Joint User Selection and Feedback Bit Allocation Based on Sparsity Constraint in MIMO Virtual Cellular Networks

Abstract

In this paper, we jointly consider the user selection and feedback design problems in a virtual cellular network (VCN), where multiple base stations (BSs) share a user set. In many practical systems, the uplink feedback channel is generally shared by multiple users. Thus, the feedback budget allocated to unselected users not only wastes the feedback resources, but harms the system throughput by decreasing the available feedback budget for the selected users. We optimize both the user selection and feedback bit allocation based on long-term average channel information of the users. We first analyze the effects of the quantization error on the average achievable rate of the VCN system. Next, we propose a user selection and feedback bit allocation protocol under each BS's sum feedback rate constraint as well as the sparsity constraint on all users' feedback sizes. We show that the joint optimization problem can be decoupled into several NP-hard subproblems, one for each BS. We describe the brute-force searching algorithm for the optimal solution, and propose an efficient algorithm with significantly reduced computational complexity by relaxing the sparsity constraint on the feedback sizes. As a result, only the selected users exploit the uplink feedback budget, and the system performance is improved.