🤖 AI Summary
This work addresses pilot-free error-control coding for ultra-reliable low-latency communication (URLLC) in 6G, specifically targeting noncoherent single-input multiple-output (SIMO) systems without channel state information (CSI) and under short blocklength constraints. To overcome the computational intractability of conventional maximum-likelihood (ML) detection, we propose the ML-based matching pursuit (MLMP) decoder—a low-complexity iterative algorithm that integrates greedy search with ML-based metric evaluation and incorporates parallel candidate exploration to accelerate convergence and improve accuracy. Built upon the sparse regression code (SPARC) framework, MLMP enhances robustness in noncoherent detection while maintaining practical complexity. Experimental results demonstrate that MLMP achieves superior block error rate (BLER) performance compared to existing orthogonal matching pursuit (OMP)-type decoders and pilot-assisted polar codes—particularly under short blocklengths and low signal-to-noise ratios—offering a viable, low-complexity solution for pilot-free noncoherent short-packet transmission.
📝 Abstract
We study the sparse regression codes over flat-fading channels with multiple receive antennas. We consider a practical scenario where the channel state information is not available at the transmitter and the receiver. In this setting, we study the maximum likelihood (ML) detector for SPARC, which has a prohibitively high search complexity. We propose a novel practical decoder, named maximum likelihood matching pursuit (MLMP), which incorporates a greedy search mechanism along with the ML metric. We also introduce a parallel search mechanism for MLMP. Comparing with the existing block-orthogonal matching pursuit based decoders, we show that MLMP has significant gains in the block error rate (BLER) performance. We also show that the proposed approach has significant gains over polar codes employing pilot-aided channel estimation.