Approximate Message Passing for Sparse Large MIMO Systems with Prior Information

Abstract

We consider large MIMO systems where not all transmit antennas are active in every timeslot resulting in a sparse transmit vector. A typical scenario is a wireless sensor network with a single aggregation node. Compressed sensing algorithms allow to reconstruct the transmitted signal at the receiver with relatively few receive antennas using the sparsity of the system. Bayesian approximate message passing exploits the sparsity and the discrete nature of the communication system with suitable priors. We expand the algorithm by an individual prior component for each element, e.g. information from a decoder, and survey its influence on the recovery performance.