A Signal-Space Aligned Network Coding Approach to Distributed MIMO

This paper studies an uplink distributed MIMO (DMIMO) system that consists of K users and K' distributed base stations (BSs), where the BSs are connected to a central unit (CU) via independent rate-constrained backhaul (BH) links. We propose anew signal-space aligned network coding scheme. First, a network coding generator matrix is selected subject to certain structural properties. Next, distributed linear precoding is employed by the users to create aligned signal-spaces at the BSs, according to the pattern determined by the network coding generator matrix. For each aligned signal-space at a BS, physical-layer network coding is utilized to compute the corresponding network-coded (NC) messages, where the actual number of NC messages forwarded to the CU is determined by the BH rate-constraint. We derive an achievable rate of the proposed scheme based on the existence of the NC generator matrix and signal-space alignment precoding matrices. For DMIMO with two and three BSs, the achievable rates and degrees of freedom (DoF) are evaluated and shown to outperform existing schemes. For example, for DMIMO with two BSs where each user and BS have N and N' antennas, respectively, the proposed scheme achieves a DoF of 2 min (N, N') - 1, if the BH capacity scales like (2 min (N, N') - 1) log SNR. This leads to greater DoF compared to that utilizes the strategy for interference channel, whose DoF is min (N, N'). Numerical results demonstrate the performance advantage of the proposed scheme.