An Improved Successive Filter-Based Dropping Algorithm for Massive MIMO With Max-Min Power Control

Abstract

In line-of-sight massive MIMO, there are use cases where the channel vectors of some users become highly correlated. Highly correlated users lead to a large reduction in the sum-rate of linear and nonlinear precoders with max-min power control. To alleviate the loss in the sum-rate, some users can be dropped and rescheduled. The optimal dropping strategy can be found by an exhaustive search. In this letter, a successive filter-based dropping algorithm (SFDA) is proposed, which improves upon the existing dropping algorithms in the literature. At each step, the user with the highest filter norm is dropped. By comparing the sum-rate of all the steps, the best set of dropped users is found. In contrast to previous threshold-based algorithms in the literature, SFDA does not require a predefined threshold for the spatial correlation of users. Compared to an exhaustive search, the complexity of SFDA is reduced. Simulations results show when a 100 antennas base station serves 10 users, SFDA improves the 5th percentile sum-rate compared to previous algorithms in the literature up to 6 bits/channel use.