Optimized Unsourced Random Access Schemes With Sparse-Correction-Based Approximate Message Passing for Massive MIMO Systems

In a massive machine-type communication (mMTC) scenario, there are large number of devices that may establish links to receivers, which causes a great burden to the system for signaling overhead, thereby reducing the communication efficiency. One potential approach involves refraining from assigning signaling overhead to users, with active users opportunistically transmitting data messages within time slots. The receiver only needs to recover these independent data messages without identifying the source of these messages, named unsourced random access (URA). In this paper, suitable URA schemes are proposed for cellular massive multiple-input multiple-output (MIMO) and cell-free massive MIMO systems. In the cellular scenario, the system completes the transmission and estimation of the number of active users and the corresponding channel large-scale fading coefficients (LSFCs) in stage one. Utilizing the channel information obtained in stage one, the receiver in stage two applies the sparse-correction-based bilinear generalized approximate message passing (SCB-BiG-AMP) algorithm proposed in this paper to restore the original data sequences sent by active users. In a cell-free scenario, active users won't transmit LSFC information similar to that in stage one of cellular scenario for the properties of massive distributed antennas. Instead, the central limit theorem (CLT) is used to estimate the average channel LSFC of all active users, which is substituted into the SCB-BiG-AMP algorithm as the equivalent channel variance. Then, the original data sequences sent by active users can be restored. According to the simulation results, the proposed URA communication scheme can achieve good bit error performance. Moreover, as a result of the small uplink user-AP distance, the system performance in the cell-free scenario is much better than that in the cellular scenario.