Dynamic Workload Scheduling in Edge Computing

Edge computing has fixed and limited resource capacity, thus low resource scalability is one of the inherent shortages. Nevertheless, on the one hand, task may be excessive of mobile users, on the other hand, task arrivals at edge nodes are usually dynamic both spatially and temporally. The cloud-assisted mobile edge computing system is a critical architecture to enhance resource scalability and servicing dynamic mobile tasks with high resource efficiency, by dynamically tuning the usage of cloud resources. However, tasks arrivals are dynamic, and distribution of mobile users and edge nodes is heterogeneous. Therefore, workload imblance leads to high task response time and low resource efficiency in the cloud-assisted mobile edge computing system. In this chapter, we present dynamic workload scheduling in the cloud-assisted mobile edge computing, including Edge-Device and Cloud-Edge workload scheduling, both aiming at optimizing task response time within resource budget limit. The problem of scheduling is challenging due to task arrival dynamics, edge node heterogeneity, and computation–communication delay trade-off. To address these challenges, we propose the Water-filling Based Dynamic Task Scheduling (WiDaS) algorithm, which leverages the Lyapunov optimization method and the idea of water filling to efficiently schedule mobile tasks among edge nodes (and the cloud) while dynamically tuning the usage of cloud resources. We conduct extensive simulations to evaluate the performance of WiDaS under trace-driven and mathematical traffic patterns. The simulation results demonstrate that WiDaS is both efficiency and effectiveness, providing a twofold benefit in terms of task response time optimization and resource efficiency improvement.