A comparison of serial and parallel LLR updates for LDPC coded massive MIMO detection with belief propagation

Recently, very large multiple-input multiple-output (MIMO) systems have been actively studied for 5G mobile radios. Belief propagation based detection is known as a technique reducing the computational complexity for uplink signal detection. When we consider an LDPC coded MIMO system, it is possible to combine both factor graphs of MIMO detection and LDPC decoding. In the paper, two types: bipartite and tripartite graphs are compared. The former one updates the LLR values on MIMO detection and parity checking simultaneously. On the other hand, the latter one updates alternatively. The simulation results show that the tripartite graph provides faster convergence with the same BER performance.