Radio resource management techniques for industrial IoT in 5G-and beyond networks

Fifth generation 5G and beyond wireless technologies are expected to support Quality of Service/Experience (QoS/QoE) requirements of new and emerging Internet of Things (IoT) applications and services. Smart manufacturing is one of the target verticals for 5G-and-beyond networks. Device-to-Device (D2D) communication is a key technology to facilitate Ultra-Reliable Low-Latency Communication (URLLC). Efficient Radio Resource Management (RRM) techniques are necessary to address the challenges posed by interference. The main objective of the research work reported in this thesis is the development of new RRM techniques for D2D communication in wireless industrial setting meeting QoS/QoE requirements of end-users in cellular networks and its deployment in Factories of the Future (FoF). The algorithms and techniques developed to address RRM challenges are a combination of centralised techniques such as mathematical optimisation and distributed approaches such as matching theory and machine learning. The key contributions of this research work are the development of new RRM techniques optimising spectrum utilisation, in terms of energy efficiency and throughput performance. The first part of the thesis focuses on developing spectrum sharing schemes for D2D communication in cellular and Multi-tier Heterogeneous Networks (HetNet), resulting in new spectrum and energy-efficient solutions. The second part of the thesis focuses on ensuring reliable communication for the deployment of D2D communication in industrial settings. A new matching technique was developed to optimise matching between D2D links and cellular resources. A new stateless reinforcement learning scheme is also presented, to ensure a low-dimension state-action mapping with the latency and reliability speciications of the D2D users and minimum QoS of the cellular users. A comparative analysis of the results in terms of trade-offs between factors including performance, signalling overheads and complexity shows that the developed distributed RRM techniques outperform centralised methods for the studied industrial scenarios.