Channel Estimation for MU-MIMO Systems with Some Inactive Users

Almost all existing channel estimation schemes, for multi-user multiple-input multiple-output (MU-MIMO) systems, were designed based on a fixed and known number of active users. However, in random access MU-MIMO systems, some users may be inactive and the estimation techniques designed with the assumption that all users are available may lead to performance losses. In this paper, we consider the popular least squares (LS) and linear minimum mean square error (LMMSE) estimations approaches. For such techniques, when the number of users varies, the channel estimation filter coefficients must be re-derived. The fundamental problem of interest here is to reduce the computational complexity of the LS and LMMSE channel estimation methods when some users are inactive. For each techniques, we propose an estimation approach with a low complexity and without performance loss. The proposed algorithms avoid the direct computation of matrix inverses required by the LS and LMMSE methods. Moreover, the mean square error (MSE) losses, due to the use of the LS and LMMSE estimator intended for the scenario without inactive users, are evaluated and compared with Monte-Carlo simulations results.