在移动边缘计算的物联网(Mobile Edge Computing-enabled Internet of Things Networks,IoT-MEC)中,物联终端的高移动性、服务请求的随机到达性以及网络流量的实时变化,导致原有应用场景下的资源配置与服务部署不再完全匹配.如何有效利用网络提供的资源以实现服务功能链(Service Function Chain,SFC)的实时部署和重构是一个重要的挑战.针对用户的高移动性和网络流量的实时变化造成的SFC性能需求和已分配资源不匹配的问题,提出IoT-MEC网络中基于用户移动和资源需求预测的SFC重构策略.建立以SFC的端到端时延和重构成本最小化为目标的整数线性规划模型;设计基于注意力机制的Encoder-Decoder移动用户轨迹预测模型和基于长短期记忆(Long Short-Term Memory,LSTM)网络的虚拟网络功能(Virtual Network Function,VNF)实例资源需求预测模型,分别准确预测用户移动轨迹和节点负载;基于预测结果提出 SFC 主动重构(Predict-based SFC Active Reconfiguration,PSAR)启发式算法,确保在服务质量(Quality of Service,QoS)下降之前,提前完成VNF迁移和路由更新,实现SFC的主动重构和无缝迁移,保证网络的一致性高质量服务.仿真结果表明,所提算法有效降低了 SFC端到端时延和重构成本.
Active Reconfiguration Method for Service Function Chain in IoT-MEC Network
In the Mobile Edge Computing-enabled Internet of Things Networks(IoT-MEC),the high mobility of IoT terminals,the random arrival of service requests,and the real-time changes of network traffic result in resource allocation and service deployment no longer fully matching in the original application scenarios.How to effectively utilize the resources provided by the network to achieve real-time deployment and reconstruction of Service Function Chain(SFC)is a great challenge.Considering the mismatch between performance requirement of SFC and the allocated network resources caused by the high mobility of users and the real-time variation of network traffic in IoT-MEC,the user mobility and resource demand prediction based SFC reconfiguration strategy is investigated.First,an integer linear programming model is established with the objective of minimizing the end-to-end latency and reconfiguration cost of SFC.Second,an attention mechanism-based encoder-decoder mobile user trajectory prediction model and a Long Short-Term Memory(LSTM)network based Virtual Network Function(VNF)instance resource demand prediction model are designed to accurately predict user mobile trajectory and node load,respectively.Finally,according to the prediction results,a heuristic algorithm,Predict based SFC Active Reconfiguration(PSAR)is developed to implement the migration of VNFs and update routing in advance,ensuring proactive reconfiguration and seamless migration of SFCs before Quality of Service(QoS)degrades,and guaranteeing consistency and high-quality services of the network.Experimental results show that the proposed algorithm can efficiently reduce the end-to-end delay and reconfiguration cost of SFC.
国家科技期刊平台
NSTL
万方数据
