A defense method against zero-dynamic attacks on sampled networked control systems
By proposing a dual-rate sampling control strategy in the sampling networked control system, the invisibility problem of zero-dynamic attacks under delay conditions is solved, effective detection and defense of zero-dynamic attacks is achieved, and the security of the system is improved.
Patent Information
- Application Number
- CN202410497959.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-24
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-04-24
AI Technical Summary
When the sampling networked control system faces zero dynamic attacks, due to the delay, it is difficult to effectively defend the signal detection difficulty of existing technology. In particular, zero dynamic attack signals can be hidden in traditional detectors, resulting in serious threats to system performance and stability. .
A security defense method for dual-rate sampling control is proposed. By establishing a model of a sampling networked control system with delay, analyzing the sufficient and necessary conditions for zero dynamic attacks, and using transfer functions to characterize the relationship between attack signals and output signals, designing dual-rate The sampling control policy is used to transfer unstable sampling zeros to the safe area to ensure system security.
This method can enable zero-dynamic attacks to be detected by traditional detectors, improve the security of the system, avoid stealth attacks caused by delays in signal transmission, and ensure that the system can effectively defend against zero-dynamic attacks.
Smart Images

Figure CN118413364B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of network technology, and in particular to a defense method for zero-dynamic attacks on a sampling networked control system. Background Art
[0002] Sampling network control system has received great attention due to its low cost, high reliability, convenient diagnosis and maintenance, as well as its significant advantages such as less information transmission and high control efficiency in sampling control. Therefore, it is widely used in the industrial field.
[0003] With the growing application of sampling networked control systems, ensuring their safe operation has become a challenging task. Because once the system is attacked by malicious attackers, the stability of the system will change dramatically. At present, malicious network attacks mainly include denial of service attacks and deception attacks. However, there is a very destructive type of attack in deception attacks: zero dynamic attacks. The attack signal of zero dynamic attacks evades the detection of traditional detectors by hiding in the zero space represented by the state space. What's worse, over time, the signal of zero dynamic attacks can be infinite, which can cause catastrophic damage to the performance and stability of the system.
[0004] At present, there are many effective strategies for the security defense of sampled networked control systems. For example, generalized holds are used in sampled networked control systems to neutralize zero dynamics attacks (Journal: Automatica,; Author: Jihan Kim, Juhoon Back and Gyunghoon Park; Publication time: 2020; Article title: Neutralizing zero dynamics attack on sampled-data systems via generalized holds; Page number: 108778); and multi-rate sampling control strategies are used to defend against zero dynamics attacks on sampled networked control systems (Journal: System & Control Letters; Author: Mohammad Naghnaeian and Nabil H. Hirzallah; Publication time: 2019; Article title: Security via multirate control in cyber–physical systems; Page number: 12-18). However, for networked control systems, there will inevitably be delays in the transmission process of signals. Due to the existence of delay, two signals will enter the physical device within one sampling interval. Therefore, this motivates us to further study the security defense method of zero dynamic attack in sampling networked control systems with delay. Summary of the invention
[0005] The purpose of the present invention is to solve the security threat caused by zero dynamic attack to the stability of a sampling networked control system with delay, and to propose a security defense method against zero dynamic attack in a sampling networked control system with delay.
[0006] A method for defending against zero-dynamic attacks on a sampled networked control system comprises the following steps:
[0007] Step 1: Use transfer function to characterize the relationship model between attack signal and output signal in the sampling networked control system with delay under zero dynamic attack;
[0008] Step 2: Propose necessary and sufficient conditions for zero-dynamic attacks to remain invisible in sampling networked control systems with delays;
[0009] Step 3: Establish a dual-rate sampling networked control system model with delay;
[0010] Step 4: Based on the necessary and sufficient conditions of step 2 and the dual-rate sampling networked control model of step 3, a defense method is given.
[0011] The beneficial effects of the present invention are:
[0012] 1. The present invention takes into account the delay in the transmission of signals through the network in a sampled networked control system, and analyzes the necessary and sufficient conditions for zero-dynamic attack to maintain stealth under the condition of delay.
[0013] 2. The dual-rate sampling control security defense method proposed in the present invention enables zero-dynamic attacks to be detected by traditional detectors without the need to add other additional equipment.
[0014] 3. The dual-rate sampling control security defense method proposed in the present invention can improve the security of the system by transferring the unstable sampling zero point to a safe area. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 Flow chart of the method of the present invention.
[0016] Figure 2 Schematic diagram of a sampled networked control system that was attacked.
[0017] Figure 3 Schematic diagram of an automatic voltage regulator under zero dynamic attack.
[0018] Figure 4 The attack signal is designed for the attacker to generate unstable zero points based on the sampling system.
[0019] Figure 5This is the state curve diagram of the system when it is attacked by zero dynamics.
[0020] Figure 6 This is the output diagram of the system under zero dynamic attack when no security defense method is taken.
[0021] Figure 7 This is the output diagram of the system subjected to zero dynamic attack when the dual-rate sampling control security defense method is adopted. DETAILED DESCRIPTION
[0022] The specific implementation of the present invention is described below in conjunction with the above drawings for better understanding by the reader. It should be noted that in the following description, when the detailed description of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.
[0023] Specific implementation method 1: The concept of the present invention is a defense method against zero dynamic attack of a sampling networked control system:
[0024] First, based on the dynamic model of a sampling networked control system with delay, the relationship between the system output and the attack signal when attacked under delay is analyzed and characterized using the transfer function.
[0025] Second, based on the definition and properties of zero-dynamic attacks and using the relationship represented by transfer functions, necessary and sufficient conditions are proposed for zero-dynamic attacks to remain invisible in sampled networked control systems with delays.
[0026] Third, based on the dual-rate sampling control strategy, a dual-rate sampling networked control system model with delay is established, and combined with the necessary and sufficient conditions for zero dynamic attack invisibility, a defense method for dual-rate sampling control to ensure system security is given.
[0027] According to the above concept, a defense method for zero-dynamic attack against a sampling networked control system is specifically described, including the following steps:
[0028] Step 1: Use transfer function to characterize the relationship model between attack signal and output signal in the sampling networked control system with delay under zero dynamic attack;
[0029] Step 2: Propose necessary and sufficient conditions for zero-dynamic attacks to remain invisible in sampling networked control systems with delays;
[0030] Step 3: Establish a model of the sampling networked control system with delayed dual-rate control under zero dynamic attack;
[0031] Step 4: Based on the necessary and sufficient conditions of step 2 and the dual-rate sampling networked control model of step 3, a defense method is given.
[0032] Specific implementation method 2: This implementation method is different from the specific implementation method 1 in that the specific process of using the transfer function in step 1 to characterize the relationship model between the attack signal and the output signal in the sampling networked control system with delay under zero dynamic attack is:
[0033] Consider a sample of continuous devices in a networked control system:
[0034] x(t)=A c x(t)+B c u(t)
[0035] y(t)=C c x(t)
[0036] Where x(t), u(t), and y(t) represent the state, input signal, and output signal of the continuous system, respectively; A c , B c , C c denote the system matrix, input matrix and output matrix of the continuous system respectively;
[0037] And the period of the sampler and the zero-order holder is the same, which is T. Due to the existence of the delay τ, there will be two input signals held by the zero-order holder into the system within one sampling interval, that is:
[0038]
[0039] Where k is a non-negative constant;
[0040] Since the purpose of zero dynamic attack is to make the attack signal injected into the system not affect the system output, we only need to study the zero dynamic attack signal and the output signal affected by the attack, that is:
[0041]
[0042] Where x(k), y(k), and a(k) represent the state of the discrete system, the output signal affected by the attack signal, and the zero dynamic attack signal, respectively; the system matrix is expressed as: A = e AcT , B=B1+B2, C=C c
[0043] Based on the above analysis, the sampling networked control system with delay under zero dynamic attack established by transfer function can be expressed as:
[0044]
[0045] in,
[0046] G1(z)=C(zI-A)-1 B1, G2(z) = C(zI-A) -1 B2.
[0047] The other steps and parameters are the same as those in the first embodiment.
[0048] Specific implementation method 3: This implementation method is different from specific implementation methods 1 or 2 in that: the specific process of the necessary and sufficient conditions for the zero-dynamic attack to remain invisible in the sampling networked control system with delay in step 2 is:
[0049] For sampled networked control systems, due to the delay in signal transmission, multiple attack signals will enter the system within a sampling interval. In this case, the traditional zero-dynamic model is no longer applicable. In order to better study the security of the system, the necessary and sufficient conditions for zero-dynamic attacks to remain invisible in sampled networked control systems with delays are proposed.
[0050] Zero dynamic attack in sampled networked control systems with delays is presented Necessary and sufficient conditions for remaining invisible:
[0051] G1 and G2 have the same zero z0 and there are non-zero parameters and At the same time, the following zero dynamic attack definition variants are met
[0052]
[0053] in, a0 is a non-zero parameter used by the attacker when designing the attack signal.
[0054] Sufficiency is reflected in:
[0055] because So Therefore, a(k), a(k-1), Satisfies formula (11);
[0056] Necessity is reflected in:
[0057] From formula (3), we can get:
[0058]
[0059] Then, the z-transform of a(k) and a(k-1) is:
[0060]
[0061] Therefore, the system output signal affected by the attack signal is:
[0062]
[0063] Therefore, it can be concluded that the attack signal designed according to (11) can evade the detection of traditional detectors and achieve the effect of invisibility; the necessary and sufficient conditions for the attack to remain invisible are that, on the one hand, the sub-transfer functions G1(z) and G2(z) need to have the same zero point z0, and on the other hand, there exists a parameter a0 such that z0 and Satisfies equation (11).
[0064] The other steps and parameters are the same as those in the first or second embodiment.
[0065] Specific implementation method 4: This implementation method is different from any one of the specific implementation methods 1 to 3 in that the specific process of establishing the model of the sampling networked control system with delayed dual-rate control under zero dynamic attack in step 3 is:
[0066] For modeling dual-rate sampling networked control systems:
[0067]
[0068]
[0069] Where g(k)=[a T (k-1)a T (k)] T ,
[0070]
[0071] C=C c .
[0072] The other steps and parameters are the same as those in Specific Embodiments 1 to 3.
[0073] Specific implementation method 5: This implementation method is different from any one of the specific implementation methods 1 to 4 in that: in step 4, based on the necessary and sufficient conditions of step 2, the specific process of the defense method of providing a dual-rate sampling control strategy to ensure system security is as follows:
[0074] According to the dual-rate sampling networked control system established in step 3, it is assumed that there is an unstable zero point z0, and the zero dynamic attack designed according to the unstable zero point It is invisible to the system. Then, according to the necessary and sufficient conditions for the zero-dynamic attack to remain invisible in the delayed sampling networked control system obtained in step 2, we can get:
[0075]
[0076]
[0077] Therefore, we can get:
[0078]
[0079]
[0080]
[0081]
[0082] For equations (12) and (13), multiply the matrix You can get:
[0083]
[0084]
[0085] Likewise, a0 and z0 also satisfy the following equation:
[0086]
[0087]
[0088] Extract the common factor Γ=[C T (CA) T …(CA m-3 ) T ] T have to:
[0089]
[0090]
[0091] Since Γ is a column-full rank matrix, then:
[0092]
[0093]
[0094] Combining formulas (14) and (16) as well as (15) and (17) we obtain:
[0095]
[0096]
[0097] Simplified, we get:
[0098] (IA)B1a0=0
[0099] [(z0I-A)B2+AB-z0B]a0=0
[0100] a0 is a non-zero vector and (IA)B1 is a column-full rank matrix. Since 1 is not an eigenvalue of the matrix A, (IA)B1a0=0 does not hold. Therefore, there is no zero-dynamic attack that can be invisible under the dual-rate sampling control strategy.
[0101] The invention provides a defense method for zero dynamic attack of sampling networked control system by numerical simulation verification, which shows that when the sampling networked control system with delay faces zero dynamic attack damage, the proposed dual-rate sampling control strategy can effectively realize attack detection, thereby ensuring the security of the system. The details are as follows:
[0102] like Figure 2 , the parameters of the automatic voltage regulator are:
[0103] K A =10,τ A =0.1,K E =1τ E =0.4,K G =1,τ G =1,K R =1,τ R =0.05.
[0104] Consider a sampler and holder with a period T = 0.5s for an automatic voltage regulator:
[0105] C d =[0005000], D d =0.
[0106] According to the above discrete system, we can get an unstable discrete zero point z0 = -1.419. Based on this zero point, the attacker can design an attack signal a(k) = (-1.419). k .
[0107] For a sampling networked control system with a delay of τ = 0.05s, a dual-rate sampling control strategy is used to improve the safety of the system. When the sampling rate is twice the holding rate, that is, m = 2, we can obtain:
[0108]
[0109]
[0110] Figure 3 , Figure 4They are the zero-dynamic attack signal designed by the attacker and held by the zero-order holder, and the state curve of the system after the attack. It can be seen from the figure that with the passage of time, the attack signal gradually increases, and the stability of the system changes dramatically. Figure 5 , Figure 6 They are the system output diagrams without taking security defense methods and the system output diagrams with taking dual-rate sampling security strategy. It can be seen from the figure that in the case of no security strategy, malicious zero-dynamic attacks will not be detected by the anomaly detector. However, in the case of taking dual-rate security defense method, the impact of zero-dynamic attacks on system output changes, so that they can be detected by the anomaly detector. This shows that the designed dual-rate security strategy can effectively detect zero-dynamic attacks and improve the security of the system.
[0111] The other steps and parameters are the same as those in Specific Implementation Methods 1 to 5.
[0112] The present invention may also have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.
Claims
1. A defense method against zero-dynamic attacks on sampled networked control systems, Features: The following steps are involved: Step 1: Using transfer function to characterize the relationship model between the attack signal and the output signal affected by the attack in the sampling networked control system with delay under zero dynamic attack; Step 2: Propose necessary and sufficient conditions for zero-dynamic attacks to remain invisible in sampling networked control systems with delays; Step 3: Establish a dual-rate sampling networked control model with delay under zero dynamic attack; Step 4: Based on the necessary and sufficient conditions of step 2 and the dual-rate sampling networked control model of step 3, a defense method is given; The specific process of step one is: Consider a sample of continuous devices in a networked control system: x(t)=A c x(t)+B c u(t) y(t)=C c x(t), Where x(t), u(t), and y(t) represent the state, input signal, and output signal of the continuous system, respectively; A c , B c , C c denote the system matrix, input matrix and output matrix of the continuous system respectively; And the period of the sampler and the zero-order holder is the same, which is T. Due to the existence of the delay τ, there will be two input signals held by the zero-order holder into the system within one sampling interval, that is: Where k is a non-negative constant; Since the purpose of zero dynamic attack is to make the attack signal injected into the system not affect the system output; therefore, the zero dynamic attack signal and the output signal affected by the attack are further studied, namely: Among them, x(k), y(k), a(k) represent the state of the discrete system, the output signal affected by the attack signal, and the zero dynamic attack signal respectively; the system matrix is expressed as: B=B 1 +B 2 , C=C c ; The sampling networked control system with delay under zero dynamic attack established by transfer function is expressed as: in, G 1 (z)=C(zI-A) -1 B 1 ,G 2 (z)=C(zI-A) -1 B 2 ; The specific process of step 2 is as follows: Zero dynamic attack in sampling networked control systems with delays Necessary and sufficient conditions for remaining invisible: G 1 and G 2 have the same zero point z 0 , and there are non-zero parameters and The following zero dynamic attack definition variations are also met: in, a 0 Non-zero parameters used by attackers when designing attack signals; Sufficiency is reflected in: because So Therefore, a(k), a(k-1), Variant (3) that satisfies the definition of zero dynamic attack; Necessity is reflected in: From formula (3), we can get: Then, the z-transform of a(k) and a(k-1) is: Therefore, the system output signal affected by the attack signal is: Therefore, it can be concluded that the attack signal designed according to (3) can evade the detection of traditional detectors and achieve the effect of invisibility; the necessary and sufficient conditions for the attack to remain invisible are that the sub-transfer function G 1 (z) and G 2 (z) have the same zero point z 0 On the other hand, there is a parameter a 0 Make z 0 and Variant (3) that satisfies the definition of zero dynamic attack; The specific process of step three is: Model dual-rate sampling networked control systems: where g(k) = [a T (k - 1) a T (k)] T , B = B 1 + B 2 , C = C c ; The specific process of step 4 is as follows: According to the dual-rate sampling networked control system model established in step 3, it is assumed that the system has an unstable zero point z 0 , and zero dynamic attack based on unstable zero point design For the system, it is invisible; then, according to the necessary and sufficient conditions for the zero-dynamic attack obtained in step 2 to remain invisible in the delayed sampling networked control system, we get: Therefore, we get: For equations (4) and (5), multiply by a matrix get: Likewise, a 0 and z 0 It also satisfies the following formula: Extract the common factor Γ=[C T (CA) T …(CA m-3 ) T ] T have to: Since Γ is a column-full rank matrix, then: Combining formula (6) and (7), we get: Simplified: (I-A)B 1 a 0 =0 [z 0 (I-A)B 2 +AB-z 0 B]a 0 =0 a 0 is a non-zero vector and since 1 is not an eigenvalue of the matrix A, (IA)B 1 is a column full rank moment, so (IA)B 1 a 0 =0 does not hold; there is no zero-dynamic attack that can be invisible in a dual-rate sampling networked control system.
