Enabling Service Continuity for Stateful Service Segmentation in Mobile Edge Computing Toward 6G

초록

Future 6G applications, such as extended reality (XR) and distributed artificial intelligence (AI), require ultra-low latency, high reliability, and scalable computing resources, driving the adoption of multi-access edge computing (MEC) to meet these requirements in practical, large-scale deployments. Recently, service segmentation, where an application is decomposed into distributed subtasks, has gained attention as a flexible deployment strategy that enhances scalability and resource utilization across MEC sites. However, stateful subtasks make pipeline reconfiguration and migration prone to service disruption, and per-instance metric collection becomes inefficient at scale. To address these challenges, we extend our previous Computing-Aware Traffic Steering for the Mobile User Plane (CATS-MUP) framework to support service continuity for segmented services in 6G MEC environments. The enhanced design introduces coordinated session-modification procedures, a pipeline-level Metric Aggregator for rapid pipeline selection, and an SRv6-based buffering mechanism to preserve end-to-end service continuity for both sequential and parallel service pipelines. A testbed-based proof-of-concept implementation validates the feasibility of the proposed architecture and shows reduced handover and traffic-steering latency while preserving continuous service delivery and efficient resource usage for segmented services.

키워드