Providing Worst-Case Latency Guarantees With Collaborative Edge Servers

Mobile Edge Computing (MEC) is a promising computing paradigm that provides cloud computing services in proximity to end users. Due to the bursty and spatially imbalanced arrival of computation tasks, the workload on different edge servers may vary wildly. To improve the Quality of Experience (QoE), peer offloading has been proposed as an effective cooperation method that offloads tasks from busy edge servers to idle ones. Although the average latency has been extensively considered in the design of peer offloading strategies, the worst-case latency, a common Quality of Service (QoS) requirement that is usually demanded by latency-sensitive applications, yet receives much less attention. In this paper, we study the task scheduling among collaborative edge servers and propose an online algorithm that aims to maximize the system utility under the worst-case latency requirement and long-term energy consumption constraints. Both theoretical analysis and simulation results demonstrate that our algorithm performs well under various situations.