Adaptive Slice Handover for Cloud-Native Vehicular Edge Networks

Cloud-native vehicular edge networks enable efficient and scalable service delivery by deploying containerized functions close to vehicles. To accommodate diverse service requirements, edge slices are configured as logically isolated virtual networks with dedicated resources and tailored performance. As vehicles traverse different edge domains and service demands evolve, slice handover becomes essential to ensure service continuity. However, most existing approaches focus mainly on mobility, overlooking the dynamics of service requirements and the heterogeneity of slice types. To address this gap, we propose a deep reinforcement learning-based adaptive slice handover (DASH) scheme for vehicular edge networks, which jointly consider mobility patterns, service dynamics, and slice diversity for optimized decision-making. In addition, a multi-factor scoring function is introduced to assess handover decisions in real time, taking into account resource availability, communication reliability, and decision stabsility. Simulation results demonstrate that DASH significantly outperforms existing methods in terms of handover success rate and decision effectiveness.