Edge-Cloud Continuum Computing: Task Offloading Optimization, Latency-Aware Scheduling, and Mobility-Driven Workload Migration in 5G-Enabled Mobile Edge Environments
Mobile Edge Computing (MEC) in 5G-enabled environments introduces a computational continuum spanning ultra-low-latency edge nodes, regional fog nodes, and centralized cloud data centers, enabling latency-sensitive applications -- including augmented reality, industrial automation, and autonomous vehicle coordination -- that cannot tolerate cloud-only round-trip latencies. Optimal task placement across this continuum requires dynamic offloading decisions that account for task computational requirements, data transfer costs, edge node capacity, user mobility patterns, and SLA constraints simultaneously. This paper presents EdgeOpt, a multi-objective task offloading optimization framework for 5G MEC environments that employs a Deep Q-Network (DQN) agent trained to balance execution latency, energy consumption, and edge resource utilization in real-time. EdgeOpt is evaluated in a 5G MEC testbed comprising three edge nodes, one fog aggregation layer, and simulated cloud infrastructure, processing workloads representative of AR rendering, industrial sensor fusion, and V2X communication scenarios. EdgeOpt achieves 38% lower mean task execution latency and 27% lower edge energy consumption compared to greedy offloading baselines, while maintaining edge utilization above 78% under high mobility scenarios. We characterize the mobility-induced workload migration problem and introduce the Mobility-Aware Migration Cost Model (MAMCM) to quantify handover-induced service disruption risk. This work provides architectural and algorithmic foundations for latency-optimized edge-cloud continuum orchestration.
