IMPACT：意图驱动多智能体策略保障云边微服务迁移 SLO
原标题：IMPACT: Intent-driven Multi-agent Policy with Attention for SLO-guaranteed Microservice Migration in Cloud-edge Systems
AI 摘要
该论文提出 IMPACT，一个面向云边系统微服务迁移与带宽控制的意图驱动多智能体强化学习框架。每个边缘云作为自主智能体，将本地 SLO 风险、迁移紧迫性和计算压力编码为语义意图，并通过双重注意力机制先聚合相关同伴意图、再过滤本地观测。在 5 边缘和 20 边缘场景实验中，IMPACT 相比最先进的分解式 MARL 和启发式基线将平均延迟降低 30–50%、尾延迟偏差降低 40–70%，并在严格阈值下实现近零 SLO 违规率。
正文节选
IMPACT: Intent-driven Multi-agent Policy with Attention for SLO-guaranteed Microservice Migration in Cloud-edge Systems Abstract Ensuring strict tail-latency service-level objectives (SLOs) in dynamic mobile edge computing (MEC) systems remains challenging because user mobility, wireless fading, bursty workloads, and partial observability jointly undermine reliable cloud-edge orchestration. Existing microservice migration methods predominantly optimize average delay and often decouple migration