A priority-driven task offloading and scheduling algorithm for edge-based health monitoring system

With the ubiquitous deployment of IoT sensors and the proliferation of mobile applications, large volumes of data are generated at the network edge. Smart health monitoring applications deployed on edge servers receive diverse requests offloaded by wearable devices attached to patients in hospitals. These continuous requests need computing resources at the edge servers rather than being offloaded to the cloud, resulting in higher latency. Although edge computing offers low-latency services to incoming traffic, its limited resource availability poses a challenge. Existing approaches focus on task size, priority, deadline, fastest-machine-first, or any of their combinations. However, they struggle to improve multiple performance metrics simultaneously, particularly for workloads with varying priorities. This paper addresses this problem by proposing a priority-driven task offloading and scheduling algorithm (PDTOS). It introduces a deadline computation mechanism to prioritize task execution. Initially, a suitable server providing the earliest finish time is selected depending on task priority. Subsequently, an evaluation mechanism that considers the task’s execution time, earliest finish time, and corresponding deadline is used to assign appropriate local or remote resources. A real-world Human Vital Sign dataset is utilized to validate the performance of the proposed algorithm against state-of-the-art benchmark techniques. The results demonstrate significant performance improvements by PDTOS compared to PTSRA, WOA, and PS in reducing makespan and network latency. PDTOS achieves a 16.5% reduction in makespan and a 23% reduction in network latency. Additionally, it decreases energy consumption while increasing resource utilization. The Friedman test is conducted to illustrate the statistical significance of the obtained results.