Back to search

University of Southampton

Machine learning assisted adaptive LDPC coded system

Abstract

dc:description.abstract

This thesis investigates machine-learning-assisted adaptive transmission techniques for next-generation wireless communication systems, with an emphasis on high spectral efficiency, low latency, and robustness under practical channel impairments. By integrating adaptive modulation, low-density parity-check (LDPC) coding, index modulation, and reconfigurable intelligent surfaces (RIS), a unified learning-assisted communication framework is developed.<br/><br/>To support low-latency transmissions, short-block-length LDPC-coded adaptive modulation schemes are first investigated. Instead of relying on conventional fixed signal-to-noise ratio (SNR) thresholds, supervised learning algorithms are employed to directly learn the relationship between channel conditions and decoding performance. In particular, a k-nearest neighbours (KNN)-based adaptive scheme is proposed to enable reliable mode selection under limited bit-error-rate (BER) resolution, where Gaussian-distributed synthetic training data are constructed to overcome the scarcity of valid training samples. The proposed learning-assisted approach achieves improved throughput, enhanced SNR efficiency, and strong robustness against channel estimation errors compared with traditional adaptive modulation and coding (AMC) schemes based on predefined lookup-table thresholds.<br/><br/>The learning-assisted framework is then extended to index modulation systems, where signal detection and adaptive transmission over fading channels become significantly more challenging. Deep-neural-network (DNN)-assisted detectors are developed for compressive sensing multiple-input multiple-output (CS-MIM) systems to accurately recover sparse index modulation patterns while reducing computational complexity. Building upon this framework, adaptive index modulation schemes based on both KNN and DNN models are introduced to dynamically adjust modulation parameters and LDPC coding rates according to channel state information (CSI). Simulation results demonstrate that the proposed learning-assisted adaptive designs consistently outperforms conventional threshold-based schemes, with the DNN-assisted approach achieving the highest throughput while maintaining low detection complexity and reliable BER performance.<br/><br/>Furthermore, the proposed adaptive LDPC-coded modulation framework is generalized to RIS-assisted wireless communication systems. By incorporating cascaded base station (BS)–RIS–user channel modeling and RIS phase-shift optimization, machine-learning-assisted adaptive transmission is enabled in RIS-enhanced propagation environments. A KNN-assisted adaptive algorithm is employed to select modulation orders and LDPC coding rates without relying on predefined SNR switching thresholds. Numerical results show that RIS-assisted transmission significantly improves BER and throughput performance, enabling the system to operate closer to the Shannon capacity limit while remaining robust under severe fading conditions and imperfect channel estimation.<br/><br/>Overall, this thesis demonstrates that machine-learning-assisted adaptive transmission provides a flexible and effective design framework for future wireless communication systems. By combining adaptive modulation, LDPC coding, machine learning, index modulation, and RIS-assisted propagation, the proposed approaches enable high-throughput, low-latency, and reliable wireless transmission across a wide range of channel conditions and system architectures.

Degree

thesis:*
Name dc:type.qualificationname
Ph.D.
Level dc:type.qualificationlevel
doctoral
Grantor dc:publisher.institution
University of Southampton
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Xie, Cong
Advisors dc:contributor.advisor
  • Ng, Michael
  • El-Hajjar, Mohammed
ORCID person:identifier.orcid

Rights

Language dc:language
English, eng

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
University of Southampton
Base URL
eprints.soton.ac.uk/cgi/oai2
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Xie, Cong. Machine learning assisted adaptive LDPC coded system. doctoral thesis, University of Southampton, 2026. https://doi.org/10.5258/SOTON/PG/T309