Orthogonal Matching Non-Negative Least Squares for Activity Detection in Unsourced Random Access

Unsourced random access (URA) is challenged to support massive connectivity with reduced latency for the future communications networks. Activity detection (AD) has a close correlation with URA under limited number of channel uses. Recently, non-Bayesian approach has rigorously improved the AD efficiency and been well-proven under the structure of non-negative least squares (NNLS). Even though NNLS-oriented algorithm guarantees convergence towards a local optimal point, further performance improvement is possible. This work points out a potential error propagation phenomenon in NNLS due to non-orthogonality among codewords, based on which an orthogonal matching-NNLS is proposed. Numerical results illustrate the viability of the proposed algorithm.