Innovative approach to enhancing MAC protocols in modern wireless networks

Abstract

Modern wireless networks face significant challenges in managing interference and efficiently accessing the medium, especially in densely deployed, dynamic environments. This thesis first presents a comprehensive review of multi-hop wireless networks, detailing the evolution of network architectures, interference management strategies, and the pivotal role of network coding—particularly Physical-layer Network Coding (PNC)—in enhancing throughput and reliability. Building on this foundation, we propose a novel AI-based PNC MAC protocol (AI-PNCMP) for Two-Way Relay Channels, where Long Short-Term Memory (LSTM) and Temporal Convolutional Network (TCN) models are employed to predict transmission behaviour at the relay node. By accurately forecasting when simultaneous transmissions occur, the relay can dynamically choose between ordinary and PNC modes, enhancing slot utilization and improving overall delay and throughput metrics. Simulation experiments using Python, TensorFlow, and Keras demonstrate that our adaptive approach mitigates traditional MAC protocols’ limitations and offers substantial gains in network performance. This work underscores the potential of integrating predictive intelligence into the MAC layer, laying a solid foundation for future research in adaptive and intelligent wireless communication systems.