A method, apparatus, system, device, and medium for evaluating terminal reliability.

Abstract

This application relates to the field of network security technology, and in particular to a method, apparatus, system, device, and medium for evaluating terminal trustworthiness, which enables lightweight and dynamic evaluation of terminal trustworthiness while reducing the transmission parameters between edge servers and central servers. The method includes: identifying multiple target terminals with historical interaction behavior with the terminal to be evaluated; determining the interaction trustworthiness corresponding to each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated; calculating the comprehensive trustworthiness of the terminal to be evaluated based on the interaction trustworthiness of multiple target terminals; and correcting the comprehensive trustworthiness using a penalty factor issued by the central server at the current moment to determine the final trustworthiness of the terminal to be evaluated.

Description

Technical Field

[0001] This application relates to the field of network security technology, and in particular to a method, apparatus, system, device and medium for evaluating terminal trustworthiness. Background Technology

[0002] Today, network technology is developing rapidly, and large-scale distributed systems are becoming increasingly common. The research and development of next-generation internet and future networks pose enormous challenges to network information security. Therefore, research on trusted networks is particularly important.

[0003] Trusted assessment mechanisms are a type of soft security assessment mechanism that addresses security control issues by sensing the dynamic behavior of terminals. Based on tracking user dynamics, it takes necessary preventative measures to prevent attackers from deploying malicious terminals to launch attacks on the system. In existing edge computing environments, considering the limitations of edge devices' memory and computing resources, and the dynamic nature of terminal access and disconnection, a lightweight trusted assessment mechanism has been proposed to ensure overall network security. This mechanism significantly reduces the transmission pressure on communication networks, accelerates security assessment response times, and prevents the leakage of sensitive information, thus achieving excellent security protection.

[0004] Most existing trust assessment mechanisms rely on manually assigning weights to influencing factors to evaluate terminal trustworthiness. This method not only leads to significant evaluation errors due to extreme values ​​within a certain time period, but also results in inaccurate overall trust assessment models due to a lack of adaptability in weight settings. To address these issues, existing technologies have proposed training edge-side models using distributed federated learning and adjusting the parameters of both the edge-side and global models through a two-step upload-download process to maintain parameter consistency. However, this approach involves very frequent parameter transmission, resulting in a large overall network throughput. For models with complex parameters, this can lead to excessively frequent parameter adjustments, causing instability in the model evaluation results. Furthermore, parameter adjustments consume significant transmission space and are not conducive to building lightweight evaluation models. Summary of the Invention

[0005] The purpose of this application is to provide a method, apparatus, system, device, and medium for evaluating terminal trustworthiness, so as to achieve lightweight and dynamic evaluation of terminal trustworthiness while reducing the transmission parameters between edge servers and central servers.

[0006] Firstly, this application provides a method for evaluating terminal trustworthiness, applied to an edge server, comprising:

[0007] Identify multiple target terminals that have historical interaction behavior with the terminal to be evaluated;

[0008] Based on the interaction behavior between each target terminal and the terminal to be evaluated, the interaction credibility of each target terminal is determined.

[0009] The overall credibility of the terminal to be evaluated is calculated based on the interaction credibility of multiple target terminals.

[0010] The overall credibility is corrected using the penalty factor issued by the central server at the current moment to determine the final credibility of the terminal to be evaluated. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and a preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated.

[0011] In one possible implementation, determining the interaction credibility of each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated includes:

[0012] Based on the number of different content data packets and the number of identical content data packets sent between each target terminal and the terminal to be evaluated, a similarity evaluation parameter for the data sent between each target terminal and the terminal to be evaluated is calculated.

[0013] Based on the correlation between the data sent by each target terminal and the data requested by the terminal to be evaluated, a consistency evaluation parameter for the data sent between each target terminal and the terminal to be evaluated is calculated.

[0014] For each target terminal, the similarity evaluation parameters and the consistency evaluation parameters are weighted and summed based on pre-configured weight coefficients to obtain the interaction credibility corresponding to each target terminal.

[0015] In one possible implementation, calculating the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals includes:

[0016] Calculate the average value of the interaction credibility corresponding to the multiple target terminals;

[0017] Based on the deviation between the interaction credibility of each target terminal and the average value, the initial value of the weight coefficient set for each target terminal is corrected to obtain the weight coefficient corresponding to each target terminal.

[0018] The accuracy rate for each target terminal is determined based on the deviation between the interaction credibility of each target terminal and the average value.

[0019] Target terminals with an accuracy rate lower than a preset accuracy rate threshold are removed, and a preset number of target terminals are selected from the remaining target terminals, wherein the preset number is more than half of the total number of target terminals.

[0020] Using the weight coefficients corresponding to each of the preset number of target terminals, the interaction credibility of the preset number of target terminals is weighted and summed to obtain the comprehensive credibility of the terminal to be evaluated.

[0021] Secondly, embodiments of this application provide a method for evaluating terminal trustworthiness, applied to a central server communicating with the edge server provided in the first aspect, comprising:

[0022] Receive the comprehensive trustworthiness and final trustworthiness of the terminal to be evaluated reported by the edge server;

[0023] Using a pre-trained neural network model for evaluating terminal credibility, the re-examination credibility of the terminal to be evaluated is determined based on the overall credibility and the network security level of the network where the edge server is located.

[0024] Calculate the absolute value of the error between the re-examination confidence level and the final confidence level;

[0025] Based on the absolute value of the error, the penalty factor at the current moment is adjusted, and the adjusted penalty factor is sent to the edge server.

[0026] In one possible implementation, adjusting the penalty factor at the current moment based on the absolute value of the error includes:

[0027] When the absolute value of the error is greater than a preset threshold, an adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is increased using the adjustment coefficient.

[0028] When the absolute value of the error is less than a preset threshold, an adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is reduced using the adjustment coefficient.

[0029] Thirdly, embodiments of this application provide a device for evaluating the credibility of a terminal, comprising:

[0030] The determination unit is used to determine multiple target terminals that have historical interaction behaviors with the terminal to be evaluated.

[0031] The first processing unit is used to determine the interaction credibility of each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated.

[0032] The second processing unit is used to calculate the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals.

[0033] The third processing unit is used to correct the overall credibility by using the penalty factor issued by the central server at the current moment, and to determine the final credibility of the terminal to be evaluated. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and the preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated.

[0034] Fourthly, embodiments of this application provide a device for evaluating the credibility of a terminal, comprising:

[0035] The receiving unit is used to receive the comprehensive trustworthiness and final trustworthiness of the terminal to be evaluated reported by the edge server;

[0036] The first processing unit is used to determine the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located, using a pre-trained neural network model for evaluating terminal credibility.

[0037] The second processing unit is used to calculate the absolute value of the error between the review credibility and the final credibility of the terminal to be evaluated reported by the edge server;

[0038] The third processing unit is used to adjust the penalty factor at the current moment based on the absolute value of the error, and send the adjusted penalty factor to the edge server.

[0039] Fifthly, embodiments of this application provide a terminal trustworthiness evaluation system, including an edge server, a central server, and multiple terminals communicatively connected to the edge server, wherein...

[0040] The edge server is used to identify multiple target terminals that have historical interaction behavior with the terminal to be evaluated. Based on the interaction behavior between each target terminal and the terminal to be evaluated, it determines the interaction credibility of each target terminal. Based on the interaction credibility of multiple target terminals, it calculates the comprehensive credibility of the terminal to be evaluated. Using the penalty factor issued by the central server at the current moment, it corrects the comprehensive credibility to determine the final credibility of the terminal to be evaluated. The comprehensive credibility and the final credibility are then reported to the central server. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and a preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated.

[0041] The central server is used to receive the comprehensive credibility and final credibility of the terminal to be evaluated reported by the edge server, and use a pre-trained neural network model for evaluating terminal credibility to determine the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located. The central server calculates the absolute value of the error between the re-examination credibility and the final credibility, adjusts the penalty factor at the current time based on the absolute value of the error, and sends the adjusted penalty factor to the edge server.

[0042] In a sixth aspect, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and when the computer program is executed by the processor, it implements the method described in the first aspect or the second aspect.

[0043] In a seventh aspect, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method described in the first or second aspect.

[0044] The terminal trustworthiness assessment method, apparatus, system, device, and medium provided in this application, when determining the trustworthiness of a terminal to be assessed on the edge server side, firstly identifies multiple target terminals that have historical interaction behavior with the terminal to be assessed. Then, based on the interaction behavior between each target terminal and the terminal to be assessed, the interaction trustworthiness corresponding to each target terminal is determined. Furthermore, based on the interaction trustworthiness of multiple target terminals, the comprehensive trustworthiness of the terminal to be assessed is calculated. The comprehensive trustworthiness is then corrected using a penalty factor issued by the central server at the current moment to determine the final trustworthiness of the terminal to be assessed. Compared with existing technologies, only one parameter, the penalty factor, needs to be transmitted between the edge server and the central server, reducing the number of transmission parameters between them. Simultaneously, given the limitations of cost, energy consumption, and computing resources on the edge server, it cannot calculate terminal trustworthiness very accurately. Correcting terminal trustworthiness through the penalty factor not only reduces the complexity of terminal trustworthiness assessment on the edge server side, achieving lightweight terminal trustworthiness assessment, but also enables dynamic assessment of terminal trustworthiness due to the dynamic adjustment of the penalty factor. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is an application scenario for an optional terminal trustworthiness evaluation method according to an embodiment of this application;

[0047] Figure 2 This is a schematic flowchart illustrating a method for evaluating terminal trustworthiness according to an embodiment of this application;

[0048] Figure 3 This is a schematic diagram of a target terminal that has historical interaction behavior with the terminal to be evaluated, according to an embodiment of this application.

[0049] Figure 4 This is a schematic diagram of a distributed trustworthiness assessment framework according to an embodiment of this application;

[0050] Figure 5 This is a schematic diagram illustrating an embodiment of the present application for calculating the interaction credibility of a target terminal;

[0051] Figure 6 This is a schematic diagram illustrating an embodiment of the present application for calculating the overall credibility of a terminal to be evaluated;

[0052] Figure 7This is a schematic flowchart illustrating another method for evaluating terminal credibility according to an embodiment of this application;

[0053] Figure 8 This is a schematic diagram of a terminal trustworthiness evaluation system according to an embodiment of this application;

[0054] Figure 9 This is a schematic diagram of the structure of a terminal credibility evaluation device according to an embodiment of this application;

[0055] Figure 10 This is a schematic diagram of the structure of another terminal credibility evaluation device according to an embodiment of this application. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0057] It should be noted that the application scenarios described in the following embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0058] To reduce the transmission parameters between edge servers and central servers, and to simultaneously achieve lightweight and dynamic assessment of terminal trustworthiness, this application provides a method, apparatus, system, device, and medium for assessing terminal trustworthiness. On the edge server side, when determining the trustworthiness of a terminal to be assessed, multiple target terminals with historical interaction behavior with the terminal to be assessed are first identified. Then, based on the interaction behavior between each target terminal and the terminal to be assessed, the interaction trustworthiness corresponding to each target terminal is determined. Furthermore, based on the interaction trustworthiness of multiple target terminals, the comprehensive trustworthiness of the terminal to be assessed is calculated. Finally, the current penalty issued by the central server is utilized... The penalty factor corrects the overall credibility and determines the final credibility of the terminal to be evaluated. Compared with existing technologies, only one parameter, the penalty factor, needs to be transmitted between the edge server and the central server, reducing the number of transmission parameters between the edge server and the central server. At the same time, given the limitations of cost, energy consumption, and computing resources on the edge server, it is not possible to calculate the terminal credibility very accurately. By correcting the terminal credibility through the penalty factor, not only can the complexity of terminal credibility evaluation on the edge server side be reduced, achieving lightweight evaluation of terminal credibility, but also dynamic evaluation of terminal credibility can be achieved due to the dynamic adjustment of the penalty factor.

[0059] The following describes, with reference to the accompanying drawings, an application scenario of a terminal trustworthiness evaluation method provided in this application. For example... Figure 1 As shown, it includes an edge server 10, a central server 11, and multiple terminals 12, wherein:

[0060] The edge server 10 communicates with multiple terminals 12 through a communication network, and also communicates with the central server 11 through a communication network. There is interactive behavior between the multiple terminals 12, that is, the multiple terminals 12 can transmit data and exchange information by sending and receiving data packets.

[0061] For any terminal to be evaluated among multiple terminals 12, when determining the credibility of the terminal to be evaluated, firstly, multiple target terminals with historical interaction behavior with the terminal to be evaluated are identified. Then, based on the interaction behavior between each target terminal and the terminal to be evaluated, the interaction credibility corresponding to each target terminal is determined. Then, based on the interaction credibility of multiple target terminals, the comprehensive credibility of the terminal to be evaluated is calculated. The comprehensive credibility is corrected using the penalty factor issued by the central server 11 at the current moment to determine the final credibility of the terminal to be evaluated. The comprehensive credibility and the final credibility of the terminal to be evaluated are reported to the central server 11. The penalty factor at the current moment is obtained by the central server 11 after adjusting the penalty factor at the previous moment based on the relationship between the credibility error and the preset threshold. The credibility error is the error between the final credibility determined by the edge server 10 and the credibility determined by the central server 11 for the previous terminal to be evaluated.

[0062] The central server 11 receives the comprehensive credibility and final credibility of the terminal to be evaluated reported by the edge server 10. Using a pre-trained neural network model for evaluating terminal credibility, it determines the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located. It calculates the absolute value of the error between the re-examination credibility and the final credibility. Based on the absolute value of the error, it adjusts the penalty factor at the current moment and sends the adjusted penalty factor to the edge server 10.

[0063] In one alternative implementation, the communication network is a wired network or a wireless network.

[0064] It should be noted that, Figure 1 The examples shown are merely illustrative. In reality, the number of terminals and servers and the communication methods are not limited. When there are multiple servers, multiple servers can form a blockchain, and the servers are nodes on the blockchain. No specific limitations are made in the embodiments of this application.

[0065] To further illustrate the technical solutions provided in the embodiments of this application, a detailed description is provided below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiments of this application provide method operation steps as shown in the following embodiments or drawings, more or fewer operation steps may be included in the method based on conventional or non-inventive effort. For steps that do not logically have a necessary causal relationship, the execution order of these steps is not limited to the execution order provided in the embodiments of this application. In actual processing or when the device executes, the method may be executed in the order shown in the embodiments or drawings, or in combination.

[0066] On the edge server side Figure 2 A flowchart illustrating a method for evaluating terminal trustworthiness provided in an embodiment of this application is shown. Figure 2 As shown, the method may include the following steps:

[0067] Step 201: Identify multiple target terminals that have historical interaction behaviors with the terminal to be evaluated.

[0068] In one example, such as Figure 3 As shown, if terminals A, B, C, and D all have historical interaction behavior with the terminal to be evaluated, then when calculating the credibility of the terminal to be evaluated, multiple target terminals with historical interaction behavior with the terminal to be evaluated are identified, and terminals A, B, C, and D are identified as target terminals.

[0069] Step 202: Determine the interaction credibility of each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated.

[0070] It should be noted that the overall trustworthiness of an edge network refers to the mutual evaluation of the trustworthiness of interacting parties by various terminal nodes in the network, based on the attributes and environment of the interaction behavior. Terminal trustworthiness is assessed through historical interaction behavior. Therefore, the edge server calculates the overall trustworthiness of the terminal to be evaluated by calculating the interaction trustworthiness of multiple target terminals that have historical interaction behavior with the terminal to be evaluated. This application embodiment uses a distributed trustworthiness evaluation framework to calculate the trustworthiness of the terminal to be evaluated.

[0071] In one example, it is still used Figure 3 The example shown takes the defined target terminals as including terminal A, terminal B, terminal C, and terminal D as an example. Figure 4As shown, when using a distributed credibility assessment framework to calculate the credibility of the terminal to be evaluated, the interaction credibility of terminal A is first determined based on the interaction behavior between terminal A and the terminal to be evaluated. The interaction credibility of terminal B is determined based on the interaction behavior between terminal B and the terminal to be evaluated. The interaction credibility of terminal C is determined based on the interaction behavior between terminal C and the terminal to be evaluated. The interaction credibility of terminal D is determined based on the interaction behavior between terminal D and the terminal to be evaluated. Then, the comprehensive credibility of the terminal to be evaluated is calculated based on the interaction credibility of terminal A, terminal B, terminal C, and terminal D.

[0072] In practical implementation, taking terminal A as an example, when determining the interaction credibility of terminal A based on the interaction behavior between terminal A and the terminal to be evaluated, the similarity and consistency of the data sent between terminal A and the terminal to be evaluated can be used. Specifically, such as... Figure 5 As shown, it includes the following steps:

[0073] Step 2021: Based on the number of different content data packets and the number of identical content data packets sent between terminal A and the terminal to be evaluated, calculate the similarity evaluation parameters of the data sent between terminal A and the terminal to be evaluated.

[0074] In practical applications, the similarity evaluation parameter of the data sent between terminal A and the terminal to be evaluated can be calculated using the following formula (1):

[0075]

[0076] In this context, based on the terminal being evaluated, NRP refers to the number of data packets with different content sent between the terminal being evaluated and terminal A, while RP refers to the number of data packets with duplicate content sent. Deep packet inspection (DPI) technology is typically used. This involves parsing Transmission Control Protocol / Internet Protocol (TCP / IP) packets, combining the context, and precisely locating and determining the values ​​of specific fields according to the protocol's rules. Then, directly comparing the values ​​at these specific locations can determine whether the data packet content is identical.

[0077] Step 2022: Based on the correlation between the data sent by terminal A and the data requested by the terminal to be evaluated, calculate the consistency evaluation parameters of the data sent between terminal A and the terminal to be evaluated.

[0078] In practical applications, the consistency of transmitted data can be used to verify whether a malicious terminal has forged data and transmitted it to an adjacent terminal. If terminal A and the terminal to be evaluated interact, then the data packets sent by terminal A are correlated with the data packets collected by the terminal to be evaluated itself. Based on this, the consistency evaluation parameter of the data transmitted between terminal A and the terminal to be evaluated can be calculated using the following formula (2):

[0079]

[0080] Wherein, CP represents the number of relevant data packets sent between terminal A and the terminal to be evaluated, and NCP represents the number of irrelevant data packets sent between terminal A and the terminal to be evaluated.

[0081] Step 2023: For terminal A, based on the pre-configured weight coefficients, the similarity evaluation parameters and consistency evaluation parameters are weighted and summed to obtain the interaction credibility corresponding to terminal A.

[0082] After obtaining the similarity evaluation parameters and consistency evaluation parameters between terminal A and the terminal to be evaluated, the similarity evaluation parameters and consistency evaluation parameters are weighted and summed based on the pre-configured weight coefficients to obtain the interaction credibility corresponding to terminal A. Specifically, it can be calculated using the following formula 3:

[0083] T i =w1similar + w2relevance (3)

[0084] Among them, T i Here, "a" represents the interaction credibility corresponding to terminal A, "similar" represents the similarity evaluation parameter between terminal A and the terminal to be evaluated, "relevance" represents the consistency evaluation parameter between terminal A and the terminal to be evaluated, and "w1" and "w2" are pre-set weight parameters. For example, both "w1" and "w2" are set to 0.5. They can also be set using the entropy weight method. This application does not limit this setting.

[0085] In this way, the interaction credibility corresponding to terminal B, terminal C, and terminal D can be calculated, and the specific details of each will not be elaborated in the embodiments of this application.

[0086] Step 203: Calculate the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals.

[0087] In practice, when calculating the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals, the following steps are taken: First, the average value of the interaction credibility of the multiple target terminals is calculated. Based on the deviation between the interaction credibility of each target terminal and the average value, the initial value of the weight coefficient set for each target terminal is corrected to obtain the weight coefficient for each target terminal. Then, based on the deviation between the interaction credibility of each target terminal and the average value, the accuracy of each target terminal is determined. Target terminals with an accuracy rate less than a preset accuracy rate threshold are eliminated, and a preset number of target terminals are selected from the remaining target terminals. The preset number is more than half of the total number of target terminals. Finally, the interaction credibility of the preset number of target terminals is weighted and summed using the weight coefficients of each target terminal in the preset number of target terminals to obtain the overall credibility of the terminal to be evaluated.

[0088] In one example, such as Figure 6 As shown, assuming there are K target terminals with historical interaction behavior with the terminal to be evaluated, and the interaction credibility of the K target terminals has been calculated, and assuming that the weight value of each target terminal is 1 / K, then when calculating the comprehensive credibility of the terminal to be evaluated, first calculate the average credibility of the interaction credibility of the K target terminals, then calculate the credibility deviation of each target terminal, that is, the difference between the interaction credibility of each target terminal and the average credibility, and update the weight value of each target terminal using the following formula (4):

[0089] w i = (1 / K)×(1-v) i (4)

[0090] Among them, w i The updated weight value for each target terminal, 1 / K is the original weight value for each target terminal, and v i The credibility deviation for each target terminal is defined by the value of i, which ranges from [1, K].

[0091] Then, the accuracy of each target terminal is calculated based on the credibility deviation corresponding to each target terminal. Specifically, the accuracy of each target terminal is the difference between 1 and the credibility deviation. Combined with a pre-set accuracy threshold, target terminals with an accuracy lower than the preset accuracy threshold are eliminated. In other words, if the accuracy of a target terminal is greater than or equal to the preset accuracy threshold, the interaction credibility of the target terminal is considered reliable, and the target terminal is retained. Conversely, if the accuracy of the target terminal is less than the preset accuracy threshold, the interaction credibility of the target terminal is considered unreliable, and the target terminal is eliminated.

[0092] Finally, from the retained target terminals, M target terminals are randomly selected, where M is greater than K / 2. Then, based on the interaction credibility of the selected M target terminals and the updated weight value of each target terminal, the following formula (5) is used to perform weighted summation to obtain the comprehensive credibility of the terminal to be evaluated.

[0093]

[0094] Among them, T ′ To assess the overall credibility of the terminal to be evaluated, w i The updated weight value T for each target terminal i The interaction credibility corresponding to each target terminal.

[0095] Of course, it should be noted that when randomly selecting M target terminals from the retained target terminals, in order to improve the accuracy of the calculated comprehensive credibility of the terminal to be evaluated, all remaining target terminals may also be selected. This application embodiment does not limit this.

[0096] Step 204: Use the penalty factor issued by the central server at the current moment to correct the overall credibility and determine the final credibility of the terminal to be evaluated. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and the preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated.

[0097] In practice, when using the penalty factor issued by the central server at the current moment to correct the overall credibility, the following formula (6) is used to multiply the overall credibility by the penalty factor to determine the credibility of the terminal to be evaluated.

[0098]

[0099] Among them, T e To assess the credibility of the terminal to be evaluated, T ′ To assess the overall credibility of the terminal to be evaluated, The penalty factor issued by the central server at the current moment.

[0100] The penalty factor limits the credibility of the terminal to be evaluated calculated by the edge server. The penalty factor is related to the edge network environment, and its parameters change when the environment's security level is different. The penalty factor at the current moment is issued by the central server. The central server adjusts the penalty factor from the previous moment based on the relationship between the credibility error and a preset threshold. The credibility error is the difference between the final credibility determined by the edge server for the previous terminal to be evaluated and the re-examination credibility determined by the central server.

[0101] In practical applications, the initial value of the penalty factor can be determined by the following formula (7):

[0102]

[0103] in, τ(t0) represents the initial value of the penalty factor, τ(t0) represents the number of malicious terminals identified by the edge server at the initial moment, and ε represents the environment classification control value. Its initial value is determined based on the actual network security level. As the network security level gradually increases, it will decrease, and conversely, it will gradually increase.

[0104] Of course, it should be noted that in specific implementation, after the edge server calculates the final credibility of the terminal to be evaluated, it needs to upload the comprehensive credibility and final credibility of the terminal to be evaluated to the central server so that the central server can review the credibility of the terminal to be evaluated and adjust the penalty factor at the current moment based on the review results to obtain the penalty factor at the next moment.

[0105] On the central server side, Figure 7 A flowchart illustrating a method for evaluating terminal trustworthiness provided in an embodiment of this application is shown. Figure 7 As shown, the method may include the following steps:

[0106] Step 701: Receive the overall trustworthiness and final trustworthiness of the terminal to be evaluated reported by the edge server.

[0107] Step 702: Using a pre-trained neural network model for evaluating terminal credibility, the credibility of the terminal to be evaluated is determined based on the overall credibility and the network security level of the network where the edge server is located.

[0108] In specific implementation, the pre-trained neural network model used to evaluate the trustworthiness of the terminal is not limited in terms of its network structure and training method in this application embodiment. Its input parameters are the overall trustworthiness of the terminal to be evaluated and the network security level of the network where the edge server is located, and its output parameter is the re-examination trustworthiness of the terminal to be evaluated.

[0109] Step 703: Calculate the absolute value of the error between the re-examination confidence level and the final confidence level.

[0110] In practice, the absolute value of the error between the re-examination credibility and the final credibility can be calculated using the following formula (8):

[0111] error = |T e -T c | (8)

[0112] Where error is the absolute value of the error, T e To determine the final credibility of the terminal to be evaluated, T c To assess the credibility of the terminal to be evaluated.

[0113] Step 704: Based on the absolute value of the error, adjust the penalty factor at the current moment and send the adjusted penalty factor to the edge server.

[0114] In practice, the penalty factor at the current moment is adjusted based on the absolute value of the error, including the following two cases:

[0115] Case 1: When the absolute value of the error is greater than the preset threshold, the adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is increased using the adjustment coefficient.

[0116] In practical applications, when the absolute value of the error is greater than the preset threshold, the penalty factor at the current moment is increased by using the following formulas (9), (10) and (11).

[0117] ε t+1 =ε t -1 (9)

[0118]

[0119]

[0120] Where τ(Δt) represents the number of malicious terminals identified by the edge server during the time interval Δt between the current time and the next time, ε t ε is the environmental classification control value at the current moment. t+1 This will be the environmental classification control value for the next time step. C is the penalty factor for the next time step, C is the adjustment coefficient, the magnitude of which is related to the absolute value of the error, and k is a constant whose value range is (0, 1]. In the embodiments of this application, the value of k can be 1.

[0121] Case 2: When the absolute value of the error is less than the preset threshold, the adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is reduced using the adjustment coefficient.

[0122] In practical applications, when the absolute value of the error is less than the preset threshold, the following formulas (12), (13) and (14) are used to reduce the penalty factor at the current moment.

[0123] ε t+1 =ε t +1 (12)

[0124]

[0125]

[0126] Where τ(Δt) represents the number of malicious terminals identified by the edge server during the time interval Δt between the current time and the next time, ε t ε is the environmental classification control value at the current moment. t+1 This will be the environmental classification control value for the next time step. C is the penalty factor for the next time step, C is the adjustment coefficient, the magnitude of which is related to the absolute value of the error, and k is a constant whose value range is (0, 1]. In the embodiments of this application, the value of k can be 1.

[0127] It should be noted that when the absolute value of the error is equal to the preset threshold, the penalty factor does not need to be processed.

[0128] Based on the same inventive concept, embodiments of this application provide a system for evaluating terminal reliability, such as... Figure 8 As shown, it includes an edge server 80, a central server 81, and multiple terminals that are communicatively connected to the edge server 80.

[0129] Edge server 80 is used to identify multiple target terminals that have historical interaction behavior with the terminal to be evaluated. For example, there are K target terminals, namely terminal 1 to terminal K. Based on the interaction behavior between each target terminal and the terminal to be evaluated, the interaction credibility of each target terminal is determined, and the interaction credibility of terminal 1 to terminal K is obtained. Based on the interaction credibility of multiple target terminals, the comprehensive credibility of the terminal to be evaluated is calculated. The comprehensive credibility is corrected using the penalty factor issued by the central server 81 at the current moment, and the final credibility of the terminal to be evaluated is determined. The comprehensive credibility and the final credibility are reported to the central server 81. The penalty factor at the current moment is obtained by the central server 81 after adjusting the penalty factor at the previous moment based on the relationship between the credibility error and the preset threshold. The credibility error is the error between the final credibility determined by edge server 80 and the review credibility determined by central server 81 for the previous terminal to be evaluated.

[0130] The central server 81 is used to receive the comprehensive credibility and final credibility of the terminal to be evaluated reported by the edge server 80. Using a pre-trained neural network model for evaluating terminal credibility, it determines the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located. It calculates the absolute value of the error between the re-examination credibility and the final credibility, adjusts the penalty factor at the current moment based on the absolute value of the error, and sends the adjusted penalty factor to the edge server 80.

[0131] It should be noted that in other embodiments of this application, the absolute value of the error between the review confidence and the final confidence is calculated, and the penalty factor at the current moment is adjusted based on the absolute value of the error, and the adjusted penalty factor is sent to the edge server 80. This part can also be processed in the control terminal that communicates with the edge server 80 and the central server 81. This application embodiment does not limit this.

[0132] The terminal trustworthiness assessment system provided in this application not only enables dynamic assessment of terminal trustworthiness on the edge server side, but also allows dynamic adjustment of the assessment model or algorithm through the adjustment of the penalty factor. Furthermore, the lightweight dynamic trustworthiness assessment allows the edge server to adjust the assessment model or algorithm for the terminal to be assessed based on the assessment accuracy of multiple terminal trustworthiness assessments. This automatic adaptation to the current network environment and the trust mechanism that adjusts according to environmental changes enhances the system's adaptability to environmental changes. In addition, the lightweight nature is further reflected in the fact that the trustworthiness assessment model on the edge server side does not require frequent parameter adjustments with the assessment model on the central server side; simply updating the penalty factor is sufficient to achieve efficient and consistent assessment of terminal trustworthiness between the system's central server and edge server networks.

[0133] This application embodiment uses a penalty factor to continuously correct the parameters of the edge server-side evaluation model, demonstrating the superiority of lightweight dynamic evaluation of the terminal. At the same time, the initial parameters of the edge server-side evaluation model do not need to be very precise. As the number of iterations increases, the edge server-side evaluation model will become more and more accurate and can change according to changes in the environment.

[0134] This application proposes a two-step model construction method involving coarse-grained and fine-grained steps. Currently, many terminals cannot possess high computing power due to limitations in cost, energy consumption, and computing resources. Considering the limitations of terminal computing resources and energy consumption, this model recognizes that terminals cannot develop highly accurate credibility assessment models based on their own environment or edge servers located near the terminal. Therefore, this application first employs a coarse-grained approach to construct the credibility model parameters for the edge server. It should be noted that this model is an initial credibility model. The edge server constructs the credibility model and derives the final credibility of the terminal to be evaluated. Then, a deep learning model is used to construct the credibility model on the central server side. A fine-grained approach is then used to construct the central server model parameters, calculate the re-verification credibility of the terminal to be evaluated, and calculate the absolute value of the error between the final credibility and the re-verification credibility. A penalty factor is adjusted based on a preset threshold, thereby achieving fine-tuning of the model parameters.

[0135] Based on the same inventive concept, embodiments of this application provide a device for evaluating the reliability of a terminal, such as... Figure 9 As shown, it includes:

[0136] The determination unit 901 is used to determine multiple target terminals that have historical interaction behaviors with the terminal to be evaluated.

[0137] The first processing unit 902 is used to determine the interaction credibility of each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated.

[0138] The second processing unit 903 is used to calculate the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals.

[0139] The third processing unit 904 is used to correct the overall credibility by using the penalty factor issued by the central server at the current moment, and to determine the final credibility of the terminal to be evaluated. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and the preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated.

[0140] In one possible implementation, the first processing unit 902 is specifically used for:

[0141] Based on the number of different content data packets and the number of identical content data packets sent between each target terminal and the terminal to be evaluated, the similarity evaluation parameters of the data sent between each target terminal and the terminal to be evaluated are calculated.

[0142] Based on the correlation between the data sent by each target terminal and the data requested by the terminal to be evaluated, the consistency evaluation parameters of the data sent between each target terminal and the terminal to be evaluated are calculated.

[0143] For each target terminal, the similarity evaluation parameters and consistency evaluation parameters are weighted and summed based on pre-configured weight coefficients to obtain the interaction credibility of each target terminal.

[0144] In one possible implementation, the second processing unit 903 is specifically used for:

[0145] Calculate the average interaction credibility of multiple target terminals;

[0146] Based on the deviation between the interaction credibility of each target terminal and the average value, the initial value of the weight coefficient set for each target terminal is corrected to obtain the weight coefficient corresponding to each target terminal.

[0147] The accuracy rate for each target terminal is determined based on the deviation between the interaction credibility of each target terminal and the average value.

[0148] Remove target terminals with an accuracy rate lower than a preset accuracy threshold, and select a preset number of target terminals from the remaining target terminals. The preset number is more than half of the total number of target terminals.

[0149] By using the weight coefficients corresponding to each of the preset number of target terminals, the interaction credibility of the preset number of target terminals is weighted and summed to obtain the comprehensive credibility of the terminal to be evaluated.

[0150] Based on the same inventive concept, embodiments of this application provide a device for evaluating the reliability of a terminal, such as... Figure 10 As shown, it includes:

[0151] The receiving unit 1001 is used to receive the comprehensive credibility and final credibility of the terminal to be evaluated reported by the edge server;

[0152] The first processing unit 1002 is used to determine the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located, using a pre-trained neural network model for evaluating terminal credibility.

[0153] The second processing unit 1003 is used to calculate the absolute value of the error between the review credibility and the final credibility of the terminal to be evaluated reported by the edge server;

[0154] The third processing unit 1004 is used to adjust the penalty factor at the current moment based on the absolute value of the error, and then send the adjusted penalty factor to the edge server.

[0155] In one possible implementation, the third processing unit 1004 is specifically used for:

[0156] When the absolute value of the error is greater than the preset threshold, the adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is increased using the adjustment coefficient.

[0157] When the absolute value of the error is less than a preset threshold, an adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is reduced using the adjustment coefficient.

[0158] Based on the same inventive concept, this application provides an electronic device, including a memory and a processor. The memory stores a computer program that can run on the processor. When the computer program is executed by the processor, it implements any of the terminal trust evaluation methods in the above embodiments.

[0159] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium that, when the instructions in the storage medium are executed by a processor, enables the processor to perform any of the terminal trust evaluation methods described in the above embodiments.

[0160] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0162] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0163] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0164] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for evaluating terminal trustworthiness, applied to edge servers, characterized in that, include: Identify multiple target terminals that have historical interaction behavior with the terminal to be evaluated; Based on the interaction behavior between each target terminal and the terminal to be evaluated, the interaction credibility of each target terminal is determined. The overall credibility of the terminal to be evaluated is calculated based on the interaction credibility of multiple target terminals. The overall credibility is corrected using the penalty factor issued by the central server at the current moment to determine the final credibility of the terminal to be evaluated. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor from the previous moment based on the relationship between the credibility error and a preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated. The re-examination credibility is determined by the central server, based on the overall credibility reported by the previous terminal to be evaluated, using a pre-trained neural network model for evaluating terminal credibility, and considering the overall credibility reported by the previous terminal to be evaluated and the network security level of the network where the edge server is located.

2. The method according to claim 1, characterized in that, The step of determining the interaction credibility of each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated includes: Based on the number of different content data packets and the number of identical content data packets sent between each target terminal and the terminal to be evaluated, a similarity evaluation parameter for the data sent between each target terminal and the terminal to be evaluated is calculated. Based on the correlation between the data sent by each target terminal and the data requested by the terminal to be evaluated, a consistency evaluation parameter for the data sent between each target terminal and the terminal to be evaluated is calculated. For each target terminal, the similarity evaluation parameters and the consistency evaluation parameters are weighted and summed based on pre-configured weight coefficients to obtain the interaction credibility corresponding to each target terminal.

3. The method according to claim 1, characterized in that, The calculation of the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals includes: Calculate the average value of the interaction credibility corresponding to the multiple target terminals; Based on the deviation between the interaction credibility of each target terminal and the average value, the initial value of the weight coefficient set for each target terminal is corrected to obtain the weight coefficient corresponding to each target terminal. The accuracy rate for each target terminal is determined based on the deviation between the interaction credibility of each target terminal and the average value. Target terminals with an accuracy rate lower than a preset accuracy rate threshold are removed, and a preset number of target terminals are selected from the remaining target terminals, wherein the preset number is more than half of the total number of target terminals. Using the weight coefficients corresponding to each of the preset number of target terminals, the interaction credibility of the preset number of target terminals is weighted and summed to obtain the comprehensive credibility of the terminal to be evaluated.

4. A method for evaluating terminal trustworthiness, applied to a central server, wherein the central server communicates with an edge server as described in any one of claims 1-3, characterized in that, include: Receive the comprehensive trustworthiness and final trustworthiness of the terminal to be evaluated reported by the edge server; Using a pre-trained neural network model for evaluating terminal credibility, the re-examination credibility of the terminal to be evaluated is determined based on the overall credibility and the network security level of the network where the edge server is located. Calculate the absolute value of the error between the re-examination confidence level and the final confidence level; Based on the absolute value of the error, the penalty factor at the current moment is adjusted, and the adjusted penalty factor is sent to the edge server.

5. The method according to claim 4, characterized in that, The adjustment of the penalty factor at the current moment based on the absolute value of the error includes: When the absolute value of the error is greater than a preset threshold, an adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is increased using the adjustment coefficient. When the absolute value of the error is less than a preset threshold, an adjustment coefficient is calculated based on the absolute value of the error, and the penalty factor at the current moment is reduced using the adjustment coefficient.

6. A device for evaluating the credibility of a terminal, characterized in that, include: The determination unit is used to determine multiple target terminals that have historical interaction behaviors with the terminal to be evaluated. The first processing unit is used to determine the interaction credibility of each target terminal based on the interaction behavior between each target terminal and the terminal to be evaluated. The second processing unit is used to calculate the overall credibility of the terminal to be evaluated based on the interaction credibility of multiple target terminals. The third processing unit is used to correct the overall credibility using the penalty factor issued by the central server at the current moment, and determine the final credibility of the terminal to be evaluated. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and a preset threshold. The credibility error is the error between the final credibility determined by the edge server for the previous terminal to be evaluated and the re-examination credibility determined by the central server. The re-examination credibility is determined by the central server based on the overall credibility reported by the previous terminal to be evaluated, using a pre-trained neural network model for evaluating terminal credibility, and the network security level of the network where the edge server is located.

7. A device for evaluating the credibility of a terminal, characterized in that, include: The receiving unit is configured to receive the comprehensive credibility and final credibility of the terminal to be evaluated reported by the edge server as described in any one of claims 1-3; The first processing unit is used to determine the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located, using a pre-trained neural network model for evaluating terminal credibility. The second processing unit is used to calculate the absolute value of the error between the review credibility and the final credibility of the terminal to be evaluated reported by the edge server; The third processing unit is used to adjust the penalty factor at the current moment based on the absolute value of the error, and send the adjusted penalty factor to the edge server.

8. A system for evaluating terminal credibility, characterized in that, It includes an edge server, a central server, and multiple terminals that are communicatively connected to the edge server, wherein, The edge server is used to identify multiple target terminals that have historical interaction behavior with the terminal to be evaluated. Based on the interaction behavior between each target terminal and the terminal to be evaluated, it determines the interaction credibility of each target terminal. Based on the interaction credibility of multiple target terminals, it calculates the comprehensive credibility of the terminal to be evaluated. Using the penalty factor issued by the central server at the current moment, it corrects the comprehensive credibility to determine the final credibility of the terminal to be evaluated. The comprehensive credibility and the final credibility are then reported to the central server. The penalty factor at the current moment is obtained by the central server adjusting the penalty factor at the previous moment based on the relationship between the credibility error and a preset threshold. The credibility error is the error between the final credibility determined by the edge server and the re-examination credibility determined by the central server for the previous terminal to be evaluated. The central server is used to receive the comprehensive credibility and final credibility of the terminal to be evaluated reported by the edge server, and use a pre-trained neural network model for evaluating terminal credibility to determine the re-examination credibility of the terminal to be evaluated based on the comprehensive credibility and the network security level of the network where the edge server is located. The central server calculates the absolute value of the error between the re-examination credibility and the final credibility, adjusts the penalty factor at the current time based on the absolute value of the error, and sends the adjusted penalty factor to the edge server.

9. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and when the computer program is executed by the processor, it implements the method of any one of claims 1 to 5.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it implements the method of any one of claims 1 to 5.

Images