Signal Processing and Coding Techniques for 6G Ultra-reliable Low-latency Wireless Machine-type Communications

The emergence of sixth-generation (6G) wireless networks imposes physical-layer requirements, particularly ultra-reliable low-latency communication (URLLC), demanding sub-millisecond latency and reliability exceeding 99.999%. Conventional long-blocklength coding schemes are incompatible with these constraints due to inherent decoding delays and interleaving latency. This dissertation develops low-complexity, high-reliability signal processing and coding techniques tailored for short-blocklength wireless machine-type communications (MTC).

The first contribution demonstrates the feasibility of coded massive multiple-input multiple-output (mMIMO) systems by integrating random linear codes (RLCs), zero-forcing spatial multiplexing, and guessing random additive noise decoding (GRAND). Results show significant reliability and latency improvements over comparable 5G polar codes, achieving significant block error rate (BLER) gains and reduced decoding time, particularly at high signal-to-noise ratios (SNRs).

The second contribution extends symbol-level GRAND to mMIMO with arbitrary constellations through antenna-reliability ordering and constrained error enumeration, reducing decoding complexity by over 90% while preserving maximum-likelihood (ML) performance.

The third contribution introduces constraint-based GRAND variants for short low-density parity-check (LDPC) codes, including set-partitioning, Sieving-GRAND, and largely constrained GRAND, enabling over 99% reduction in candidate error patterns while maintaining ML optimality.

Finally, uniquely decodable Bose–Chowla–Lindström signature sequences are developed for grant-free random access over noisy adder channels, enabling reliable user identification and collision resolution.