ML-BASED TASK OFFLOADING FOR EDGE-CLOUD HYBRID SYSTEMS: A SIMULATION-BASED APPROACH

Abstract

Edge–cloud hybrid computing is rapidly transforming how we manage computational workloads in IoT and mobile environments. By combining the strengths of edge and cloud resources, these systems offer faster processing and better service quality. Yet, deciding when and where to offload tasks remains a tough challenge especially with fluctuating network conditions, diverse hardware capabilities, and the constant balancing act between speed, energy use. This study introduces a machine learning–driven simulation framework designed to optimize task offloading decisions in such hybrid systems. Using Support Vector Machines (SVM) and Logistic Regression, the framework predicts offloading choices based on key parameters like latency, throughput, uplink/downlink traffic, task duration, radio access type (RAT), and network resource availability.