DDQN-SFCAG: A service function chain recovery method against network attacks in 6G networks

Service Function Chain (SFC) as an effective solution can satisfy the diverse service requirements of six application scenarios in Network Function Virtualization (NFV)-enabled 6G networks. Resilience as a new key capability indicator in 6G networks puts forward higher requirements for the Quality of Service (QoS) of SFCs. In this article, we study the resilience recovery method in network attack scenarios. We make full use of the monitoring and early warning capability of the monitoring function and propose a proactive recovery method called Double Deep Q-Network based on SFC Attack Graph (DDQN-SFCAG). We fully consider the characteristics of network attacks to generate SFC attack graphs and determine the recovery strategy of SFC according to the service security requirements to provide guidance for recovery. Among them, we design three recovery modes for the recovered Virtual Network Functions (VNFs) to determine the optimal recovery strategy and avoid resource waste. We formalize the SFC recovery problem, which aims to minimize the recovery cost while meeting the recovery strategy. In order to shorten the interruption time, we use DDQN to quickly solve the recovery solution to ensure optimal recovery performance. Our extensive evaluation shows DDQN-SFCAG has excellent recovery performance in network attack scenarios and can reduce the recovery cost by at least 31% compared to the state-of-the-art methods.