Resource-Aware Scheduling Mechanism for Real-Time Tasks in Lightweight Edge Systems

Abstract

Lightweight edge systems are complicated due to their limited execution capabilities and communication problems. Due to the highly complex and dynamic nature of real-time tasks, scheduling services to which nodes faces the unique challenges of resource allocation and network bandwidth coordination. To address these challenges, in this paper we introduce a scheduling mechanism called RASM. Using this mechanism, intelligent real-time tasks at the lightweight edge can be efficiently assigned to the appropriate computing nodes without waiting in long queues at the edge nodes or being allocated to the cloud indiscriminately. Our approach is a novel technology for scheduling resources in the server environment. Particularly, our method defines a cache list to record the task execution location, priority and system remaining resources, so as to unload the task to the appropriate node for execution and maximize the quality of service. Our approach is scalable and efficient with the provided infrastructure resources. Our evaluation shows that over the same conditions, RASM reduces the delay by more than 15% compared with the Cloud Only environment, and solves more tasks than the Edge Only environment.