Sparse parity random linear codes for URLLC in channels with memory

The use of short error-correcting codes is vital for achieving an Ultra-Reliable Low Latency Communications capability for supporting the Internet of Things with minimal decoding complexity. Here we examine the performance and decoding complexity of two code-agnostic decoders for a new class of full-rank sparse random linear codes derived from the MacKay-Neal LDPC construction. We consider decoding by a noise guessing decoder such as GRAND-MO and by a binary matching pursuit decoder to which list decoding is added. Comparisons are made in the context of error correction performance and computational complexity.