Subset Vector Perturbation Pre-Coding for MU-MIMO Downlink Systems

Abstract

Multi-user multiple input multiple output (MU-MIMO) is a basic technique which has been widely investigated for its ability to increase system capacity. However, the inter-user interference (IUI) is one of the obstacles to obtain system capacity. With the help of pre-coding technique, the IUI can be canceled or reduced at the base station (BS). Vector perturbation (VP) is one of the most celebrated pre-coding schemes that can approach system capacity. By perturbing users’ data vector with a certain integer vector offset at the BS, the VP scheme can offer the users a higher effective signal-to-noise ratio. To reach the theory bound of VP, the BS has to select the optimal integer vector from all candidate integer vectors globally. This exhaustive search has been proved to be NP-hard, which is hard to be implemented. In this paper, we analyze several state-of-the-art modified low-complexity VP schemes and propose the subset VP (SVP) scheme to reduce the complexity of the VP. The basic principle of the SVP is that some vectors are selected as the optimal vector with high probability, whereas the others are scarcely selected. So, we try to form a subset consists of vectors with a relatively higher probability. First, we calculate the probability of all candidate vectors which have been selected as the optimal vector. Second, we propose a method to divide all vectors into several subgroups. Vectors in the same subgroup have a similar probability, whereas vectors in different subgroups have distinguished probability. Third, we build a subset that is comprised of several subgroups in which vectors have a relatively higher probability. Simulation results show that we get almost the same bit-error rate performance with a significant complexity reduction.