DYNAMIC NETWORK RESOURCE MANAGEMENT IN IOT

Abstract

This dissertation presents innovative solutions to address the challenges in the 5G & IoT network, including self-adaptive load balancing, software-defined dynamic 5G network slice management, unsupervised traffic classification, and dynamic network slicing traffic prediction. We propose self-adaptive load balancing approaches and software-defined dynamic 5G network slice management middleware for industrial IoT use cases, respectively. These solutions provide effective self-adaptation, self-healing, and predictable communication for real-time and autonomous needs of IoT applications while effectively sharing network resources. To further improve the network resource utilization and minimize the end-to-end latency, we propose two works for classifying and predicting IoT network traffic, which aims to improve the efficiency, reliability, and predictability of the network.