MIMO LMMSE Transceiver Design with Imperfect CSI at Both Ends

Abstract

Abstract-This paper presents a new result on minimum total mean squared error (MSE) joint transceiver design for multiple-input multiple-output (MIMO) systems, with imperfect channel state information (CSI) at both ends and subject to a total power constraint. The channel knowledge here is the channel mean and transmit correlation information. The joint design is formulated into an optimization problem, to which the closed-form optimum solution is found. The optimum precoder consists of a linear filter, a matrix which collects effective channel eigenmodes, and a diagonal power allocation matrix. The linear filter balances the suppression of channel noise and the additional noise induced by channel estimation error. Given the precoder, the optimum decoder is simply a linear minimum MSE (LMMSE) data estimator. The performance degradations due to imperfect channel estimation and/or transmit correlation are demonstrated by simulation results. The relation between the minimum total MSE design and the maximum mutual information design is determined as well, under the above mentioned imperfect CSI.

Joint optimization for multiuser MIMO uplink systems with imperfect CSI

“…As in [2][5] [7], we cannot show that the iterative algorithm in Table I is guaranteed to achieve the globally optimum solution (except when K = 1 [15]), despite the fact that the global minimum exists. This is because the objective function in (1) is not convex in…”

MIMO Minimum Total MSE Transceiver Design With Imperfect CSI at Both Ends

“…Correspondingly, the KKT conditions associated with (9) can be derived, as given by (24)- (27) (24) (25) (26) (27) It can be shown that Lemma 1 still holds here, i.e., for any solution which satisfies the KKT conditions (24)- (27). Define (28) Consider the following eigenvalue decomposition: (29) where the subscript "g" means the general case, the matrices , , , and are similarly defined as those in (15), and denotes the rank of the matrix in (29), i.e., . The diagonal entries of are arranged in decreasing order.…”

“…However, the scalars , and need to be determined numerically, which can be a difficult task. Alternatively, (9) can be solved using an iterative algorithm developed from the KKT conditions [12], [13], [28], which is given in Table I. This algorithm converges according to [13].…”

Transceiver Designs for MIMO Relaying Systems with Imperfect Channel State Information

“…ese existing schemes, however, do not consider the scenario where there is ICSI. In the presence of ICSI, robust designs have been studied in [16], but this study does not consider the MSE decomposition and relaxation property [3].…”