mobility management

Existing surveys on live migration predominantly emphasize theoretical mechanisms while neglecting practical deployment constraints and technical adaptability. This paper systematically compares live migration techniques for containers and virtual machines across three dimensions—migration mechanisms, migration units, and infrastructure characteristics—to analyze performance, overhead, and compatibility trade-offs. It introduces, for the first time, an integrated evaluation framework that jointly considers migration objectives (e.g., cloud-edge coordination), operational constraints (e.g., resource and network limitations), and adoption heterogeneity, thereby enabling scenario-aware technology assessment and evolutionary guidance. Through multidimensional empirical analysis—including pre-copy/post-copy strategies, dirty-page tracking, CRIU-based checkpointing, KVM/QEMU live migration, and CRI-O hot migration—the study identifies five pervasive challenges. The findings provide actionable, deployment-oriented insights for technology selection and optimization in elastic scheduling and other production-critical scenarios.

To address service disruption and degraded Quality of Experience (QoE) caused by stateful microservice migration in edge networks, this paper proposes the first KPI-driven, end-to-end cooperative lightweight orchestration framework. It integrates incremental memory snapshotting with container live migration for low-overhead state capture, and combines multi-objective optimization scheduling with real-time QoE-aware feedback control to guarantee strict service continuity in latency-sensitive scenarios. Its key innovation lies in unifying network-layer (e.g., latency, bandwidth) and application-layer (e.g., availability, response time) KPIs into a dynamic decision-making model—overcoming the traditional decoupling of state synchronization and resource scheduling. Experimental evaluation demonstrates up to 77% reduction in migration downtime; in representative use cases—autonomous drone control and multi-object tracking—the framework achieves end-to-end latency ≤50 ms and service availability ≥99.99%, fully satisfying stringent edge computing requirements for real-time performance and high reliability.

In cloud-edge collaborative continuum environments, microservice scheduling faces challenges including complex call-graph dependencies, dynamic inter-node communication latency, and fluctuating bandwidth—conditions that existing evaluation frameworks struggle to reproduce realistically and controllably. To address this, we propose iDynamics, the first modular dynamic simulation framework specifically designed for evaluating cloud-edge microservice scheduling policies. Our approach decouples the modeling of call-graph topology, time-varying network latency, and bandwidth dynamics; enables fine-grained environmental configuration and systematic, multi-policy comparison; and integrates graph dynamical analysis, programmable latency injection, and containerized deployment. Empirical validation on real-world cloud-edge platforms demonstrates that iDynamics significantly enhances the realism, reproducibility, and comparability of scheduling evaluations.

To address service interruptions and inefficient resource allocation in multi-edge Internet of Vehicles (IoV) caused by high vehicle mobility and limited base station coverage, this paper proposes the SR-CL framework. Methodologically, it first decouples the mixed-integer nonlinear programming (MINLP) problem into a learnable service migration subproblem and an analytically tractable resource allocation subproblem; second, it designs an asynchronous-update Actor-Critic deep reinforcement learning algorithm to enable seamless service migration; third, it derives a real-time optimal resource allocation policy via convex optimization theory. Simulation results based on real-world vehicular trajectory data demonstrate that SR-CL achieves a 32% faster convergence rate and reduces end-to-end latency to 18.7 ms (mean), significantly outperforming state-of-the-art baselines. The framework thus enables efficient, dynamic, and quality-guaranteed collaborative optimization in highly mobile edge-IoV environments.

Service deployment for mobile users in edge-cloud multi-tier networks faces significant challenges under low-latency and highly dynamic conditions, particularly regarding service placement and migration. Method: This paper proposes the first decentralized asynchronous service deployment framework that operates without global visibility or a centralized orchestrator. It integrates a distributed consensus protocol, an asynchronous event-driven scheduler, and a resource-aware lightweight placement algorithm to enable elastic scaling and fault tolerance while guaranteeing latency constraints. Contribution/Results: Evaluated on large-scale simulations driven by real-world mobility traces, the framework achieves performance close to that of optimal centralized solutions. It incurs negligible communication overhead (effectively zero bandwidth consumption) and reduces deployment costs by 37%. The design significantly enhances system scalability and robustness, demonstrating practical viability for dynamic edge-cloud environments.

To address high latency, substantial coordination overhead, and weak fault tolerance arising from centralized scheduling in cloud-edge collaborative microservice systems, this paper proposes a decentralized scheduling architecture leveraging service mesh sidecar proxies. The core innovation embeds lightweight, autonomous scheduling logic directly into each sidecar, enabling fully localized service discovery, load-aware routing, and real-time decision-making—eliminating dependence on a central controller. By exploiting the distributed traffic control capabilities and programmability of service meshes, the architecture supports dynamic topology adaptation and self-healing under failures. Experimental evaluation demonstrates that, under diverse workload pressures, the approach reduces average response latency by 37.2% and scheduling coordination overhead by 89.5% compared to conventional centralized schemes. Moreover, system throughput scales nearly linearly with node count, confirming strong scalability and real-time performance.

This work addresses the critical need for personalized, real-time AI inference at AI-native user equipment in 6G mobile networks, a capability hindered by the absence of user-centric edge resource allocation in existing Net4AI architectures. To bridge this gap, we propose SubEdge, a novel system that introduces the “computation context”—a data structure binding user identity to dedicated inference containers, edge nodes, and access permissions. SubEdge enables joint migration of communication and computation resources through a mobility event–driven cooperative handover mechanism, operating without modifications to the 3GPP core network. Leveraging NEF APIs and containerized deployment, our implementation demonstrates significant performance gains: the 95th-percentile latency drops from 22.9 ms to 12.2 ms, achieves zero packet loss across six handovers, sustains a 99.92% frame delivery ratio for 30-fps inference tasks, and supports 50 concurrent users completing 1,560 migrations.

This work addresses the significant performance overhead in service meshes caused by frequent context switches due to sidecar proxies. It proposes L7FP, the first fast-path framework that automatically compiles high-level Layer 7 policies—such as HTTP/2 and TLS—into eBPF programs and offloads their execution into the kernel, accelerating microservice communication without requiring application code modifications. For policies unsupported in-kernel, L7FP transparently falls back to user-space proxying. This approach achieves, for the first time, automated synthesis and execution of Layer 7 policies entirely within kernel space while preserving full policy compatibility. Experimental results demonstrate substantial performance gains: compared to existing service meshes, L7FP reduces median request latency by up to 6× and improves throughput by up to 3×.

This work addresses the challenge of frequent service function chain (SFC) reconfigurations in multi-access edge computing caused by user mobility, which, when combined with the time-consuming lifecycle operations of virtual network functions (VNFs), can severely degrade service quality if their dynamics are ignored. The paper presents the first joint modeling of VNF lifecycle dynamics and user mobility prediction, proposing a proactive SFC deployment and reconfiguration approach. By forecasting connectivity changes and integrating uncertainty-aware modeling with resource scheduling optimization, the method enables lifecycle-aware SFC embedding. Experimental results demonstrate that, under realistic VNF lifecycle constraints, the proposed scheme significantly reduces service disruption and achieves performance close to that of an idealized instantaneous deployment benchmark.

This work addresses the challenge of dynamic server selection in edge computing under stringent latency SLOs, where decisions must jointly manage tail-risk violations and switching stability. The authors propose a lightweight, interpretable decision framework that uniquely co-optimizes tail-risk control and switching stability: it estimates SLO violation risk using normal approximation and the Cantelli inequality, while incorporating a hysteresis mechanism to suppress excessive switching. Experimental results under a 0.5-second SLO demonstrate that, compared to a baseline relying solely on mean latency, the proposed method reduces deadline miss rate from 39% to 34%, cuts switching frequency by 88% (down to 5.5%), and maintains average latency stably at 0.45 seconds, thereby significantly enhancing both system robustness and efficiency.

This paper identifies a critical gap in high-performance data transfer research: an overemphasis on network bandwidth while neglecting end-to-end bottlenecks—including latency, TCP congestion control, host CPU limitations, and virtualization—leading to severe discrepancies between benchmark results and real-world production performance. To address this, the authors propose a hardware–software co-design paradigm and develop a latency-programmable testbed. Leveraging high-fidelity wide-area network (WAN) modeling and cross-continental 100 Gbps measurements (Switzerland–California), they systematically isolate key constraints at the network edge and host side. Results demonstrate that primary bottlenecks reside predominantly at the network edge—not the core—and that stable, predictable data movement is achieved across 1–100+ Gbps. This significantly enhances performance fidelity in complex, heterogeneous environments.