Q-Learning-Based Task Scheduling for Low-Latency Edge Offloading in MEC Systems

Mobile Edge Computing (MEC) handles energy constraints and enhances performance by facilitating the effective offloading of applications that are delay-sensitive and computationally demanding from mobile devices. Nevertheless, high computing complexity, network limits, and the possibility of task failures brought on by user mobility and resource constraints make efficient task scheduling difficult. To address the limitations, the Q-Optimize OffloadPro Framework (QOOPF) is proposed as a task scheduling and offloading system designed to manage high virtual machine utilization, reduce latency, and improve resource efficiency in MEC. The framework incorporates the OffloadPro Scheduling Method (OPSM), which optimizes task assignment by prioritizing tasks based on a critical path approach to ensure effective offloading. To ensure that task offloading choices in edge computing settings are made dynamically, this technique is augmented by a Deep-Q-Driven Policy-Value Optimizer that has been trained on large amounts of task data. QOOPF dynamically balances computational loads, reduces task failures, and increases resource consumption by combining Policy Value Optimization (PVO) with Q learning. The experimental findings demonstrate QOOPF achieves a makespan of 720 seconds and variance of 30.03 for 300 tasks, with VM results showing a makespan of 445.88 seconds and variance of 4.58 for 16 VMs, scaling efficiently with up to 608.54 seconds and 6.08 variance for 32 VMs for high-demand MEC situations. This method provides an efficient, scalable solution for dynamic computing requirements while successfully addressing scheduling constraints.