Find in Library

This paper introduces SATS, an innovative online task scheduling strategy tailored for hierarchical Mobile Edge Computing (MEC) architectures. Leveraging a Simulated Annealing-based approach, SATS extends the applicability of this method beyond offline scheduling, demonstrating its efficacy in real-time task allocation. Crucially, the paper underscores the pivotal role of accurate service request predictions in enhancing SATS’ performance. Through a comparative analysis of three prediction models, neutral, conservative, and optimistic, the paper reveals that SATS achieves optimal outcomes when paired with a conservative predictor, which deliberately overestimates service request volumes. In fact, employing this predictor yields significantly higher acceptance rates and reduced processing times. Remarkably, SATS, coupled with a conservative predictor, achieves an acceptance ratio within a mere 5% deviation from an ideal scenario where the frequency of service request arrivals is known in advance, maintaining consistency across various experimental scenarios.