QoS-aware Scheduling of Periodic Real-time Task Graphs on Heterogeneous Pre-occupied MECs

In latency-sensitive applications, efficient task scheduling is crucial for maintaining Quality of Service (QoS) while meeting strict timing constraints. This paper addresses the challenge of scheduling periodic tasks structured as directed acyclic graphs (DAGs) within heterogeneous, pre-occupied Mobile Edge Computing (MEC) networks. We propose a modified version of the Heterogeneous Earliest Finish Time (HEFT) algorithm designed to exploit residual processing capacity in preoccupied MEC environments. Our approach dynamically identifies idle intervals on processors to create a feasible hyperperiodic schedule that specifies an allocated virtual machine (VM), task version, and start time for each task. This scheduling strategy maximizes the aggregate QoS by optimizing task execution without disrupting the existing periodic workload, while also adhering to periodicity, precedence, and resource constraints.Experimental results demonstrate that our method achieves enhanced load balancing and resource utilization, highlighting its potential to improve performance in heterogeneous MEC infrastructures supporting real-time, periodic applications.