The Network Interference Management Research Based On Joint Base Staion Selection

Interference management is the foundation of a stable and an efficient network composed of multiple base stations(BSs) and multiple users. Each user is covered by multiple BSs and a sub-set of BSs are active in a collaborative way to achieve the channel interference mitigation or the congestion control. It’s necessary to balance the system performance improvement and the backhaul overhead, as well as the scheduling complexity when an actual network scenario is concerned.In this paper, we plan to select some coordinate BSs to perform joint reception in a single input and multiple output(SIMO) uplink network. To provide fairness among users, we consider a Signal to Interference plus Noise Ratio(SINR) based max-min fairness(MMF) problem by jointly optimizing the cooperating BSs, the power allocation and the beamformers. However; if we limit the size of the cooperating BSs,the MMF problem is a nonlinear mixed-integer program which is a challenging problem to solve. From the perspective of sparse optimization, we seek for an efficient approximating solution, then reformulate the problem by employing the uplinkdownlink duality transformation to eliminate trivial non-sparse solutions which is caused by the scaling ambiguity. Based on the idea above, the BS selection and the beamforming algorithms developed in this article, mainly focus on the following two power constraints:1. Total power constraint.2. Per-user power constraint.After the duality transformation, the scaling ambiguity disappeared and the optimization problem is convex. Generally, the convex optimization problem can be solved by CVX in a centralized way. However, the increasing dimensions of the matrix and vectors in the problem model leads to a large data exchange and a huge computing load, with the expanding of the users and the BSs, as well as the antennas. In such kind of situations we prefer the distributed method than the centralized one. However, if we want to design a distributed algorithm, we need to separate the coupled variables. So, a custom-made distributed solution is developed for the approximated problem by utilizing the alternating direction method of multipliers(ADMM) framework and the proposed algorithm is computationally efficient since the sub-problems in each step ofADMM can be solved in a separate manner.