Resource allocation for ultra-reliable and low-latency 5G communication

Recently, emerging applications and services such as augmented/virtual reality, autonomous driving, and smart factory are enabled by 5G networks in which ultra-reliable low latency communication (URLLC) is a key enabler for mission-critical applications. The vast majority of URLLC system designs consider average latency or outage probability as a performance metric or quality-of-service requirement. However, focusing on an average-based system design is inadequate for ensuring reliability and latency guarantees in mission-critical applications. Instead, this requires further looking into higher-order statistics, characterization of the extreme event with a very low occurrence probability, or other worst-case metrics. Additionally, an efficient resource allocation is instrumental in enabling URLLC when the data traffic and wireless link quality are dynamic and non-deterministic.

In this thesis, we study the problems of power minimization in URLLC scenarios and propose dynamic resource allocation solutions in three different use cases which include multi-access edge computing (MEC), vehicular communication, and industrial Internet of things (IIoT). In the MEC study, we trade off allocated communication and computational resources for task offloading and local computation while alleviating the impact of delay threshold violation events. In the vehicular communication study, we minimize a network-wide maximal latency by allocating vehicles' transmit power. In the IIoT investigation, we propose a joint resource allocation and data-updating policy, taking into account the freshness of the updated information and the information decoding error due to the finite blocklength transmission. Simulation results validate the effectiveness of the proposed resource allocation approaches and their superiority over several baselines.