipartite network, in which two types of nodes represent cells and UEs respectively, while edges are quantified to denote the channel gain. Meanwhile a projection method is introduced to project the bipartite network onto a Euclidean space where distance between UEs and cells is easily calculated. Finally the efficient K-mean method is utilized to extract nearby UEs and cells out of the space and form the clusters. The proposed scheme hardly incurs extra overheads since only UE´s large scale fading information is required. The simulation results demonstrate a considerable throughput gain attained by UEs over comparing algorithm.