Elastic Memory Event Triggered Control System and Method for a Handoff System under DoS Attacks

Abstract

The present invention relates to the field of handover system control, and discloses a resilient memory event-triggered control system and method for a handover system under DoS attacks. The system includes a handover system, a sampler, an event trigger with a buffer, a local controller, an edge cloud, and an attack detector. The event trigger with a buffer is used to determine whether the sampled data can be released. The edge cloud consists of a cloud controller with a buffer and a handover identifier. The attack detector is used to detect whether there is a DOS attack. When no DoS attack is detected, the cloud controller is scheduled; when a DoS attack is detected, the detection of the event trigger condition is stopped, and the local controller is scheduled. The disclosed control system and method adopt a resilient memory event-triggered scheme, which can release system data with small errors but large state values, thereby greatly shortening the transient response, saving a large amount of network resources and ensuring good control performance.

Description

Technical Field

[0001] The present invention relates to the field of handover system control, and particularly to an elastic memory event-triggered control system and method for a handover system under DoS attack. Background Art

[0002] With the rapid development of network communication technology, the research on network handover systems has received extensive attention. For example, unmanned ground vehicles, smart grids, and water distribution systems. The network handover system consists of a traditional handover system and network control technology. A unique aspect of the network handover system is the ability to share sampled information among sensors, controllers, and actuators through network channels. However, with the increase in the amount of information collected from network sensors, the network handover system requires a large amount of computing resources to achieve control objectives, which leads to the problem of insufficient performance of network controllers.

[0003] Since the cloud controller is set in the cloud, its openness leads to some malicious attacks on the network channel. Among many types of network attacks, the denial-of-service (DoS) attack is particularly harmful because its initiator does not need to have in-depth knowledge of the handover system. The DoS attack will cut off the connection between the cloud controller and the system, resulting in a slow system convergence speed, severe system chattering, and even instability. In addition, the cloud-controlled handover system also faces the problem of limited communication resources. A suitable event-triggering scheme can effectively reduce the number of data transmissions.

[0004] Nowadays, different triggering technologies have been proposed to address the above problems. For example, there are periodic event triggering; dynamic event triggering; time triggering; adaptive triggering, etc. However, the above triggering technologies all perform triggering judgments based on the error between the current signal and the latest released signal. If the error is large, a new data packet is more likely to be released, otherwise it will not be released. This results in the inability to release data with a small error but a large state value, and the system response process cannot be shortened.

[0005] With the increasing requirements for control performance, people have conducted more in-depth research on how to reduce the convergence time and the amplitude of system chattering. Therefore, on the premise of ensuring the exponential stability of the system, designing an efficient triggering mechanism and controller to improve the control performance of the system is one of the challenging tasks. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides an elastic memory event-triggered control system and method for a handover system under DoS attack, so as to achieve the purpose of saving a large amount of network resources and ensuring good control performance.

[0007] To achieve the above object, the technical solution of the present invention is as follows:

[0008] An elastic memory event-triggered control system for a switching system under DoS attacks, comprising a switching system, a sampler, an event trigger with a buffer, a local controller, an edge cloud, and an attack detector;

[0009] The switching system consists of multiple subsystems and corresponding switching instructions;

[0010] The sampler is used to sample the state of the switching system and transmit the sampled data to the event trigger with a buffer;

[0011] The event trigger with a buffer is used to determine whether the sampled data can be released, and the buffer is used to store the trigger data that meets the event trigger conditions;

[0012] The local controller is used to control the switching system to ensure that the system does not have a large amplitude of chattering;

[0013] The edge cloud consists of a cloud controller with a buffer and a switching identifier; the switching identifier is used to identify the occurrence of switching between different modes of the switching system and the system mode according to the trigger data, so as to select the corresponding cloud controller, and the cloud controller is used to control the switching system to ensure the stability of the system;

[0014] The attack detector is used to detect whether there is a DOS attack. When no DoS attack is detected, it schedules the cloud controller; when a DoS attack is detected, it stops detecting the event trigger conditions and schedules the local controller.

[0015] An elastic memory event-triggered control method for a switching system under DoS attacks, using the control system described above, comprising the following steps:

[0016] Step 1, based on the signal transmission process of the networked closed-loop control system under the elastic memory event-triggered mechanism, establish a switching system dynamics model;

[0017] Step 2, construct a DoS attack model, and construct an elastic memory event-triggered mechanism and a cloud / local controller model based on the DoS attack detection method;

[0018] Step 3, combine the switching system dynamics model, the elastic memory event-triggered mechanism, and the cloud / local controller model to establish a switching system closed-loop control dynamics model;

[0019] Step 4, conduct a stability analysis on the switching system closed-loop control dynamics model to obtain the stability conditions of the switching system closed-loop control dynamics model;

[0020] Step 5, use a solver to solve the stability conditions of the switching system closed-loop control dynamics model to obtain the event-triggered control parameters of the switching system under DoS attacks.

[0021] In the above solution, in step 1, the established switching system dynamic model is as follows:

[0022] ;

[0023] where, is the state vector varying with time t, is the derivative of the state vector varying with time t, is the control input varying with time t; , are the system parameters of the switching system, is the switching instruction.

[0024] In the above solution, in step 2, the established DOS attack model is as follows:

[0025] ;

[0026] where, is the end time of the nth detected DoS attack, , is the occurrence time of the nth DoS attack under ideal conditions, is the duration of the nth DoS attack, is the maximum detection time of the attack detector, represents the detected DoS attack interval, represents the detected DoS attack dormant interval, is the minimum detection time of the attack detector.

[0027] In the above solution, in step 2, the DoS attack detection method is as follows:

[0028] First, record and store the average data traffic value q without DoS attacks; subsequently, during normal operation, compare the observed data traffic value with q, and the comparison result is used as the output of the attack detector, which is also used to turn on or off the event trigger, and its definition is as follows:

[0029] ;

[0030] where, the output of the attack detector indicates the occurrence of a DoS attack; the output of the attack detector indicates that no DoS attack has occurred.

[0031] In the above solution, in step 2, the established elastic memory event trigger mechanism is as follows:

[0032] ;

[0033] When a DoS occurs, the attack detector outputs a signal , and at this time the event trigger mechanism immediately stops working; once it is detected that the DoS attack has ended, the event trigger will force the release of data at the end point of the DoS attack .

[0034] When there is no DoS attack, the attack detector outputs a signal , and if the event trigger condition is satisfied, the latest sampled data can be successfully released;

[0035] Among them, is the sampling error, is transpose, is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of the historical signals is 1, that is , is the system state vector at the latest sampling moment; is the average value of storing system state vectors in the buffer, is transpose, is the th trigger moment system state vector, is the trigger matrix, is the th trigger moment switching instruction, f is the trigger parameter, is the end time of the nth detected DoS attack.

[0036] In the above scheme, in step 2, the established cloud / local controller model is as follows:

[0037] ;

[0038] Among them, is the control input. When there is no DoS attack, the attack detector outputs a signal , and the cloud controller is called, which is the sum value when the historical output signal r takes different values, is the cloud controller gain, is the th trigger moment switching instruction, is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of historical signals is 1, i.e., , is the th trigger moment of the system state vector; when the maximum number of memory items stored in the buffer , the cloud controller is a memoryless controller;

[0039] When a DoS attack occurs, the attack detector outputs a signal , and the local controller is called, is the local controller gain, which is the sum of the local controller gains of p subsystems, is the local controller gain of the i-th subsystem, is the system state vector varying with time t.

[0040] In the above scheme, in step 3, the switched system closed-loop control dynamic model is divided into three states:

[0041] The switched system closed-loop control dynamic model in the asynchronous control state:

[0042] ;

[0043] The switched system closed-loop control dynamic model in the synchronous control state:

[0044] ;

[0045] The switched system closed-loop control dynamic model in the DOS attack state:

[0046] ;

[0047] Among them, is the sampling error, is the state vector varying with time t, is the derivative of the state vector varying with time t, is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of historical signals is 1, i.e., ; is the asynchronous cloud control matrix, , and are the system parameters of the i-th subsystem, is the cloud controller gain of the j-th subsystem when the historical output signal is r; is the synchronous cloud control matrix, , is the cloud controller gain of the i-th subsystem when the historical output signal is r; is the local control matrix, , is the local controller gain, which is the sum of the local controller gains of p subsystems, is the local controller gain of the i-th subsystem.

[0048] In the above scheme, in step 4, the stability condition of the switched system closed-loop control dynamic model is:

[0049] ;

[0050] where, is the asynchronous cloud control algebraic matrix, , the asynchronous cloud control coefficient , the subscript 0 represents the attack detector output signal without DoS attack , is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight of the released data, satisfying that the sum of the weights of the historical signals is 1, that is ; is the cloud controller algebraic gain, is the matrix transpose, , is the cloud controller gain of the j-th subsystem when the historical output signal is r; is the matrix P of the j-th subsystem without DoS attack j0 inverse matrix, f is the event trigger parameter, is the event trigger algebraic matrix, , is the event trigger matrix of the j-th subsystem; is the system parameter A of the i-th subsystem i transpose, is the system parameter B of the i-th subsystem i transpose;

[0051] ;

[0052] where, is the synchronous cloud control algebraic matrix, , the synchronous cloud control coefficient , the subscript 0 represents the attack detector output signal without DoS attack , is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of historical signals is 1, i.e., ; is the algebraic gain of the cloud controller, is the matrix transpose, , is the cloud controller gain of the i-th subsystem when the historical output signal is r; is the matrix P of the i-th subsystem without DoS attack i0 inverse matrix, f is the event-triggering parameter, is the event-triggering algebraic matrix, , is the triggering matrix of the i-th subsystem; is the system parameter A of the i-th subsystem i transpose, is the system parameter B of the i-th subsystem i transpose;

[0053] ;

[0054] Among them, is the local control algebraic matrix, , local control coefficient , the subscript 1 represents the output signal of the attack detector when there is a DoS attack , is the local controller algebraic gain, which is the sum of the local controller algebraic gains of p subsystems, is the local controller algebraic gain of the i-th subsystem, , K i is the local controller gain of the i-th subsystem, is the matrix P of the i-th subsystem when a DoS attack occurs i1 inverse matrix, is the system parameter A of the i-th subsystem i transpose, is the system parameter B of the i-th subsystem i transpose;

[0055] ;

[0056] ;

[0057] ;

[0058] ;

[0059] ;

[0060] Among them, is the inverse matrix of the matrix of the j-th subsystem without DoS attack ; is the inverse matrix of the matrix of the i-th subsystem without DoS attack ; is the inverse matrix of the matrix of the j-th subsystem during DoS attack ; is the inverse matrix of the matrix of the i-th subsystem during DoS attack ; the transition parameter at the system switching point , the transition parameter from the DoS attack interval to the DoS attack dormant interval > 0, the transition parameter from the DoS attack dormant interval to the DoS attack interval > 0, the product of the transition parameters ; the synchronous cloud control coefficient , the asynchronous cloud control coefficient , the local control coefficient , is the maximum asynchronous time, and are DoS attack parameters, is the average dwell time of the switching system.

[0061] In the above solution, in step 5, the event-triggered control parameters of the switching system are as follows:

[0062] Cloud controller gain: ;

[0063] Local controller gain: ;

[0064] Event-triggered matrix: ;

[0065] Among them, is the cloud controller algebraic gain of the i-th subsystem when the historical output signal is r, is the matrix of the i-th subsystem without DoS attack; is the sum of the local controller gains of p subsystems, is the local controller algebraic gain of the i-th subsystem, is the matrix of the i-th subsystem during DoS attack, is the event-triggered algebraic matrix of the i-th subsystem.

[0066] Through the above technical solutions, the elastic memory event-triggered control system and method for a switching system under DoS attack provided by the present invention have the following beneficial effects:

[0067] 1. The edge cloud has low latency and can provide a large amount of computing and storage resources. By migrating the computing to the cloud, the control system can reduce its energy cost. Therefore, the present invention adopts a cloud controller with a buffer. Different from other network controllers, the cloud controller gain is related not only to the number of subsystems but also to the number of stored data;

[0068] 2. The present invention designs a scheduling mechanism for the local controller and the cloud / local controller based on the output signal of the attack detector. When there is no DoS attack, this mechanism schedules the cloud controller; when a DoS attack occurs, it schedules the local controller, thus ensuring better control performance of the system;

[0069] 3. The present invention adopts an elastic memory event-triggering scheme, which can release the system data with small errors but large state values, thereby greatly shortening the transient response. Since the elastic memory event-triggering scheme can adjust the weight coefficient to increase the triggering parameter, the system not only saves a large amount of network resources but also ensures good control performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art.

[0071] Figure 1 Schematic diagram of an elastic memory event-triggering control system for a switched system under DoS attack disclosed in an embodiment of the present invention;

[0072] Figure 2 Maximum asynchronous time disclosed in an embodiment of the present invention Schematic diagram of the switching instruction of the switched system and the controller at this time;

[0073] Figure 3 Schematic diagram showing the variation of the DOS attack time region and the elastic memory event-triggering moment with time under the action of cloud control and local control disclosed in an embodiment of the present invention;

[0074] Figure 4 Schematic diagram showing the variation of the state vector of the switched system with time under the action of cloud control and local control disclosed in an embodiment of the present invention;

[0075] Figure 5 Schematic diagram showing the variation of the DOS attack time region and the elastic memory event-triggering moment with time under the action of cloud control;

[0076] Figure 6 Schematic diagram showing the variation of the state vector of the switched system with time under the action of cloud control. DETAILED DESCRIPTION OF THE INVENTION

[0077] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention.

[0078] The present invention provides an elastic memory event-triggered control system for a switching system under DoS attacks, as Figure 1 shown, including a switching system, a sampler, an event trigger with a buffer, a local controller, an edge cloud, and an attack detector;

[0079] The switching system consists of multiple subsystems and corresponding switching instructions;

[0080] The sampler is used to sample the state of the switching system and transmit the sampled data to the event trigger with a buffer;

[0081] The event trigger with a buffer is used to determine whether the sampled data can be released, and the buffer is used to store the trigger data that meets the event trigger conditions;

[0082] The local controller is used to control the switching system to ensure that the system does not have a large amplitude of chattering;

[0083] The edge cloud consists of a cloud controller with a buffer and a switching identifier; the switching identifier is used to identify the occurrence of switching between different modes of the switching system and the system mode according to the trigger data, so as to select the corresponding cloud controller, and the cloud controller is used to control the switching system to ensure the stability of the system;

[0084] The attack detector is used to detect whether there is a DOS attack. When no DoS attack is detected, it schedules the cloud controller; when a DoS attack is detected, it stops detecting the event trigger conditions and schedules the local controller.

[0085] The present invention also discloses an elastic memory event-triggered control method for a switching system under DoS attacks. Using the above control system, it includes the following steps:

[0086] Step 1, based on the signal transmission process of the networked closed-loop control system under the elastic memory event-triggered mechanism, establish a switching system dynamics model.

[0087] The established switching system dynamics model is as follows:

[0088] ;

[0089] Among them, is the state vector that changes with time t, is the derivative of the state vector that changes with time t, is the control input that changes with time t; , are the system parameters of the switching system, is the switching instruction.

[0090] Step 2: Construct a DoS attack model, and build an elastic memory event trigger mechanism and a cloud / local controller model based on the DoS attack detection method.

[0091] The established DOS attack model is as follows:

[0092] ;

[0093] where, is the end time of the nth detected DoS attack, , is the occurrence time of the nth DoS attack under ideal conditions, is the duration of the nth DoS attack, is the maximum detection time of the attack detector, represents the detected DoS attack interval, represents the detected DoS attack dormant interval, is the minimum detection time of the attack detector.

[0094] The DoS attack detection method is as follows:

[0095] First, record and store the average data traffic q without DoS attacks; subsequently, during normal operation, compare the observed data traffic value with q, and the comparison result is used as the output of the attack detector , which is also used to turn on or off the event trigger, and its definition is as follows:

[0096] ;

[0097] where, the output of the attack detector , indicates the occurrence of a DoS attack; the output of the attack detector , indicates that no DoS attack has occurred.

[0098] The established elastic memory event trigger mechanism is as follows:

[0099] ;

[0100] When a DoS occurs, the attack detector outputs a signal , and at this time the event trigger mechanism immediately stops working; once the end of the DoS attack is detected, the event trigger will force the release of data at the end point of the DoS attack;

[0101] When there is no DoS attack, the attack detector outputs a signal , if the event trigger condition is satisfied, the latest sampled data can be successfully released;

[0102] Among them, is the sampling error, is the transpose of, is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of historical signals is 1, that is , is the system state vector at the latest sampling moment; is stored in the buffer the average value of system state vectors, is the transpose of, is the th trigger moment of the system state vector, is the trigger matrix, is the th trigger moment when the switching instruction, f is the trigger parameter, is the end time of the nth detected DoS attack.

[0103] The established cloud / local controller model is as follows:

[0104] ;

[0105] Among them, is the control input. When there is no DoS attack, the attack detector outputs a signal , and the cloud controller is called, which is the sum value when the historical output signal r takes different values, is the cloud controller gain, is the th trigger moment when the switching instruction, is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of historical signals is 1, that is , is the th trigger moment of the system state vector; when the maximum number of memory items stored in the buffer is, the cloud controller is a memoryless controller;

[0106] When a DoS attack occurs, the attack detector outputs a signal , and calls the local controller , is the local controller gain, which is the sum of the local controller gains of p subsystems, is the local controller gain of the i-th subsystem, is the system state vector that changes with time t.

[0107] Step 3: Combine the switched system dynamics model, the elastic memory event-triggering mechanism, and the cloud / local controller model to establish the switched system closed-loop control dynamics model.

[0108] The switched system closed-loop control dynamics model is divided into three states:

[0109] (1) The switched system closed-loop control dynamics model in the asynchronous control state:

[0110] ;

[0111] (2) The switched system closed-loop control dynamics model in the synchronous control state:

[0112] ;

[0113] (3) The switched system closed-loop control dynamics model in the DOS attack state:

[0114] ;

[0115] Among them, is the sampling error, is the state vector that changes with time t, is the derivative of the state vector that changes with time t, is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight of the released data, and the sum of the weights of the historical signals is 1, that is ; is the asynchronous cloud control matrix, , and are the system parameters of the i-th subsystem, is the cloud controller gain of the j-th subsystem when the historical output signal is r; is the synchronous cloud control matrix, , is the cloud controller gain of the i-th subsystem when the historical output signal is r; is the local control matrix, , is the local controller gain, which is the sum of the local controller gains of p subsystems. is the local controller gain of the i-th subsystem.

[0116] Step 4: Conduct a stability analysis on the closed-loop control dynamic model of the switched system to obtain the stability conditions of the closed-loop control dynamic model of the switched system.

[0117] The stability conditions of the closed-loop control dynamic model of the switched system are:

[0118] ;

[0119] where is the asynchronous cloud control algebraic matrix, , the asynchronous cloud control coefficient , the subscript 0 represents the output signal of the attack detector without a DoS attack , is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of the weights of historical signals is 1, i.e., ; is the cloud controller algebraic gain, is the transpose of the matrix , , is the cloud controller gain of the j-th subsystem when the historical output signal is r; is the matrix P of the j-th subsystem without a DoS attack j0 's inverse matrix, f is the event-triggering parameter, is the event-triggering algebraic matrix, , is the event-triggering matrix of the j-th subsystem; is the transpose of the system parameter A of the i-th subsystem i , is the transpose of the system parameter B of the i-th subsystem i ;

[0120] ;

[0121] where is the synchronous cloud control algebraic matrix, , the synchronous cloud control coefficient , the subscript 0 represents the output signal of the attack detector without a DoS attack , is the historical output signal, is the maximum number of memory items stored in the buffer, is the weight for releasing data, satisfying that the sum of weights of historical signals is 1, that is . is the algebraic gain of the cloud controller, is the transpose of the matrix , , is the cloud controller gain of the i-th subsystem when the historical output signal is r; is the matrix P of the i-th subsystem without DoS attack i0 's inverse matrix, f is the event-triggering parameter, is the event-triggering algebraic matrix, , is the triggering matrix of the i-th subsystem; is the system parameter A of the i-th subsystem i 's transpose, is the system parameter B of the i-th subsystem i 's transpose;

[0122] ;

[0123] Among them, is the local control algebraic matrix, , local control coefficient , the subscript 1 represents the output signal of the attack detector when there is a DoS attack , is the local controller algebraic gain, which is the sum of the local controller algebraic gains of p subsystems, is the local controller algebraic gain of the i-th subsystem, , K i is the local controller gain of the i-th subsystem, is the matrix P of the i-th subsystem when a DoS attack occurs i1 's inverse matrix, is the system parameter A of the i-th subsystem i 's transpose, is the system parameter B of the i-th subsystem i 's transpose;

[0124] ;

[0125] ;

[0126] ;

[0127] ;

[0128] ;

[0129] Among them, is the inverse matrix of the j-th subsystem without DoS attack , is the inverse matrix of the i-th subsystem without DoS attack , is the inverse matrix of the j-th subsystem during DoS attack , is the inverse matrix of the i-th subsystem during DoS attack , the transition parameter of the system switching point , the transition parameter from the DoS attack interval to the DoS attack dormant interval >0, the transition parameter from the DoS attack dormant interval to the DoS attack interval >0, the product of transition parameters ; the synchronous cloud control coefficient , the asynchronous cloud control coefficient , the local control coefficient , is the maximum asynchronous time, and are the DoS attack parameters, is the average dwell time of the switched system

[0130] Step 5, use the solver to solve the stability condition of the switched system closed-loop control dynamics model to obtain the event-triggered control parameters of the switched system under DoS attack

[0131] The event-triggered control parameters of the switched system are as follows

[0132] Cloud controller gain: ;

[0133] Local controller gain: ;

[0134] Event-triggered matrix: ;

[0135] Among them, is the cloud controller algebraic gain of the i-th subsystem when the historical output signal is r, is the matrix of the i-th subsystem without DoS attack; is the sum of the local controller gains of p subsystems, is the local controller algebraic gain of the i-th subsystem, is the matrix of the i-th subsystem during DoS attack, is the event-triggered algebraic matrix of the i-th subsystem

[0136] Simulation comparison:

[0137] In a specific embodiment, a switching system with 2 subsystems is adopted, and the parameter values are as follows:

[0138] , , , ;

[0139] f = 0.27, , , , , , , , , , , h = 0.1s, s, , , = 1.6.

[0140] The stability condition of the closed-loop control dynamic model of the switching system can be solved by LMI to obtain the control parameters.

[0141] Applying an elastic memory event-triggered control method for a switching system under a DOS attack, the control parameters of the closed-loop control dynamic model of the switching system are as follows:

[0142] ;

[0143] .

[0144] Simulating the above control parameters, the results shown in Figures 2 to 4 are obtained.

[0145] To reflect the advantages of the control method of the present invention, the following comparison is made from two major aspects.

[0146] When comparing with the use of a memoryless event-triggered control method, that is, setting the maximum number of stored data in the buffer , the stability condition described in step 4 above is unsolvable. Therefore, using the memoryless event-triggered control method cannot stabilize the system, while from Figure 3 and Figure 4 it can be seen that when using the elastic memory event-triggered control method, a large amount of network resources are saved while ensuring good control performance of the switching system.

[0147] When comparing with only setting the cloud controller without setting the local controller, that is, substituting the local controller gain into the local control algebraic matrix described in step 4 above Next, the control parameters of the closed-loop control dynamic model of the switched system obtained by solving the stability conditions described in step 4 are as follows:

[0148] ;

[0149] .

[0150] Simulating the above control parameters can obtain the results as shown in Figures 5 to 6 . At this time, by comparing Figures 3 to 6 it is found that when setting the local controller, the switched system has a faster convergence speed and a smaller amplitude of system chattering, so that the system has better control performance.

[0151] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for triggering control of elastic memory events of a switching system under DoS attack, characterized in that: The steps include: Step 1, based on the signal transmission process of the networked closed-loop control system under the elastic memory event triggering mechanism, a switching system dynamics model is established; Step 2: Build a DoS attack model, and build an elastic memory event trigger mechanism and a cloud / local controller model based on the DoS attack detection method; Step 3, combining the switching system dynamics model, elastic memory event triggering mechanism and cloud / local controller model to establish a closed-loop control dynamics model of the switching system; Step 4, performing stability analysis on the closed-loop control dynamics model of the switching system to obtain a stability condition of the closed-loop control dynamics model of the switching system; Step 5, using a solver to solve the stability condition of the closed-loop control dynamics model of the switching system, and obtaining event-triggered control parameters of the switching system under DoS attack; In step 2, the elastic memory event triggering mechanism established is as follows: ； When DoS occurs, the attack detector outputs a signal , then the event trigger mechanism stops working immediately; once the DoS attack is detected to be over, the event trigger will be Forced release of data; When there is no DoS attack, the attack detector outputs a signal If the event triggers the condition If true, the latest sampling data can be released successfully; in, is the sampling error, for The transpose of is the historical output signal, is the maximum number of memory items stored in the cache, is the weight of the released data, and the sum of the weights of the historical signals is 1, that is , is the system state vector at the latest sampling moment; Store in cache The average value of the system state vector, for The transpose of For the Trigger moment The system state vector, To trigger the matrix, It is Trigger moment The switching instruction is, f is the trigger parameter, is the end time of the nth DoS attack detected; In step 2, the cloud / local controller model is established as follows: ； in, For control input, when no DoS attack occurs, the attack detector outputs a signal , calling the cloud controller , which is the sum of the historical output signal r when it takes different values, is the cloud controller gain, For the Trigger moment The switching instruction when is the historical output signal, is the maximum number of memory items stored in the cache, is the weight of the released data, and the sum of the weights of the historical signals is 1, that is , For the Trigger moment The system state vector; when the maximum number of memory items stored in the buffer is When, the cloud controller is a memoryless controller; When a DoS attack occurs, the attack detector outputs a signal , calling the local controller , is the local controller gain, which is the sum of the local controller gains of p subsystems, is the local controller gain of the ith subsystem, is the system state vector that changes with time t; In step 3, the closed-loop control dynamics model of the switching system is divided into three states: Closed-loop control dynamics model of switching system under asynchronous control state: ； Closed-loop control dynamics model of switching system under synchronous control state: ； Closed-loop control dynamics model of switching system under DOS attack state: ； in, is the sampling error, is the state vector that changes with time t, is the derivative of the state vector with respect to time t, is the historical output signal, is the maximum number of memory items stored in the cache, is the weight of the released data, and the sum of the weights of the historical signals is 1, that is ; For the asynchronous cloud control matrix, , and is the system parameter of the ith subsystem, is the cloud controller gain of the jth subsystem when the historical output signal is r; To synchronize the cloud control matrix, , is the cloud controller gain of the ith subsystem when the historical output signal is r; is the local control matrix, , is the local controller gain, which is the sum of the local controller gains of p subsystems, is the local controller gain of the ith subsystem.

2. The elastic memory event triggering control method for switching system under DoS attack according to claim 1 is characterized in that: In step 1, the established switching system dynamics model is as follows: ； in, is the state vector that changes with time t, is the derivative of the state vector with respect to time t, is the control input that varies with time t; , is the system parameter of the switching system, It is a switching instruction.

3. The elastic memory event triggering control method for switching system under DoS attack according to claim 1 is characterized in that: In step 2, the DOS attack model established is as follows: ； in, is the end time of the nth DoS attack detected, , is the ideal time when the nth DoS attack occurs, is the duration of the nth DoS attack, is the maximum detection time of the attack detector, Indicates the detected DoS attack interval, Indicates the detected DoS attack sleep interval. It is the minimum detection time of the attack detector.

4. The elastic memory event triggering control method for switching system under DoS attack according to claim 1 is characterized in that: In step 2, the DoS attack detection method is as follows: First, the average data flow rate q when there is no DoS attack is recorded and stored; then, during normal operation, the observed data flow rate value Compare with q, and the comparison result is used as the output of the attack detector , which is also used to turn on or off event triggers, and is defined as follows: ； Among them, the output of the attack detector , indicating that a DoS attack has occurred; the output of the attack detector , indicating that no DoS attack has occurred.

5. The elastic memory event triggering control method for switching system under DoS attack according to claim 1 is characterized in that: In step 4, the stability condition of the closed-loop control dynamics model of the switching system is: ； in, is the asynchronous cloud control algebraic matrix, , asynchronous cloud control coefficient , subscript 0 represents the attack detector output signal when there is no DoS attack , is the historical output signal, is the maximum number of memory items stored in the cache, is the weight of the released data, and the sum of the weights of the historical signals is 1, that is ; is the cloud controller algebraic gain, For the matrix The transpose of , is the cloud controller gain of the jth subsystem when the historical output signal is r; is the matrix P of the jth subsystem when there is no DoS attack j0 The inverse matrix of , f is the event trigger parameter, is the event trigger algebraic matrix, , is the event trigger matrix of the jth subsystem; is the system parameter A of the ith subsystem i The transpose of is the system parameter B of the ith subsystem i The transpose of ； in, is the synchronization cloud control algebraic matrix, , Synchronous cloud control coefficient , subscript 0 represents the attack detector output signal when there is no DoS attack , is the historical output signal, is the maximum number of memory items stored in the cache, is the weight of the released data, and the sum of the weights of the historical signals is 1, that is ; is the cloud controller algebraic gain, For the matrix The transpose of , is the cloud controller gain of the ith subsystem when the historical output signal is r; is the matrix P of the ith subsystem when there is no DoS attack i0 The inverse matrix of , f is the event trigger parameter, is the event trigger algebraic matrix, , is the trigger matrix of the i-th subsystem; is the system parameter A of the ith subsystem i The transpose of is the system parameter B of the ith subsystem i The transpose of ； in, is the local control algebra matrix, , local control coefficient , subscript 1 represents the output signal of the attack detector when there is a DoS attack , is the local controller algebraic gain, which is the sum of the local controller algebraic gains of p subsystems, is the local controller algebraic gain of the ith subsystem, , K i is the local controller gain of the ith subsystem, is the matrix P of the i-th subsystem when a DoS attack occurs i1 The inverse matrix of is the system parameter A of the ith subsystem i The transpose of is the system parameter B of the ith subsystem i The transpose of ； ； ； ； ； in, is the matrix of the jth subsystem when there is no DoS attack The inverse matrix of is the matrix of the ith subsystem when there is no DoS attack The inverse matrix of is the matrix of the jth subsystem when a DoS attack occurs The inverse matrix of is the matrix of the i-th subsystem when a DoS attack occurs The inverse matrix of the system switching point transition parameter , transition parameters from DoS attack interval to DoS attack sleep interval >0, transition parameter from DoS attack sleep interval to DoS attack interval >0, transition parameter product ; Synchronous cloud control coefficient , asynchronous cloud control coefficient , local control coefficient , is the maximum asynchronous time, and is the DoS attack parameter, is the average dwell time of the switching system.

6. The elastic memory event triggering control method for switching system under DoS attack according to claim 1 is characterized in that: In step 5, the event triggering control parameters of the switching system are as follows: Cloud Controller Gains: ; Local controller gain: ; Event trigger matrix: ; in, is the algebraic gain of the cloud controller of the ith subsystem when the historical output signal is r, is the matrix of the i-th subsystem when no DoS attack occurs; is the sum of the local controller gains of the p subsystems, is the local controller algebraic gain of the ith subsystem, is the matrix of the ith subsystem when a DoS attack occurs, is the event trigger algebraic matrix of the ith subsystem.

7. A flexible memory event triggering control system for switching systems under DoS attacks using the method of claim 1, characterized in that: Includes switching system, sampler, event trigger with buffer, local controller, edge cloud and attack detector; The switching system is composed of multiple subsystems and corresponding switching instructions; The sampler is used to sample the state of the switching system and transmit the sampled data to the event trigger with a buffer; The event trigger with a buffer is used to determine whether the sampled data can be released, and the buffer is used to store the trigger data that meets the event trigger condition; The local controller is used to control the switching system to ensure that the system does not have a large amplitude of chattering; The edge cloud is composed of a cloud controller with a buffer and a switching identifier; the switching identifier is used to identify the switching occurrence and system mode between different modes of the switching system according to the trigger data, so as to select the corresponding cloud controller, and the cloud controller is used to control the switching system to ensure the stability of the system; The attack detector is used to detect whether there is a DOS attack. When no DoS attack is detected, the cloud controller is dispatched; when a DoS attack is detected, the detection of event triggering conditions is stopped and the local controller is dispatched.

Images