CRC-Aided Sparse Regression Codes for Unsourced Random Access
Haiwen Cao, Jiongyue Xing, Shansuo Liang
Abstract
This letter considers a coding scheme for unsourced random access (URA) based on sparse regression codes (SPARCs). In particular, an efficient concatenated coding scheme is proposed, which concatenates SPARCs and cyclic redundancy check-based block Markov superposition transmission (CRC-BMST) codes. A hybrid decoder consisting of a successive cancellation algorithm and a simplified approximated message passing (AMP) algorithm is presented for inner SPARCs, and an improved tree decoder is proposed for outer CRC-BMST codes by introducing a pruning technique. Simulation results show the proposed coding scheme outperforms the coded compressed sensing (CCS) scheme with lower computational complexity.
Topics & Concepts
Computer scienceAlgorithmRandom accessCoding (social sciences)Markov chainDecoding methodsComputational complexity theoryPermissionBlock (permutation group theory)Theoretical computer scienceMathematicsComputer networkGeometryPolitical scienceMachine learningLawStatisticsSparse and Compressive Sensing TechniquesCooperative Communication and Network CodingError Correcting Code Techniques