Energy-Efficient Multi-Stream Sparse Regression RaptorQ Codes for Unsourced Random Access

In the paper, we propose a novel RaptorQ-based unsourced random access (URA) scheme that integrates RaptorQ codes and sparse regression codes (SPARCs) to design access schemes tailored for massive machine-type communication scenarios. The proposed scheme eliminates the need for conventional outer tree codes by employing a dynamic preamble sequence, which serves as a temporary user identifier and facilitates the stitching of data segments across sub-slots. Furthermore, a dynamic approximate message passing (AMP) algorithm is utilized to jointly recover multiple data substreams transmitted concurrently by each user within a given sub-slot. In addition, the RaptorQ-based outer code, functioning as an erasure code, is capable of simultaneously correcting both missed detection and false alarm errors in the decoded sub-segments. Simulation results demonstrate that the proposed scheme achieves improved energy efficiency when operating under moderate and large load conditions. Moreover, it exhibits a significant performance gain over existing URA schemes in additive white Gaussian noise channels.