Lattice reduction aided linear detection with linear channel code using modulo operation for MIMO systems

Abstract

Linear signal process for MIMO system is prefer to non-linear one due to computational complexity. If transmitter can be fed back right-unitary-matrix by Singular value decomposition from receiver, channel-capacity can be achieved with linear processes in point to point wireless MIMO communication. However, if transmitter can not be fed back right-unitary-matrix from receiver (no-feedback), the optimal linear detector at receiver is minimum-mean-variance-estimator, of which capacity is far from channel-capacity. In first part of this paper, we propose the limited feedback system for linear process. The proposed scheme has near-channel-capacity, while the amount of feedback information is only integer matrix, which is much less than that of right-unitary-matrix via SVD. The key ideas of proposed scheme are Lattice reduction and modulo operation. Further, we minimize the amount of feedback information as binary matrix with multi-level/multi-stage encode and decode. In second part of this paper, we propose two-stage decoding with lattice reduction aided linear detection in multiuser uplink system. Since channel feedback is hard to be expected in uplink system, receiver side equalization is important. In communication applications, these processing techniques can be broadly categorized into two classes, linear and non-linear approaches, which have a trade-off between complexity and performance. Complexity of the proposed scheme for approximated posteriori probabilities of received symbols, can be reduced as much as that of Zero-forcing, approaching a near performance of Maximum likelihood detection. We consider powerful turbo channel codes and show that excellent performance at very high data rates. We compare our simulation results with Shannon capacity limits for ergodic multiple-antenna channel.