Uplink Massive MIMO Detector Based on Vector Approximate Message Passing

In Massive MIMO systems because of increasing the number of antennas at the base station, designing uplink detector with perfect bit error rate and low computing costs becomes a key issue. Recently the AMP algorithm that was proposed for compressive sensing has been exploited in massive MIMO detection. Although it enjoys low complexity and perfect BER performance, it has some limitation like diverging at mild ill-conditioned channel matrices. VAMP is one of the algorithms that has been introduced to overcome the limitation of AMP. In this paper we investigate SVD form VAMP algorithm in massive MIMO uplink detection. According to the simulation results, the VAMP not only is robust against ill-conditioned matrices but also it unlike AMP shows perfect BER performance on both low and high SNRs. Meanwhile, after initial singular value decomposition, it has the same computational complexity with AMP.