Research On Ultra-dense Networks Via Graph Theory

Ultra-Dense Network(UDN), as one of the key technology in next generation wireless communications, is widely accepted as a promising enabling technology to realize high energy and spectrum efficiency by deploying a lot of low-power small cell base stations(SBSs). However, the massive deployment of small cell base stations and the overlapping of coverage will inevitably face critical inter-cell interference. Meanwhile, cell association based on the strongest reference signal receiving power(RSRP) at each user may cause a load unbalancing. These issues should be carefully and properly covered before achieving the promised performance gains. Hence, this work aims at studying ultra-dense network for the next generation wireless communication and the main contributions are as follows:Firstly, we investigate the performance of small cell networks with different small cell base station(SBS) densities and different transmission powers via using a system-level simulation platform and some real traffic data provided by China Mobile Research Institute. Simulation results numerically demonstrate that the load unbalancing and critical inter-cell interference that must be faced in practical communication system may limit the system performance in ultra-dense network.Secondly, we exploit factor graphs to design a distributed cell association and resource allocation algorithm for ultra-dense networks. The proposed belief propagation based distributed algorithm can decompose the joint optimization problem into a series of low complexity subproblems and the original optimization problem can be efficiently solved via solving these subproblems separately. In addition, based on the proposed algorithm the amounts of exchanging information overhead between the resulting subproblems are also reduced. And the proposed algorithm is simulated and proved to have better performance compared with the existed algorithm. Meanwhile, these algorithms can be modified or extended to any 0-1 problems in cellular networks.Finally, we propose a graph-based low complexity dynamic clustering algorithm. The key idea behind the proposed algorithm is that dividing the whole network into a number of clusters under size constraint and the maximum intra-cluster interference and minimum inter-cluster interference. The logic is maximum intra-cluster can be effectively controlled by the coordination within each cluster and can increase the network throughput. Meanwhile, graph-based algorithm is exploited to further reduce implementation complexity. Simulation results numerically demonstrate that the proposed low complexity algorithm has almost the same performance compared to the existing high performance algorithm but the complexity is much lower.