基于主动弹性防御策略的网络控制系统零动态攻击检测

Zero Dynamic Attack Control of Networked Control Systems Based on Ac-tive Elastic Defense Strategy

摘要:

研究了一种适用于零动态攻击的网络控制系统（NCSs）的弹性控制策略，基于控制输入和测量数据设计的隐式虚假数据注入攻击.当网络化控制系统的控制信号受到网络攻击时，根据被动模型的故障检测和隔离方案，发现在攻击过程中，对设备状态变量的零动态攻击的结果是不可检测的，但在攻击结束后会变得明显.对此，利用广义似然比检测器的主动版本给出的信息在线更新卡尔曼滤波器，设计了一种具有快速恢复攻击端后闭环系统行为能力的弹性线性二次高斯控制器.实验结果显示，所提算法明显具有更高的攻击检测质量，并具有更佳的计算效率和内存利用效率，验证了算法有效性.

Abstract:

In this paper, an elastic control strategy for network control systems (NCSs) with zero dynamic attacks is studied，in which an implicit false data injection attack is designed based on input control and data measurement. When the control signal of networked control system is attacked by network, according to the fault detection and isolation scheme of passive model, it is found that the result of zero dynamic attack on device state variables is undetectable during the attack process, but it will become obvious after the attack. To solve this problem, an Elastic Linear Quadratic Gauss Controller (ELQGC) with the ability to quickly recover the behavior of the closedloop system at the attack end is designed by using the information provided by the active version of the Generalized Likelihood Ratio Detector (GLR) to update the Kalman filter online. The experimental results show that the proposed algorithm has significantly higher attack detection quality, better computational efficiency and memory utilization efficiency, which demonstrate the effectiveness of the algorithm.