Queue-Aware Multi-Objective Edge Offloading and Resource Provisioning for Dense IoT Networks Under Synthetic Workload Dynamics

Abstract

The intensive growth of Internet of Things (IoT) devices has led to the explosion of the quantity of data produced at the network edges. The high-energy efficiency and the ability to provide low latency demand distributed computing architectures to process such data efficiently. Edge computing is one of the promising paradigms with the ability to perform computational tasks at a position nearer to the sources of data and thus alleviating congestion in the network and enhancing response time. Nonetheless, thick IoT networks present the major issues associated with resources distribution, task scheduling, and queuing in edge servers. Traditional offloading techniques commonly focus on a single objective optimization relying on minimization of latency or energy use, and this might result in poor performance when operating in dynamic network constraints. The study is a multi-objective edge offloading framework based on queues that is suggested to operate in dense IoT networks with nonuniform workloads and variable resource allocation. The suggested system incorporates the queue length predictor, the resource-sensitive schedule, and the resource on-demand provisioning innovations into a single decision system. Large-scale IoT traffic patterns are synthetically created as models of workload without the use of real-time data gathering. The experiments with simulations show that the suggested queue-aware offloading strategy can decrease the task completion time by a significant margin, and the energy consumption of edge nodes is balanced. The results point to the efficiency of considering the behaviour of queues when managing edge resources of next-generation IoT systems.