Research On Security Problem Of Networked Control Systems For DoS Attacks

Traditional local control systems need to design the communication channels when dealing with the remote control,so costs of designing,wiring and maintenance are increased.With the popularity of the internet,networked control has become one of the hot topics in the international control fields,and has been widely used in many fields.Networked Control Systems(NCSs)use the internet channel for transmission of control and measurement of signals.Although it can save all costs,there are many problems and challenges to be solved in theory and practice.Closed control system accessing to the Internet brings many problems such as network induced delay,packet loss,disorder,communication bandwidth constraints and network security,especially the network security problem is very prominent.Network attacks on the internet protocol are gradually matured,but defense strategies can only take the follow up step by step way updating strategies.The analysis on NCSs under threats of cyber attacks is as a new interdisciplinary research field,the related research is still in its infancy and many key problems need to be further studied.This paper analyzes and models the NCSs under the Denial of Service(DoS)attacks,and designs the coping strategies at the control level.Main contents are given as follows:Firstly,DoS attacks are modeled under the optimal scheduling,and stability problem of NCSs with actuator saturation are studied under the attacks.If there exist no control signals arriving for a long time,the states of a non-self-stabilization system with actuator saturation will diverge to outside of domain of attraction,then the original controller can not stabilize the NCSs.The attacker is to rely on this disadvantage for continuous DoS attacks,then the optimal scheduling of DoS attacks can make NCSs out of control at the minimum cost.The minimum number of successive attacks that can make the system out of control is estimated by means of the domain of attraction for NCSs.The predictive controller is designed for the attacks,and the stability conditions of predictive control systems are obtained.The effectiveness of the proposed algorithm is verified by a simulation of the inverted pendulum system.Secondly,intrusion Detection Systems(IDSs)is usually configured at the controller side to prevent various attacks from networks.Considering the facts that both IDSs and DoS attackers have unknown policy information to each other in the actual situations.Combined with the existing correlation modeling methods,DoS attacks are described by attack frequencies and durations.The mixed strategy Nash equilibrium game is used to model the mutual actions between the offensive and defensive sides,and the mixed Nash equilibrium strategy under incomplete strategy information is obtained.Then with the proof of switched systems,the stability conditions of NCSs with prediction and compensation are given under DoS attacks.The equilibrium solution of both sides of the game is compared with unknown strategies and complete strategies.The effectiveness of the proposed method is verified by simulation of networked ball and beam system.Thirdly,a single form of DoS attacks is often difficult to achieve the desired attack targets for attackers.In practice,NCSs often suffers from multiple network attacks.In this paper,a type of multiple attack is studied,which uses DoS attacks as the main means to attack the control system forward channel,and auxiliary to deceive attack feedback channel,so as to make the system performance worse or even out of control.In order to deal with this kind of attack,the detector is configured at the controller end to detect the deception attack,and the predictive controller is designed to deal with DoS attacks.The stability conditions of the networked predictive control system with multiple attacks are given by the proof method of switching systems.The advantages of multiple attacks over single attacks are compared,and the effectiveness of the proposed method is verified by simulation results.Finally,in the wireless networked control system(WNCSs),due to the exposure of the communication channel to the air,it is vulnerable to malicious interference attacks,which causes packet loss.From interactions between the sender and the sender,the H infinite controller is designed as a resilient defense strategy by a Stackelberg game.Considering the unknown information existing in the actual game,the jammer can not accurately obtain the sender's transmission power,so that they can not get the optimal attack strategy.In this paper,a dynamic updating algorithm is designed from theviewpoints of jammer.The real-time observation is used to control the state of the system to eliminate the observation error,so as to adjust the interference strategy and achieve the optimal.The effectiveness of the design method is verified by the simulation of networked uninterrupted system.