Enhanced ODMA with Pattern Collision Resolution and Parameter Design for Unsourced Multiple Access

An enhanced on-off division multiple access (ODMA) transmission scheme is introduced for unsourced multiple access networks. Building upon the foundational ODMA transmission scheme, we have implemented further refinements to the original joint on-off pattern and data detection algorithm. Specifically, we propose a pattern collision resolution technique that can blindly recognize the collision degree of each on-off pattern, and then iteratively recover the data of collided users over a joint factor graph. Furthermore, we introduce a finite-length performance analysis for on-off pattern detection and iterative multi-user decoding. Through this analysis, we derive numerous numerical results, revealing the impact of various parameters on the performance of collision degree detection and multi-user decoding, respectively. By summarizing the rules observed from these numerical results, we formulate design strategies for these parameters, aiming to optimize the overall performance of our scheme. The inherent super sparse property of ODMA ensures that our scheme maintains low decoding complexity. Numerical results demonstrate that, with the implementation of our pattern collision resolution method and meticulous parameter design, the proposed scheme achieves a gap of less than 1.2 dB compared to the random coding bound for up to 300 active users. This performance surpasses state-of-the-art schemes across a broad range of user numbers.