Exploiting Matrix Information Geometry for Integrated Decoding of Massive Uncoupled Unsourced Random Access

Abstract

In this paper, we explore an efficient uncoupled unsourced random access (UURA) scheme for 6G massive communication. UURA is a typical framework of unsourced random access that addresses the problems of codeword detection and message stitching, without the use of check bits. Firstly, we establish a framework for UURA, allowing for immediate decoding of sub-messages upon arrival. Thus, the processing delay is effectively reduced due to the decreasing waiting time. Next, we propose an integrated decoding algorithm for sub-messages by leveraging matrix information geometry (MIG) theory. Specifically, MIG is applied to measure the feature similarities of codewords belonging to the same user equipment, and thus sub-message can be stitched once it is received. This enables the timely recovery of a portion of the original message by simultaneously detecting and stitching codewords within the current sub-slot. Furthermore, we analyze the performance of the proposed integrated decoding-based UURA scheme in terms of computational complexity and convergence rate. Finally, we present extensive simulation results to validate the effectiveness of the proposed scheme in 6G wireless networks.