A trust model training method and device
By combining threshold decision and algorithm clustering, the manual labeling of data is reduced, improving the training efficiency and accuracy of the trust model. This solves the problem of high manual costs in existing technologies and achieves high efficiency and robustness in trust assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HUAWEI TECH CO LTD
- Filing Date
- 2021-09-28
- Publication Date
- 2026-05-15
AI Technical Summary
Existing machine learning-based trust models require a large amount of manually labeled data, resulting in high labor costs, low training efficiency, and insufficient accuracy and robustness in trust assessment.
A method combining threshold decision and algorithm clustering is adopted. The label data of network devices is initially determined by threshold conditions, and the clustering algorithm is used to reduce manual labeling operations. The trust model is trained by combining machine learning model.
It reduces the amount of manual labeling, improves the training efficiency and accuracy of the trust model, reduces labor costs, and enhances the robustness of trust assessment.
Smart Images
Figure CN115878991B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the fields of communication technology and artificial intelligence technology, and in particular to a training method and apparatus for a trust model. Background Technology
[0002] With the rapid development of communication technology, and to meet the diverse needs of users, the number of network devices configured in communication networks is increasing. When a network device communicates with other network devices, it may be subject to malicious attacks, rendering it unusable. Alternatively, the network device's own private data may be illegally obtained by other network devices, leading to the leakage of its privacy data.
[0003] To address this, existing technologies typically employ machine learning-based trust models to assess the security of individual network devices, thereby determining their trust levels. Network devices can then decide whether to communicate with other network devices based on their trust levels. However, machine learning-based trust models require extensive training with large amounts of labeled data, which generally needs to be manually labeled. Therefore, acquiring labeled data incurs significant manual costs. Summary of the Invention
[0004] This application provides a training method and apparatus for a trust model, which uses a combination of threshold decision and algorithm clustering to label network devices, thereby reducing the manual cost of obtaining label data for training the trust model.
[0005] Firstly, this application provides a method for training a trust model. The method includes: acquiring communication data from multiple network devices; determining feature data of the multiple network devices based on the communication data; determining label data for each first network device among the multiple network devices based on threshold conditions and the feature data of the multiple network devices; the multiple network devices include at least one first network device and multiple second network devices; the label data indicates the trust level of the network devices; dividing the multiple second network devices into a preset number of clusters, and acquiring the label data corresponding to each cluster as the label data for each second network device in each cluster; the preset number corresponds to the number of trust levels; and updating the parameters of the trust model based on the feature data and label data of the multiple network devices.
[0006] The training samples for the trust model include labeled data from multiple network devices. This labeled data is typically obtained manually by labeling the network devices. When there are many network devices, the cost and time required to acquire this labeled data increase, leading to lower efficiency in training the model.
[0007] The training method for the trust model in this application first uses a threshold condition to determine the label data of the first network device among multiple network devices. Then, the second network devices among the multiple network devices are clustered to obtain cluster groups, and the labels manually assigned to each cluster group are used as the label data of each second network device in the cluster group. This reduces the amount of manual labeling, saves labor costs, and improves the efficiency of training the model.
[0008] In one possible implementation, determining the tag data of each first network device among the multiple network devices based on a threshold condition and the feature data of multiple network devices includes: when the feature data of a network device among the multiple network devices meets the threshold condition, determining the network device as the first network device; and obtaining the tag data corresponding to the threshold condition as the tag data of the first network device.
[0009] Using threshold conditions to determine the label of the first network device can reduce the amount of data processing in subsequent clustering processes.
[0010] In one possible implementation, the threshold conditions include: the feature data is less than a first threshold, and / or the feature data is greater than a second threshold.
[0011] When clustering network devices, the goal is to group devices with similar trust levels into the same cluster, while avoiding clusters with too few devices. The threshold conditions shown above prevent network devices from forming independent clusters due to their feature data being too large or too small, thus affecting the clustering results.
[0012] In one possible implementation, the communication data includes: the number of successful data transmissions and the number of failed data transmissions, and the feature data includes: the data transmission success rate.
[0013] Secondly, this application provides a trust assessment method. The method includes: acquiring communication data from a network device; a feature model determining feature data of the network device based on the communication data; and a trust model determining the trust level of the network device based on the feature data. The trust model is trained using the trust model training method provided in the first aspect of this application.
[0014] In one possible implementation, the trust assessment method further includes storing the trust level of network devices on a blockchain.
[0015] In one possible implementation, the trust assessment method further includes storing the hash value corresponding to the trust level of the network device in a blockchain, and storing the trust level of the network device in a storage system.
[0016] In one possible implementation, the trust assessment method further includes broadcasting the trust level of network devices.
[0017] Thirdly, this application also provides a training apparatus for a trust model. The apparatus includes: an acquisition module, a feature extraction module, a first determination module, a second determination module, and a training module.
[0018] The acquisition module is used to acquire communication data from multiple network devices.
[0019] The feature extraction module is used to determine the feature data of the multiple network devices based on the communication data of the multiple network devices using a feature model.
[0020] The first determining module is used to determine the tag data of each first network device among the multiple network devices based on threshold conditions and characteristic data of multiple network devices; the multiple network devices include at least one first network device and multiple second network devices; the tag data indicates the trust level of the network device.
[0021] The second determining module is used to divide multiple second network devices into a preset number of clusters and obtain the label data corresponding to each cluster group as the label data of the second network devices in each cluster group; the preset number corresponds to the number of trust levels.
[0022] The training module is used to update the parameters of the trust model based on the feature data and label data of multiple network devices.
[0023] In one possible implementation, the first determining module is specifically used to: determine the network device as the first network device when the feature data of the network device among multiple network devices meets the threshold condition; and obtain the tag data corresponding to the threshold condition as the tag data of the first network device.
[0024] In one possible implementation, the threshold conditions include: the feature data is less than a first threshold, and / or the feature data is greater than a second threshold.
[0025] In one possible implementation, the communication data includes: the number of successful data transmissions and the number of failed data transmissions, and the feature data includes: the data transmission success rate.
[0026] Fourthly, this application also provides a trust assessment device. The device includes: an acquisition module, a feature extraction module, and an assessment module.
[0027] The acquisition module is used to acquire communication data from network devices.
[0028] The feature extraction module is used to determine the feature data of the network device based on the communication data of the network device using a feature model.
[0029] The evaluation module is used to determine the trust level of the network device based on the feature data of the network device using a trust model.
[0030] In one possible implementation, the evaluation module is further configured to: store the trust level of the network device in a blockchain; or, store the hash value corresponding to the trust level of the network device in the blockchain and store the trust level of the network device in a storage system; or, broadcast the trust level of the network device.
[0031] Fifthly, this application also provides a computing device. The computing device includes a processor and a memory. The processor is configured to execute a computer program stored in the memory to perform any method of the first aspect of this application and its possible embodiments, or to perform any method of the second aspect of this application and its possible embodiments.
[0032] Sixthly, this application also provides a computer-readable storage medium. The computer-readable storage medium includes instructions that, when executed on a computer, cause the computer to perform any method of the first aspect of this application and any method of the second aspect of this application and any method of the second aspect of this application and any method of the second aspect of this application.
[0033] Sixthly, this application also provides a computer program product. The computer program product includes program code that, when a computer runs the computer program product, causes the computer to perform any method of the first aspect of this application and its possible embodiments, or any method of the second aspect of this application and its possible embodiments.
[0034] Any of the apparatus, computing device, computer-readable storage medium, or computer program product provided above is used to perform the methods provided above. Therefore, the beneficial effects that can be achieved can be referred to the beneficial effects of the corresponding solutions in the corresponding methods provided above, and will not be repeated here. Attached Figure Description
[0035] Figure 1 This is a schematic diagram of a heterogeneous network structure provided in an embodiment of this application;
[0036] Figure 2 This is a flowchart of a trust model training method provided in an embodiment of this application;
[0037] Figure 3 This is a flowchart of a method for trust assessment of network devices provided in an embodiment of this application;
[0038] Figure 4 This is a flowchart of a communication method for a network device provided in an embodiment of this application;
[0039] Figure 5 This is a flowchart of another communication method for a network device provided in an embodiment of this application;
[0040] Figure 6 This is a flowchart of another communication method for a network device provided in an embodiment of this application;
[0041] Figure 7 This is a schematic diagram of the structure of a training device for a trust model provided in an embodiment of this application;
[0042] Figure 8 This is a schematic diagram of the structure of a trust assessment device provided in an embodiment of this application;
[0043] Figure 9 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation
[0044] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be described below with reference to the accompanying drawings.
[0045] In the description of the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.
[0046] In the description of the embodiments in this application, the term "and / or" is merely a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, B existing alone, and A and B existing simultaneously. Furthermore, unless otherwise stated, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals.
[0047] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and their variations all mean "including but not limited to," unless otherwise specifically emphasized.
[0048] To better understand the embodiments of this application, a brief introduction to the communication network will be given below.
[0049] A communication network is a link that physically connects various isolated devices to enable information exchange between people, between people and computers, and between computers, thereby achieving the goals of resource sharing and communication. A communication network includes network devices located on the user side, servers located in the cloud, and switches, routers, and base stations connecting the network devices and servers.
[0050] Communication networks can be networks using the same access technology, such as cellular networks, Internet Protocol (IP) networks, and satellite networks. Network devices can be smartwatches, smartphones, laptops, etc. Communication networks can also be highly integrated heterogeneous networks consisting of two or more single networks using different access technologies, or heterogeneous networks consisting of two or more single networks using the same access technology but belonging to different operators.
[0051] Heterogeneous networks are becoming increasingly important in the communications field due to their unique advantages. Compared to independent and closed communication networks such as cellular networks, IP networks, and satellite networks, heterogeneous networks can expand network coverage and make the network more scalable. Heterogeneous networks can also make full use of existing network resources, reduce operating costs, and enhance competitiveness. They can also provide a variety of services to different users, better meeting the diverse needs of network users. Furthermore, heterogeneous networks can improve network reliability and resistance to attacks.
[0052] Figure 1 This is a schematic diagram of a heterogeneous network structure provided in an embodiment of this application. For example... Figure 1 As shown, this heterogeneous network includes Network 1 and Network 2. Network 1 can be a cellular network, and Network 2 can be an IP network. It is understood that... Figure 1 The structure shown is merely an example of a heterogeneous network in this application embodiment.
[0053] Network 1 and Network 2 include multiple network devices, such as devices 1-3 in Network 1 and devices 4-6 in Network 2. Devices 1-3 in Network 1 can communicate through base stations within it. Devices 4-6 in Network 2 can establish communication through routers within it. Devices in Network 1 can also communicate with devices in Network 2.
[0054] Optionally, devices 1 to 6 can be one of devices such as smartphones, tablets, and laptops.
[0055] Taking the communication between network 3 and device 6 as an example, if device 6 is manipulated to attack device 3, there is a risk that the private data in device 3 will be leaked, or device 3 will become unusable.
[0056] To improve the security of network devices in heterogeneous networks, when a network device needs to communicate with other network devices, it can do so only after verifying that the other network devices are trustworthy.
[0057] In one example, a network device can determine whether other network devices are trustworthy based on their trust levels. Specifically, the trust levels of network devices can be obtained by evaluating them using a trust model. This trust model can be based on mathematical theory or machine learning.
[0058] The method of establishing trust models based on mathematical theory involves using mathematical theory to build a trust model that represents the trust relationship between network devices. This modeling method has two main problems. First, the trust relationship between network devices exists within a specific scenario; therefore, the trust model is highly dependent on the application scenario and has poor transferability. Second, mathematical theory often involves many weighting factors and other parameters when modeling trust relationships, and these parameters are generally determined empirically. This leads to uncertainty in the accuracy of trust assessment by the trust model, resulting in low robustness.
[0059] The method of building trust models based on machine learning involves training a machine learning model using a large number of training samples. This method does not rely on experience to determine the parameters in the model and can use different training samples to train the model for different application scenarios. Therefore, this method is not only adaptable to different application scenarios, but also yields a highly robust trust model. While this method solves the problem of building trust models using mathematical theory, it also faces new challenges. In practical applications, the label data in the training samples often needs to be manually labeled to obtain the feature data. With a large amount of data, the manual cost of labeling increases, leading to an increase in the cost of obtaining label data. It should be understood that the label data of a network device indicates the trust level of that network device.
[0060] To address the high cost of acquiring labeled data when building trust models using machine learning, this application provides a method for training a trust model, which is applied to a model training device. This method combines threshold decision-making and algorithmic clustering to determine the labeled data of network devices. Then, a machine learning model is trained based on the labeled data of the network devices to obtain a trust model for trust assessment of the network devices.
[0061] In one possible scenario, the model training device can be a device located in a different environment. For example, the model training device could be a server located in the cloud, or it could be a local network device. When the model training device is a local network device, it could be specifically such as... Figure 1 Any one of the network devices shown. The following section combines... Figure 2 This application provides a detailed description of the trust model training method provided in its embodiments.
[0062] Figure 2 This is a flowchart of a trust model training method provided in an embodiment of this application.
[0063] like Figure 2 As shown, the training method includes the following steps S201-S205.
[0064] Step S201: Obtain communication data from multiple network devices.
[0065] The model training device can acquire communication data from multiple network devices. These network devices can also send communication data to the model training device after each recording. Optionally, the network devices can be... Figure 1 The network devices shown in the heterogeneous network can also be network devices in other communication networks.
[0066] Optionally, the communication data may include one or more of the following: the number of successful data transmissions and the number of failed data transmissions, the number of successful direct data transmissions and the number of failed direct data transmissions, and the number of successful indirect data transmissions and the number of failed indirect data transmissions. The number of successful data transmissions can be the sum of the number of successful direct data transmissions and the number of successful indirect data transmissions. Figure 1 Taking the communication between device 3 and device 6 as an example, direct transmission means that device 3 sends data directly generated by device 3 to device 6; indirect transmission means that device 3 sends data sent to device 3 by other devices to device 6.
[0067] Step S202: The feature model determines the feature data of multiple network devices based on the communication data of multiple network devices.
[0068] The model training device can input communication data from multiple network devices into the feature model to obtain feature data from multiple network devices.
[0069] The feature model may include a mathematical model. Specifically, the feature model is determined based on the communication data.
[0070] When communication data includes both the number of successful and failed data transmissions, the feature model may include a mathematical model for calculating the success rate and / or a mathematical model for calculating the failure rate. It should be understood that the feature data may include both the success rate and the failure rate of data transmission.
[0071] When the communication data consists of the number of successful direct data transmissions and the number of failed direct data transmissions, the feature model may further include a mathematical model for calculating the success rate of direct data transmissions and / or a mathematical model for calculating the failure rate of direct data transmissions. It should be understood that the feature data may also include the success rate and the failure rate of direct data transmissions.
[0072] When the communication data consists of the number of successful and failed indirect data transmissions, the feature model may further include a mathematical model for calculating the success rate and / or a mathematical model for calculating the failure rate of indirect data transmissions. It should be understood that the feature data may also include the success rate and failure rate of indirect data transmissions.
[0073] Step S203: Based on the threshold conditions and the characteristic data of multiple network devices, determine the tag data of at least one first network device among the multiple network devices.
[0074] The model training device allows for pre-setting threshold conditions and corresponding trust levels. Specifically, threshold conditions can be set based on feature data. Taking trust levels including trusted and untrusted as an example, when the feature data includes the success rate of data transmission, the threshold conditions can be set to two conditions: a success rate greater than 0.8 and a success rate less than 0.3. Furthermore, the trust level corresponding to a success rate greater than 0.8 can be set to trusted, and the trust level corresponding to a success rate less than 0.3 can be set to untrusted.
[0075] The model training device matches the feature data of each network device among multiple network devices with threshold conditions to determine whether the feature data of each network device meets the threshold conditions. When the feature data of a network device meets the threshold conditions, that network device is designated as the first network device, and the trust level corresponding to that threshold condition is used as the label data for the first network device. For example, if the feature data of a network device is less than the minimum value, the label data for that network device is the trust level corresponding to the threshold condition of less than the minimum value, i.e., untrustworthy; if the feature data of a network device is greater than the maximum value, the label data for that network device is the trust level corresponding to the threshold condition of greater than the maximum value, i.e., trustworthy.
[0076] It is understandable that after step S203, the model training device divides the multiple network devices into two main categories: a first network device and a second network device that obtain label data through threshold conditions. The second network device cannot determine label data through threshold conditions. Therefore, the label data of the second network device is determined by the clustering method in step S204.
[0077] This step uses threshold conditions to determine the label data of the first network device. This can prevent the first network device from becoming an independent cluster group due to the large difference between the feature data of the first network device and the feature data of the second network device, thus affecting the final clustering result.
[0078] Step S204: Divide the multiple second network devices into a preset number of clusters, and obtain the label data corresponding to each cluster as the label data of the second network devices in each cluster.
[0079] The model training device can use a clustering algorithm to divide multiple second network devices into a predetermined number of clusters. Each cluster contains at least one second network device. The predetermined number can be determined based on a pre-defined number of trust levels. For example, when the trust levels are determined to be two levels, trusted and untrustworthy, the clustering algorithm can be configured to divide the historical trust judgment values of multiple network devices into two clusters.
[0080] After obtaining cluster groups, the model training device can send each cluster group to the user for manual labeling and obtain the label data corresponding to each cluster group returned by the user. Specifically, the user can judge the feature data corresponding to each network device in the cluster group to determine the label data of that cluster group. The model training device can also use an algorithm to automatically label each cluster group and determine the label data corresponding to each cluster group. Specifically, this application does not specifically limit the method for automatically labeling each cluster group. After determining the label data corresponding to the cluster group, the model training device can use the label data corresponding to the cluster group as the label data for each second network device in that cluster group.
[0081] Alternatively, the clustering algorithm can be any one of the following: K-means clustering, mean-shift clustering, and expectation-maximum clustering.
[0082] Step S205: Update the parameters of the trust model based on the feature data and tag data of multiple network devices.
[0083] After steps S203 and S204, the model training device can obtain the label data of each of the multiple network devices. Therefore, the model training device can use the feature data and label data of the multiple network devices to train the trust model and update its parameters.
[0084] In one example, the model training device can use feature data from multiple network devices as input data to the trust model, obtaining the output data of the trust model. The device can use an error function to calculate the error value between the output data of the trust model and the label data from the multiple network devices, and update the parameters of the trust model using gradient descent based on this error value. When the model training device determines that the error value or the number of training iterations meets preset requirements, it terminates the training of the trust model, using the trust model at the end of training as the final trust model.
[0085] Optionally, the trust model can be one of the machine learning models such as convolutional neural networks, backpropagation neural networks, and deep neural networks, or it can be one of other networks used for machine learning.
[0086] When the trust model is a deep neural network, the work of training each layer in the deep neural network can be expressed mathematically. To describe it. Among them, For the input data of a layer, The input and output data can be represented as vectors, serving as the output data for this layer. In essence, the input to the first layer of a deep neural network is the feature data of the network device, and the output of the last layer is the deep neural network's prediction of the network device's trust level.
[0087] Specifically, from a physical perspective, the work of each layer in a deep neural network can be understood as transforming the input space (the set of input data) to the output space (i.e., from the row space to the column space of a matrix) through five operations on the input space. These five operations include: 1. Dimensionality increase / decrease; 2. Magnification / scaling; 3. Rotation; 4. Translation; 5. "Bending". Operations 1, 2, and 3 are... The operation 4 is completed using +b, and the operation 5 is implemented using a(). The term "space" is used here because the objects being classified are not individual things, but a class of things; space refers to the set of all individuals within this class of things. Here, W is the weight vector, where each value represents the weight of a neuron in that layer of the neural network. This vector W determines the spatial transformation from the input space to the output space, as described above; that is, the weights W of each layer control how the space is transformed. The purpose of training a deep neural network is to ultimately obtain the weight matrix of all layers of the trained neural network (a weight matrix formed by the vectors W from many layers). Therefore, the training process of a neural network is essentially learning how to control spatial transformation, more specifically, learning the weight matrix.
[0088] Because the goal is for the output of a deep neural network to be as close as possible to the actual predicted value, we can compare the network's predicted data with the desired label data and update the weight vector of each layer based on the difference (of course, there is usually an initialization process before the first update, i.e., pre-configuring the parameters of each layer in the deep neural network). For example, if the network's predicted data is too high, the weight vector is adjusted to predict it lower, and this adjustment is continued until the neural network can predict the actual label data. Therefore, it is necessary to predefine "how to compare the difference between the predicted data and the label data," which is the loss function or objective function. These are important equations used to measure the difference between the predicted data and the label data. Taking the loss function as an example, a higher output value (loss) indicates a greater difference, so training a deep neural network becomes a process of minimizing this loss as much as possible.
[0089] In one example, before inputting feature data from multiple network devices into the trust model, the model training device can use the feature model to convert the feature data into data that the trust model can recognize, and then input the data output by the feature model into the trust model. For an introduction to the feature model, please refer to the preceding text. Figure 2 The descriptions in the method embodiments shown are not repeated here.
[0090] In one example, when a network device is in cold start mode, it does not generate characteristic data. The trust assessment device can make a comprehensive assessment based on objective factors such as the network device's manufacturer, scope of use, importance, and deployment location to determine the trust level of the network device.
[0091] Based on the above Figure 2 The embodiment of the trust model training method shown in this application also provides a trust evaluation method for network devices. This method is applied to a trust evaluation device. The trust evaluation device can input the feature data of the network device into the trust model, and then determine the trust level of the network device based on the output of the trust model.
[0092] In one possible scenario, trust assessment devices can be used. Figure 2 The training method for the trust model shown is used to obtain the trust model. In one possible scenario, the trust model in the trust assessment device can be obtained through the model training device; that is, after obtaining the trust model, the model training device sends the model training device to the trust assessment device.
[0093] In one possible scenario, the trust assessment device can be a device located in a different environment. For example, the trust assessment device could be a server located in the cloud, or it could be a local network device. When the trust assessment device is a local network device, it can specifically be as follows: Figure 1 Any one of the network devices shown.
[0094] The following is in conjunction with the appendix Figure 3 This section details how trust assessment equipment uses trust models to perform trust assessments on network devices in heterogeneous networks.
[0095] Figure 3 This is a flowchart of a trust assessment method for network devices provided in an embodiment of this application.
[0096] like Figure 3 As shown, the trust assessment method for this network device specifically includes the following steps S301-S303.
[0097] Step S301: Obtain communication data from the network device.
[0098] The trust assessment device can send a first instruction to the network device. Upon receiving the first instruction, the network device sends its own communication data to the trust assessment device. The network device can also be configured to record the current communication data after each communication session. Optionally, the network device can be... Figure 1 The network devices shown in the heterogeneous network can also be network devices in other communication networks. For a description of the communication data, please refer to the preceding text. Figure 2 The description of step S201 in the method embodiment shown will not be repeated here.
[0099] Step S302: The feature model determines the feature data of the network device based on the communication data of the network device.
[0100] Trust assessment devices can input the communication data of network devices into a pre-established feature model and obtain the feature data of the network device output by the feature model.
[0101] For an introduction to feature data and feature models, please refer to the preceding text. Figure 2 The method described in step S202 in the illustrated embodiment will not be repeated here.
[0102] Step S303: The trust model determines the trust level of the network device based on the characteristic data of the network device.
[0103] Trust assessment devices can input network device characteristic data into a pre-established trust model and determine the trust level of the network device based on the output of the trust model. The process of obtaining the trust model will be described in detail later and will not be repeated here.
[0104] In one example, after obtaining the trust level of a network device, the trust assessment device can store the network device's identifier and the associated trust level in the blockchain. By using blocks to store the trust level, the trust assessment device can ensure that the trust level is fully public and that it cannot be tampered with.
[0105] In one example, after obtaining the trust level of a network device, the trust assessment device can first determine the hash value corresponding to that trust level. Then, the trust assessment device stores the network device's identifier and the hash value corresponding to its trust level in a blockchain, and stores the network device's trust level and its corresponding hash value in a storage system. The storage system could be the InterPlanetary File System (IPS). Storing the hash value corresponding to the trust level on the blockchain can alleviate the data storage pressure on the blockchain.
[0106] In one example, after obtaining the trust level of a network device, the trust evaluation device can also broadcast that network device's trust level. By broadcasting the trust level, the trust evaluation device can save network devices the cost and time of obtaining the trust levels of other network devices, thereby improving communication efficiency.
[0107] The above Figure 3 The trust assessment method shown uses a trust model based on identical communication data across network devices to assess their trust levels, enabling trust assessment of network devices in heterogeneous networks. Specifically, when a primary network device needs to communicate with a secondary network device, it determines the secondary network device's trust level. If the primary network device confirms that the secondary network device is trustworthy, it can then communicate with it, thus improving the security of the primary network device.
[0108] Based on the above Figure 3 The trust assessment method embodiment shown in this application also provides a communication method for a network device. This communication method is applied to a first network device. For example, when the first network device needs to communicate with a second network device, it can obtain the trust level of the second network device and determine whether to communicate with it based on the trust level.
[0109] Figure 4 This is a communication method for a network device provided in an embodiment of this application.
[0110] like Figure 4 As shown, the method includes the following steps S401-S402.
[0111] Step S401: Obtain the trust level of the second network device from the blockchain.
[0112] When a first network device determines that it needs to communicate with a second network device, it can obtain the trust level of the second network device from the blockchain based on the second network device's identifier. The first and second network devices can be... Figure 1 Any device in any subnetwork of the heterogeneous network shown can also be a network device in other communication networks.
[0113] After obtaining the trust level of each network device, the trust assessment device can store the network device's identifier and trust level associated on the blockchain, such as... Figure 4 As shown in step S400.
[0114] Alternatively, the identifier of a network device can be one of the network device's name, IP address, or identity (ID).
[0115] Step S402: Determine whether to communicate with the second network device based on the trust level of the second network device.
[0116] The first network device can determine whether to communicate with the second network device based on the trust level of the second network device and pre-set communication conditions. The first network device can establish a communication connection with the second network device when it determines that the trust level of the second network device meets the communication conditions.
[0117] Specifically, communication conditions can be set according to the trust level category. For example, when trust levels are divided into trusted and untrusted, the communication condition can be that the network device to be establishing a communication connection has a trusted trust level. As another example, when trust levels are divided into extremely untrusted, untrusted, trusted, and very trusted, the communication condition can be that the network device to be establishing a communication connection has a trusted or very trusted trust level. Furthermore, when trust levels are divided into level one, level two, level three, and level four, the communication condition can be that the network device to be establishing a communication connection has a trust level of at least level three.
[0118] Figure 5 This is another communication method for a network device provided in the embodiments of this application.
[0119] like Figure 5 As shown, the method includes the following steps S502-S504.
[0120] Step S502: Obtain the hash value corresponding to the trust level of the second network device from the blockchain.
[0121] When a first network device determines that it needs to communicate with a second network device, it can obtain the hash value corresponding to the trust level of the second network device from the blockchain. Specifically, the first network device can obtain the hash value corresponding to the trust level of the second network device from the blockchain based on the identifier of the second network device.
[0122] After obtaining the trust level of each network device, the trust assessment device can determine the hash value corresponding to the trust level of each network device, and then store the network device's identifier and the hash value corresponding to its trust level on the blockchain, such as... Figure 5 As shown in step S500. Then, the trust evaluation device stores the hash value corresponding to the trust level of the network device and the network device's trust level in the storage system, such as... Figure 5 As shown in step S501.
[0123] Alternatively, the storage system could be the Interplanetary File System (IPFS). IPFS is a media protocol based on blockchain technology that uses distributed storage and content addressing to transform peer-to-peer (P2P) transmission into multi-point-to-multi-point (P2P) transmission. Storing the hash values corresponding to the trust levels of network devices in IPFS can alleviate the pressure on storing data on the blockchain.
[0124] Step S503: Determine the trust level of the second network device based on the hash value corresponding to the trust level of the second network device.
[0125] The first device can obtain the trust level of the second network device from the storage system based on the hash value corresponding to the trust level of the second network device.
[0126] Step S504: Determine whether to communicate with the second network device based on the trust level of the second network device.
[0127] The specific process of this step is the same as described above. Figure 4 The description of step S402 in the method embodiment is the same, and will not be repeated here.
[0128] Figure 6 This is another communication method for a network device provided in the embodiments of this application.
[0129] like Figure 6 As shown, the method includes the following steps S601-S603.
[0130] Step S601: Receive broadcast signal.
[0131] The first network device can be configured to receive broadcast signals in real time. When there is no communication requirement, the first network device can discard the received broadcast signals.
[0132] After obtaining the trust levels of each network device, the trust assessment device broadcasts these trust levels to other network devices via a broadcast signal. Figure 6 As shown in step S600.
[0133] In one example, the trust assessment device can also send the trust levels of each network device to the broadcast device after obtaining those trust levels. Upon receiving the trust levels of each network device, the broadcast device sends a broadcast signal.
[0134] Step S602: Determine the trust level of the second network device based on the broadcast signal.
[0135] When the first network device receives a broadcast signal, it parses the broadcast signal to obtain the identifier and trust level of the network device contained in the broadcast signal. Then, based on the identifier of the second network device, the first network device obtains the trust level of the second network device from the parsed trust level of the network device.
[0136] Step S603: Determine whether to communicate with the second network device based on the trust level of the second network device.
[0137] The specific process of this step is the same as described above. Figure 4 The description of step S402 in the method embodiment is the same, and will not be repeated here.
[0138] based on Figure 2 The embodiment of the trust model training method shown in this application also provides a trust model training device.
[0139] Figure 7 This is a schematic diagram of the structure of a trust model training device 700 provided in an embodiment of this application. The training device 700 is used to implement... Figure 2 Steps S201-S205 in the process. For example... Figure 7 As shown, the training device 700 includes: an acquisition module 701, a feature extraction module 702, a first determination module 703, a second determination module 704, and a training module 705.
[0140] The acquisition module 701 is used to acquire communication data from multiple network devices.
[0141] The feature extraction module 702 is used to determine the feature data of multiple network devices based on the communication data of multiple network devices using a feature model.
[0142] The first determining module 703 is used to determine the tag data of at least one first network device among the multiple network devices based on threshold conditions and feature data of multiple network devices; the multiple network devices include at least one first network device and multiple second network devices; the tag data indicates the trust level of the network device.
[0143] The second determining module 704 is used to divide multiple second network devices into a preset number of clusters and obtain the label data corresponding to each cluster as the label data of the second network devices in each cluster; the preset number corresponds to the number of trust levels.
[0144] The training module 705 is used to update the parameters of the trust model based on the feature data and label data of multiple network devices.
[0145] It should be noted that, Figure 7 The training device 700 provided in the illustrated embodiment, when executing the trust model training method, is only illustrated by the division of the above-described functional modules. In practical applications, the functions performed by each module in the training device 700 can be assigned to other different functional modules as needed, that is, the internal structure of the training device 700 can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the training device 700 provided in the above embodiment and... Figure 2 The trust model training method embodiments shown belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.
[0146] based on Figure 3 The trust assessment method embodiment shown in this application also provides a trust assessment device.
[0147] Figure 8 This is a schematic diagram of the structure of a trust assessment device 800 provided in an embodiment of this application. The trust assessment device 800 is used to implement... Figure 3 Steps S301-S303 in the process. For example... Figure 8 As shown, the trust assessment device 800 includes: an acquisition module 801, a feature extraction module 802, and an assessment module 803.
[0148] The acquisition module 801 is used to acquire communication data from network devices.
[0149] The feature extraction module 802 is used to determine the feature data of the network device based on the communication data of the network device using the feature model.
[0150] The evaluation module 803 is used to determine the trust level of a network device based on the characteristic data of the network device using a trust model.
[0151] It should be noted that, Figure 8 The trust assessment device 800 provided in the illustrated embodiment is only described using the above-described division of functional modules as an example when executing the trust assessment method. In practical applications, the functions performed by each module in the trust assessment device 800 can be assigned to other different functional modules as needed, that is, the internal structure of the trust assessment device 800 can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the trust assessment device 800 provided in the above embodiment and... Figure 3 The trust assessment method embodiments shown belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.
[0152] Figure 9 This is a schematic diagram of the hardware structure of a computing device 900 provided in an embodiment of this application.
[0153] The computing device 900 can be a network device in the aforementioned communication network, the aforementioned trust assessment device, or a model training device. See also Figure 9 The computing device 900 includes a processor 910, a memory 920, a communication interface 930, and a bus 940, which are interconnected via the bus 940. The processor 910, memory 920, and communication interface 930 may also be connected using other methods besides the bus 940.
[0154] The memory 920 can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), flash memory, optical storage, hard disk, etc.
[0155] The processor 910 may be a general-purpose processor, which can be a processor that performs specific steps and / or operations by reading and executing contents stored in memory (e.g., memory 920). For example, the general-purpose processor may be a central processing unit (CPU). The processor 910 may include at least one circuit to perform... Figure 2-6 All or part of the steps of the method provided in the illustrated embodiments.
[0156] The communication interface 930 includes input / output (I / O) interfaces, physical interfaces, and logical interfaces for interconnecting devices within the network device 900, as well as interfaces for interconnecting the network device 900 with other devices (such as other network devices or user equipment). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.
[0157] The bus 940 can be any type of communication bus used to interconnect the processor 910, memory 920 and communication interface 930, such as a system bus.
[0158] The aforementioned devices can be disposed on separate chips, or at least partially or entirely on the same chip. Whether to dispose of the devices independently on different chips or integrate them on one or more chips often depends on the needs of the product design. This application does not limit the specific implementation of the aforementioned devices.
[0159] Figure 9 The computing device 900 shown is merely an example. In the implementation process, the computing device 900 may also include other components, which will not be listed one by one in this article.
[0160] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).
[0161] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application. It should be understood that in the embodiments of this application, the order of the process numbers does not imply the order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.
[0162] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of this application should be included within the scope of protection of this application.
Claims
1. A method for training a trust model, characterized in that, The method includes: Acquire communication data from multiple network devices; The feature model determines the feature data of the multiple network devices based on their communication data. Based on threshold conditions and characteristic data of the plurality of network devices, label data for each first network device among the plurality of network devices is determined; the plurality of network devices includes at least one first network device and a plurality of second network devices; the label data indicates the trust level of the network device; The plurality of second network devices are divided into a preset number of clusters, and the label data corresponding to each cluster is obtained as the label data for each second network device in each cluster; the preset number corresponds to the number of trust levels. The parameters of the trust model are updated based on the feature data and tag data of the multiple network devices.
2. The method according to claim 1, characterized in that, The step of determining the tag data of each first network device among the plurality of network devices based on threshold conditions and feature data of the plurality of network devices includes: When the feature data of a network device among the plurality of network devices meets the threshold condition, the network device is determined to be the first network device; Obtain the tag data corresponding to the threshold condition, and use it as the tag data of the first network device.
3. The method according to claim 1 or 2, characterized in that, The threshold conditions include: the feature data is less than a first threshold, and / or the feature data is greater than a second threshold.
4. The method according to any one of claims 1-3, characterized in that, The communication data includes: the number of successful data transmissions and the number of failed data transmissions; the feature data includes: the data transmission success rate.
5. A trust assessment method, characterized in that, The method includes: Acquire communication data from network devices; The feature model determines the feature data of the network device based on the communication data of the network device; The trust model determines the trust level of the network device based on the feature data of the network device, and the trust model is trained by the method described in any one of claims 1-4.
6. The method according to claim 5, characterized in that, The method further includes: Store the trust level of the network device in a blockchain; or The hash value corresponding to the trust level of the network device is stored in the blockchain, and the trust level of the network device is stored in the storage system; or Broadcast the trust level of the network device.
7. A training device for a trust model, characterized in that, The device includes: The acquisition module is used to acquire communication data from multiple network devices. The feature extraction module is used to determine the feature data of the multiple network devices based on the communication data of the multiple network devices using a feature model; A first determining module is configured to determine the tag data of at least one first network device among the plurality of network devices based on threshold conditions and characteristic data of the plurality of network devices; the plurality of network devices includes the at least one first network device and a plurality of second network devices; the tag data indicates the trust level of the network device; The second determining module is used to divide the plurality of second network devices into a preset number of clusters and obtain the label data corresponding to each cluster group as the label data of the second network devices in each cluster group; the preset number corresponds to the number of trust levels. The training module is used to update the parameters of the trust model based on the feature data and label data of the multiple network devices.
8. The apparatus according to claim 7, characterized in that, The first determining module is specifically used for: When the feature data of a network device among the plurality of network devices meets the threshold condition, the network device is determined to be the first network device; Obtain the tag data corresponding to the threshold condition, and use it as the tag data of the first network device.
9. The apparatus according to claim 7, characterized in that, The threshold conditions include: the feature data is less than a first threshold, and / or the feature data is greater than a second threshold.
10. The apparatus according to any one of claims 7-9, characterized in that, The communication data includes: the number of successful data transmissions and the number of failed data transmissions; the feature data includes: the data transmission success rate.
11. A trust assessment device, characterized in that, The device includes: The acquisition module is used to acquire communication data from network devices; The feature extraction module is used to determine the feature data of the network device based on the communication data of the network device using a feature model; An evaluation module is used to determine the trust level of the network device based on the feature data of the network device using a trust model, wherein the trust model is trained by the method described in any one of claims 1-4.
12. The apparatus according to claim 11, characterized in that, The evaluation module is also used for: Store the trust level of the network device in a blockchain; or The hash value corresponding to the trust level of the network device is stored in the blockchain, and the trust level of the network device is stored in the storage system; or Broadcast the trust level of the network device.
13. A computing device, characterized in that, The computing device includes a processor and a memory, the processor being configured to execute a computer program stored in the memory to perform the method of any one of claims 1 to 4, or to perform the method of claim 5 or 6.
14. A computer-readable storage medium, characterized in that, Includes instructions that, when executed on a computer, cause the computer to perform the method of any one of claims 1 to 4, or the method of claim 5 or 6.
15. A computer program product, characterized in that, Includes program code that, when a computer runs the computer program product, causes the computer to perform the method of any one of claims 1 to 4, or to perform the method of claim 5 or 6.