Message Passing Based Joint Active User and Data Detection for mMTC Systems

This paper investigates the issue of active user and their data detection in the uplink of massive machine type communication (mMTC) systems. By leveraging the sparsity of the active user distribution and the characteristic of multiple time slot synchronous transmission, the problem of active user and their data detection is formulated as an equivalent sparse signal recovery problem represented by a multiple measurement vector (MMV) model. To tackle this problem, a low-complexity method based on the message passing algorithm for MMV sparse Bayesian learning (MSBL), named the BP-MF-MSBL algorithm, is proposed in this work. We introduce some hard constraint factors in BP-MF-MSBL to extend the factor graph, which can effectively integrate the belief propagation (BP) and mean field (MF) message passing criteria to optimize the performance. Furthermore, some messages are approximated to alleviate the computational complexity of the algorithm. The simulation results demonstrate that BP-MF-MSBL outperforms the traditional MMV-OMP in terms of active user detection rate and signal reconstruction performance.