Optimizing IoT edge device task offloading using adaptive search strategies framework and contextual bandit algorithm

Edge computing processes Internet of Things (IoT) data locally, reducing the need for remote data centers. However, resource-limited wireless IoT devices face significant computational challenges. This paper presents the adaptive search strategies framework with probabilistic heuristic choice (ASFPHC), a new approach designed to improve task offloading for these devices. ASFPHC combines the contextual bandit algorithm (CBA) with an Epsilon-decreasing strategy to optimize resource distribution and job outsourcing in mobile edge computing. The framework targets two main quality of service goals: minimizing computation energy and time, while also accounting for device mobility. ASFPHC balances exploration and exploitation using CBA, allowing for adaptive and efficient decision-making. Performance tests show that ASFPHC significantly enhances efficiency in resource-constrained environments. Specifically, the model HGWRPO achieved a high accuracy of 97.46%, demonstrating its effectiveness in solving complex optimization problems. This paper highlights ASFPHC's potential to improve task offloading and resource management for IoT edge devices.