Risk-Aware SFC Placement Method in Edge Cloud Environment

Network Function Virtualization (NFV) is a technology that separates software from hardware, NFV increases the manageability, scalability, and flexibility of network function configuration. However, with these conveniences, it is the scheduling and management of complex virtual network resources created by NFV. The key to these problems is the deployment of the Service Function Chain (SFC). There are two major problems in SFC deployment: 1) Complexity of resource scheduling. 2) Vulnerability of SFC. To solve the above problems, we propose a riskaware SFC deployment method. LSTM (Long Short-Term Memory) is used to predict possible attacks, and DQN (Deep Q Network) uses its results to complete SFC deployments. Our model is validated in a simulation network. The results show that proposed risk-aware SFC deployment is significantly better than traditional resource-oriented deployment in terms of network elasticity, and is not inferior to it in terms of latency.