An AI behavior data-based internet of things device binding method and system

Abstract

The application discloses an AI behavior data-based Internet of Things device binding method and system, relates to the technical field of Internet of Things, and comprises the following steps: defining an Internet of Things device as a node, defining a space-time interaction feature set as a node attribute, obtaining the weight of the edge between nodes by calculating behavior correlation, completing the construction of a behavior correlation topology graph, verifying the behavior consistency of a new device by using a trained GNN through the behavior correlation topology graph, obtaining the legitimacy probability of the new device, setting a trigger threshold, collecting implicit bioelectrical impedance data of the new device when the legitimacy probability of the new device is smaller than the trigger threshold, performing cross-modal feature fusion, generating a user behavior identification vector, and performing final verification to obtain the final verification result of the new device; and the application realizes the mining of deep behavior relations and solves the limitation of static topology structure on dynamic interaction mode mining.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of Internet of Things, and in particular to an Internet of Things device binding method and system based on AI behavior data. BACKGROUND

[0002] As a core link in the fields of smart home, industrial Internet of Things and smart city, the Internet of Things device binding technology aims to ensure that only legitimate devices can access the network and complete binding by verifying the interaction behavior between users and devices. In recent years, with the wide application of Internet of Things devices, the binding technology has gradually evolved from traditional static authentication based on passwords or device identifiers to dynamic verification methods based on user behavior analysis. Early methods relied on single modal data, and gradually developed into comprehensive verification using multi-modal behavior data. Graph neural networks were introduced into device behavior correlation analysis due to their advantages in modeling complex relationships, capturing the spatio-temporal interaction characteristics between devices by constructing behavior correlation topological graphs. In addition, cross-modal feature fusion technology combined with biological signals and behavior data significantly improved the verification accuracy. Generative adversarial networks have also been applied to adversarial behavior detection, training discriminators by generating fake behavior feature vectors to enhance the ability to identify abnormal behavior. These technological advances have promoted the development of Internet of Things device binding from static security to dynamic and intelligent, adapting to diverse application scenarios.

[0003] Existing Internet of Things device binding methods have significant limitations in behavior consistency verification and adversarial behavior detection. For example, traditional methods based on single modal data cannot fully capture the multi-dimensional characteristics of user-device interaction, resulting in insufficient discrimination of behavior verification and being easily bypassed by imitation attacks. Moreover, existing behavior correlation analysis based on graph neural networks can model the relationship between devices, but the evaluation of the behavior consistency of new devices relies on static topological structures, lacking in-depth mining of dynamic interaction patterns, which limits the verification accuracy. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides an Internet of Things device binding method based on AI behavior data to solve the problem of being unable to fully capture the multi-dimensional characteristics of user-device interaction and relying on static topological structures for the evaluation of the behavior consistency of new devices.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides an Internet of Things device binding method based on AI behavior data, which comprises,

[0008] Collect multi-modal behavior data, preprocess and extract features to obtain a set of spatio-temporal interaction features, the multi-modal behavior data including initial interaction data of a new device and interaction data of a bound device;

[0009] Define the Internet of Things device as a node, and the set of spatio-temporal interaction features as node attributes, calculate the weight of the edge between nodes through the behavior correlation, and complete the construction of the behavior correlation topology graph;

[0010] Through the behavior correlation topology graph, use the trained GNN to verify the behavior consistency of the new device, and obtain the legitimacy probability of the new device;

[0011] Set a trigger threshold, when the legitimacy probability of the new device is less than the trigger threshold, collect the implicit bioelectrical impedance data of the new device, perform cross-modal feature fusion to generate a user behavior identification vector, and perform final verification to obtain the final verification result of the new device;

[0012] According to the user behavior identification vector, use the reverse mechanism to detect and strip the false binding request, and finally complete the binding of the Internet of Things device.

[0013] As a preferred scheme of the Internet of Things device binding method based on AI behavior data, wherein: obtaining the set of spatio-temporal interaction features refers to collecting multi-modal behavior data of the bound device and the new device, performing time series alignment and noise removal, then performing feature extraction and standardization processing to arrange the set of spatio-temporal interaction features.

[0014] As a preferred scheme of the Internet of Things device binding method based on AI behavior data, wherein: define the Internet of Things device as a node, and the set of spatio-temporal interaction features as node attributes, calculate the weight of the edge between nodes through the behavior correlation, and complete the construction of the behavior correlation topology graph, specifically including the following steps,

[0015] Each Internet of Things device is represented as a node of the behavior correlation topology graph, and the set of spatio-temporal interaction features is used as a node attribute;

[0016] Quantify the behavior similarity between devices by calculating the Pearson correlation coefficient, and calculate the spatial distance between each pair of Internet of Things devices in the behavior correlation topology graph using Euclidean distance;

[0017] Calculate the edge weight of the behavior correlation topology graph according to the behavior similarity and the spatial distance of each pair of nodes of the behavior correlation topology graph, and complete the construction of the behavior correlation topology graph.

[0018] As a preferred scheme of the AI behavior data-based Internet of Things device binding method, wherein: the behavior consistency of the new device is verified by using the trained GNN through the space-time interaction feature set and the behavior correlation topology graph, and the legality probability of the new device is obtained, and the specific steps include the following steps,

[0019] The historical behavior correlation topology graph is constructed based on historical behavior data, the graph neural network is trained, the weight matrix and the bias matrix of the graph neural network are optimized by supervised learning, and the trained graph neural network is obtained.

[0020] The behavior correlation topology graph is input into the trained graph neural network, the graph neural network aggregates the behavior relationship between devices through graph convolution, evaluates the behavior consistency of the new device and the bound device, and outputs the legality probability of the new device.

[0021] As a preferred scheme of the AI behavior data-based Internet of Things device binding method, wherein: the trigger threshold is set, when the legality probability of the new device is less than the trigger threshold, the implicit bioimpedance data of the new device is collected, the cross-modal feature fusion is performed, and the user behavior identification vector is generated, and the specific steps include the following steps,

[0022] Based on the confidence requirement, the trigger threshold is set, when the legality probability of the new device is less than the trigger threshold, the implicit bioimpedance data and the touch behavior data are collected;

[0023] The multi-modal neural network is trained based on historical real behavior data, the historical implicit bioimpedance data and the historical touch behavior data are input into the multi-modal neural network, the attention weight matrix and the attention bias vector of the multi-modal neural network are optimized in a supervised learning manner, and the trained multi-modal neural network is obtained, and the historical real behavior data includes historical implicit bioimpedance data and historical touch behavior data.

[0024] The implicit bioimpedance data and the touch behavior data are input into the trained multi-modal neural network, and the multi-modal neural network allocates weights through the attention mechanism to obtain the user behavior identification vector.

[0025] As a preferred scheme of the AI behavior data-based Internet of Things device binding method, wherein: the final verification result of the new device is obtained, and the specific steps include the following steps,

[0026] The similarity threshold is set based on the historical user identification vector, and the cosine similarity between the user behavior identification vector and the historical user identification vector is calculated;

[0027] When the cosine similarity is higher than the similarity threshold, it is marked as legal, otherwise it is marked as illegal.

[0028] As a preferred scheme of the AI behavior data-based Internet of Things device binding method, the final completion of the Internet of Things device binding refers to inputting the user behavior identification vector, detecting the abnormality of the user behavior identification vector through the trained adversarial generation network, rejecting the binding and recording the abnormal features if the detection is a false behavior, and completing the binding if the detection passes and the verification result is legal.

[0029] In a second aspect, the application provides an AI behavior data-based Internet of Things device binding system, comprising a collection module, a preprocessing module, a feature extraction module, a construction module, a judgment module, a verification module, and a binding module.

[0030] The construction module defines the Internet of Things device as a node, defines the spatiotemporal interaction feature set as a node attribute, calculates the weight of the edge between the nodes through the behavior correlation, and completes the construction of the behavior correlation topology graph.

[0031] The judgment module verifies the behavior consistency of the new device through the behavior correlation topology graph and the trained GNN, and obtains the legality probability of the new device.

[0032] The verification module sets a trigger threshold, collects the implicit bioelectrical impedance data of the new device when the legality probability of the new device is less than the trigger threshold, performs cross-modal feature fusion to generate a user behavior identification vector, and performs final verification to obtain the final verification result of the new device.

[0033] The binding module detects and strips false binding requests through a reverse mechanism according to the user behavior identification vector, and finally completes the binding of the Internet of Things device.

[0034] In a third aspect, the application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to implement any step of the AI behavior data-based Internet of Things device binding method according to the first aspect of the application.

[0035] In a fourth aspect, the application provides a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement any step of the AI behavior data-based Internet of Things device binding method according to the first aspect of the application.

[0036] The application has the beneficial effects that: by collecting multi-modal behavior data and generating a set of space-time interaction features, multi-dimensional behavior representation of user and device interaction is realized, and high-resolution input is provided for behavior consistency analysis. Compared with traditional single-modal data, multi-modal data enriches the behavior dimension and enhances the behavior distinguishability. Then, through the attention mechanism of the multi-modal neural network, an implicit bioelectric impedance feature vector and a touch feature vector are fused to generate a high-distinguishability user behavior identification vector, and the verification accuracy is further improved. In addition, by constructing a dynamic behavior association topology graph, dynamic modeling of the space-time interaction relationship between devices is realized. By using the trained graph neural network, behavior consistency analysis is performed based on the dynamic behavior association topology graph, deep behavior relationship mining is realized, and the limitation of static topology structure on dynamic interaction mode mining is solved. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Fig. 1 Flowchart of the method for binding the Internet of Things device based on AI behavior data.

[0039] Fig. 2 Schematic diagram of the system for binding the Internet of Things device based on AI behavior data.

[0040] Fig. 3 Schematic diagram of the behavior association topology graph construction process.

[0041] Fig. 4 Schematic diagram of cross-modal feature fusion. DETAILED DESCRIPTION

[0042] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0043] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0044] Second, the "one embodiment" or "an embodiment" referred to herein means a specific feature, structure, characteristic, or combination of features and / or characteristics described herein that can be included in at least one implementation of the present application. The various appearances of "in one embodiment" or "an embodiment" in the specification are not necessarily all referring to the same embodiment.

[0045] Referring to Figs. 1-4 For one embodiment of the present application, the embodiment provides an AI behavior data-based Internet of Things device binding method, comprising the following steps:

[0046] S1, collecting multi-modal behavior data, preprocessing and feature extraction, obtaining spatio-temporal interaction feature set;

[0047] Specifically, the method comprises the following steps,

[0048] S1.1, collect multi-modal behavior data through user-bound devices and new devices, and preprocess, specifically: the smart phone collects sliding trajectory, operation timestamp and geographic position through the touch screen, the sliding trajectory includes x, y coordinate sequence, sliding speed and pressure; the smart watch collects heart rate data, acceleration data and timestamp, wherein the acceleration data reflects wrist movement; the smart light records switch time and brightness adjustment log; the new device collects initial interaction data, including voiceprint features of voice commands and touch operation data, wherein the touch operation data includes click position and click pressure, for example, the new device is a smart speaker.

[0049] Further explanation, through the collection of multi-modal behavior data, covering various interaction modes and biological signals of users and devices, greatly enriching the dimension of behavior information

[0050] All the collected data are encrypted and transmitted to the cloud server through the MQTT protocol for preprocessing, the preprocessing content is divided into time series alignment and noise removal, the time series alignment is to unify the multi-device data to the millisecond time axis based on the linear weighting of adjacent timestamps through linear interpolation, to ensure that the data of different devices are aligned in the time dimension; the noise removal uses median filter to process sliding trajectory data, to eliminate invalid points, and applies low-pass filter to heart rate data and voiceprint data to remove high-frequency noise, after preprocessing, the data are formatted into a unified time series data set, including sliding trajectory, operation timestamp and geographic position of the smart phone, heart rate and acceleration data of the smart watch, switch time and brightness adjustment log of the smart light, and voiceprint features and touch operation data of the new device smart speaker.

[0051] S1.2, according to the pre-processed unified time series data set, extract features, generate the spatio-temporal interaction feature set of each device, specifically: for the sliding track of the smart phone, calculate the curvature, sliding speed mean and pressure mean through the second derivative of the coordinate sequence; for the heart rate data of the smart watch, extract the main frequency and heart rate variance of the heart rate data through fast Fourier transform, and extract the acceleration mean and standard deviation of the acceleration data of the smart watch; for the switch time and brightness adjustment log of the smart lamp, calculate the switch time interval and brightness change frequency; for the voiceprint features of the smart speaker, convert the audio signal to the frequency domain through fast Fourier transform, apply the mel filter bank to extract the log energy, calculate the discrete cosine transform, and generate the mel frequency cepstral coefficient; for touch operation data, calculate the click pressure mean and click interval.

[0052] All the extracted features are integrated into the spatio-temporal interaction feature set after standardization processing, and the spatio-temporal interaction feature set contains time features, space features and interaction features, covering the behavior characteristics of smart phones, smart watches, smart lamps and smart speakers.

[0053] Further explanation, the spatio-temporal interaction feature set comprehensively captures the user behavior pattern and device usage characteristics, providing high-resolution input data for subsequent AI-based behavior consistency verification, significantly improving the accuracy and reliability of the binding process.

[0054] S2, define the Internet of Things device as a node, and the spatio-temporal interaction feature set as a node attribute, calculate the weight of the edge between nodes through the behavior correlation, and complete the construction of the behavior correlation topology graph;

[0055] Specifically, the following steps are included,

[0056] S2.1, define each Internet of Things device as a node of the behavior correlation topology graph, including the bound smart phone, smart watch, smart lamp and new device smart speaker, and the node attribute directly uses the features in the spatio-temporal interaction feature set; the new device is a temporary node, and the node attribute also uses the same feature vector in the spatio-temporal interaction feature set, which is generated based on the initial interaction data.

[0057] S2.2, for each pair of nodes, use the Pearson correlation coefficient to calculate the behavior similarity of the spatio-temporal interaction feature set of each pair of nodes, which is expressed as:

[0058] ;

[0059] Wherein, represents and the Pearson correlation coefficient between them, used to measure the behavior similarity between the spatio-temporal interaction feature sets of and , representing device spatiotemporal interaction feature set, representing device spatiotemporal interaction feature set, representing device spatiotemporal interaction feature set, representing device spatiotemporal interaction feature set.

[0060] According to the operation timestamp in the spatiotemporal interaction feature set of each node, the interaction time difference between device and device is calculated by comparing the operation timestamps of the smart phone, smart watch, smart light and smart speaker, such as sliding time, heart rate recording time, switching time, voice command time, to obtain the absolute time difference.

[0061] According to the geographical position in the spatiotemporal interaction feature set of each node, the spatial distance between device and device is calculated. Here, the Euclidean distance formula is used to solve, and the expression is:

[0062] ;

[0063] wherein, represents the spatial distance between device and device , in meters, represents the coordinate value of the geographical position of device on the x-axis, represents the coordinate value of the geographical position of device on the x-axis, represents the coordinate value of the geographical position of device on the y-axis, represents the coordinate value of the geographical position of device on the y-axis;

[0064] For each pair of nodes, the edge weight of the behavior correlation topology graph is calculated using the NumPy library according to the behavior similarity, interaction time difference and spatial distance of each pair of nodes, ensuring that recent and close-range interactions have higher weights, and the expression is:

[0065] ;

[0066] wherein, represents the edge weight between device and device in the behavior correlation topology graph, represents the weight coefficient of the behavior similarity term, a weight coefficient representing a time decay term, representing a device and the interaction time difference between devices, representing a time decay parameter, representing an exponential function, representing a weight coefficient of a spatial decay term, representing a spatial decay parameter.

[0067] S2.3, using the NetworkX library to generate a directed behavior association topology graph, specifically: the node set of the behavior association topology graph contains smart phones, smart watches, smart lights and smart speakers (temporary nodes), the node attribute is the spatiotemporal interaction feature set, an edge is established for each pair of nodes based on the edge weight, and the behavior association topology graph is generated, representing the spatiotemporal behavior consistency between devices. The behavior association topology graph is stored in the Neo4j database, containing node attributes and edge weights, supporting efficient query and dynamic update.

[0068] Further explanation, by generating the behavior association topology graph, integrating node attributes and edge weights, dynamically capturing the spatiotemporal behavior consistency between devices, through multi-dimensional behavior features (time, space, interaction) and weighted topology structure, the accuracy of behavior consistency analysis in the binding process is significantly improved, and the calculation of edge weight also considers behavior similarity, time difference and spatial distance, enhancing the accuracy of the behavior association topology graph for complex scenarios (such as multi-device collaborative operation), ensuring that the binding process can accurately distinguish between legal and abnormal devices.

[0069] S3, through the behavior association topology graph, using the trained GNN to verify the behavior consistency of the new device, obtaining the legality probability of the new device;

[0070] Specifically, the following steps are included,

[0071] S3.1, based on historical behavior data, constructing a historical behavior association topology graph and training a graph neural network, the historical behavior data has already contained the spatiotemporal interaction feature set and labels of the bound devices (such as smart phones, smart watches, smart lights), the labels are divided into legal and illegal, legal representing real user devices, illegal representing abnormal or unauthorized devices, based on the historical behavior data, constructing a historical behavior association topology graph, the nodes are bound devices, and the node attributes are the spatiotemporal interaction feature set in the historical behavior data.

[0072] GNN (graph neural network) takes the historical behavior association topology graph as input, adopts a two-layer graph convolution structure, and the expression is:

[0073] ;

[0074] wherein, representing a device​ In the node representation of the l-th layer of the graph neural network, the form is a vector, denotes the ReLU activation function applied to the convolution output of the graph neural network for nonlinear transformation, denotes the weight matrix, which is a trainable parameter of the graph neural network, denotes the mean aggregation function, which is used to aggregate the representations of all neighbor nodes of the device in the l-th layer of the graph neural network, denotes the device in the l-th layer of the graph neural network, denotes the device in the l-th layer of the graph neural network, the form is a vector, denotes the neighbor node set of the device in the behavior correlation topology graph, denotes the bias matrix, which is a trainable parameter of the graph neural network, denotes the device in the l-th layer of the graph neural network, the form is a vector.

[0075] For the weight matrix and the bias matrix , training is performed, and the training process optimizes the graph neural network weight matrix and the bias matrix in a supervised learning manner, using the labels of historical behavior data, aiming to maximize the behavior consistency of legal device nodes and minimize the connection weights of illegal nodes. The training is performed on a cloud server using the TensorFlow framework, with a batch size of 32, a learning rate of 0.01, and 100 epochs of training. The trained weight matrix and the bias matrix are saved to obtain the trained GNN.

[0076] S3.2, input the behavior correlation topology graph into the trained GNN, aggregate neighbor node information through two layers of graph convolution, the first layer of convolution aggregates the spatio-temporal interaction feature set of direct neighbors to generate an intermediate representation; the second layer of convolution further aggregates to generate the final node representation, reflecting the deep behavior consistency between devices. Forward propagation is performed on the new device node (smart speaker) to output the legality probability (range 0-1), reflecting the consistency of its behavior with the bound device.

[0077] Further, by capturing the complex spatio-temporal behavior relationship between devices through the graph neural network, the accuracy of the new device behavior consistency verification is significantly improved, ensuring that only high-probability legal devices pass the binding directly, enhancing the security of the binding process. Moreover, the historical behavior association topology graph provides abundant training data, and the real-time behavior association topology graph captures the dynamic characteristics of the current device interaction. The combination of the two enables the graph neural network to adapt to changes in behavior patterns in different scenarios, enhancing the accuracy of the verification. Through the synergistic effect of historical and real-time data, complex or abnormal behaviors are effectively dealt with, and the misjudgment rate is reduced.

[0078] S4, set a trigger threshold, when the legality probability of the new device is less than the trigger threshold, collect the implicit bioimpedance data of the new device, perform cross-modal feature fusion, generate a user behavior identification vector, and perform final verification to obtain the final verification result of the new device;

[0079] Specifically, the method comprises the following steps,

[0080] S4.1, based on the high confidence requirement of behavior consistency verification, set the trigger threshold to 0.9. If the legality probability is higher than the trigger threshold, it is considered that the new device behavior is consistent, and the trigger signal is "no further verification". If the legality probability is lower than 0.9, the trigger signal is "bioimpedance verification is required" to start the subsequent cross-modal fusion.

[0081] S4.2, when the trigger signal is "bioimpedance verification is required", first collect the implicit bioimpedance data and touch behavior data through the capacitive touch sensor on the new device. The specific process is as follows: the capacitive touch sensor measures the impedance fluctuation when the user's finger touches with an alternating current signal, and simultaneously collects touch behavior data, including sliding trajectory, click pressure and timestamp.

[0082] The impedance data is subjected to fast Fourier transform to extract the frequency spectrum features, generating an implicit bioimpedance feature vector, and the touch behavior data is subjected to statistical feature extraction, including sliding speed mean value, trajectory curvature, and click interval, generating a touch feature vector.

[0083] S4.3, train the multi-modal neural network based on historical real behavior data, which includes historical implicit bioimpedance feature vectors and historical touch feature vectors as well as labels. The historical implicit bioimpedance feature vectors and the historical touch feature vectors are spliced into input feature vectors. The structure of the multi-modal neural network includes an input layer, an attention mechanism layer, three layers of fully connected hidden layers, and an output layer. The input layer is used to receive the input feature vector; the attention mechanism layer allocates weights to each dimension of the input feature vector, enhancing the key features; the three layers of fully connected hidden layers each have 256 nodes; and the output layer generates a user behavior identification vector.

[0084] The core of cross-modal fusion is to assign weights to each dimension of the input feature vector using the attention mechanism layer of the multi-modal neural network, and the expression is:

[0085] ;

[0086] wherein, represents the input feature vector processed by the multi-modal neural network attention mechanism, represents the input feature vector, represents the Softmax function, represents the attention weight matrix, represents the attention bias vector.

[0087] The training process is to optimize the attention weight matrix and the attention bias vector of the multi-modal neural network in a supervised learning manner, using the labels of historical real behavior data, with the goal of minimizing the classification loss, the batch size is 32, the learning rate is 0.001, and the training is 150 epochs. The trained attention weight matrix and attention bias vector are saved to complete the training of the multi-modal neural network.

[0088] The real-time implicit bioelectrical impedance feature vector and the touch feature vector are spliced to obtain the real-time input feature vector, and input into the trained multi-modal neural network. The input layer of the trained multi-modal neural network receives the real-time input feature vector, and through the attention mechanism layer with optimized attention weight matrix and attention bias vector, the contribution of key features of the input feature vector is enhanced, and then input into the three-layer fully connected hidden layer for nonlinear transformation and deep feature extraction. Map low-dimensional features to high-dimensional space and gradually reduce dimensionality to capture complex interaction patterns between implicit bioelectrical impedance feature vectors and touch feature vectors, generate more discriminative feature representations, and finally use linear transformation through the output layer to generate user behavior identification vectors reflecting the comprehensive behavior characteristics of the new device.

[0089] S4.4, compare the obtained user behavior identification vector with the historical user identification vector stored in the cloud to generate the final verification result of the new device. The similarity comparison adopts the cosine similarity method to measure, and the expression is:

[0090] ;

[0091] wherein, represents the cosine similarity between the user behavior identification vector and the historical user identification vector stored in the cloud, represents the user behavior identification vector, represents the historical user identification vector.

[0092] The similarity threshold is set based on the statistical characteristics of the historical user identification vector. If the cosine similarity is higher than the similarity threshold, it is considered that the behavior of the new device is highly consistent with the historical user behavior, and the final verification result of the new device is marked as "legal". Otherwise, it is marked as "illegal", for example, the similarity threshold is set to 0.95. Only when the cosine similarity is higher than 0.95, it is considered that the behavior of the new device is highly consistent with the historical user behavior.

[0093] Further, the implicit bioimpedance feature vector and the touch feature vector are used to realize robust behavior verification, significantly reducing the risk of unauthorized device binding and enhancing security, and the attention mechanism layer of the multi-modal neural network is used to fuse the implicit bioimpedance feature vector and the touch feature vector, thereby generating a highly differentiated user behavior identifier vector. This cross-modal method improves the robustness to behavior changes or attacks, ensuring reliable verification in different IoT scenarios.

[0094] S5、According to the user behavior identification vector and the final verification result, a reverse mechanism is used to detect and strip false binding requests, and finally complete the Internet of Things device binding.

[0095] Specifically, the following steps are included,

[0096] S5.1, based on historical real behavior data and user behavior identification vector and final verification result, training generative adversarial network, generative adversarial network contains generator and discriminator, generator takes random noise as input, generates fake behavior feature vector through three-layer fully connected network, fake behavior feature vector dimension matches user behavior identification vector format; the discriminator takes the user behavior identification vector as input, and outputs the abnormal probability through a four-layer fully connected network, the range is 0-1.

[0097] The training process uses historical real behavior data and fake behavior feature vectors generated by the generator to optimize the loss function of the generative adversarial network. The training is performed on a cloud server using the TensorFlow framework, with a batch size of 64, a learning rate of 0.001, and a training of 200 epochs. The trained generator and discriminator parameters are stored in the PostgreSQL database.

[0098] S5.2, the discriminator takes the user behavior identification vector as input, and calculates the abnormal probability through a four-layer fully connected network, the first three layers use ReLU activation function, and the last layer uses Sigmoid function. The abnormal probability represents the possibility of the user behavior identification vector being a fake behavior, and also reflects the deviation degree of the user behavior identification vector from the historical real behavior data.

[0099] By setting an abnormal threshold value through experience, if the abnormal probability is higher than the abnormal threshold value, for example, the abnormal threshold value is 0.3, the abnormal characteristics are analyzed, for example, the micro-jitter frequency of the sliding track, the proportion of high-frequency Fourier components is calculated by fast Fourier transform, the unnatural fluctuation in the counterfeit behavior is reflected, and the activation value of the discriminator layer 2 full connection is extracted through the reverse mechanism to construct an abnormal characteristic space, if the abnormal probability is lower than the abnormal threshold value, the user behavior identification vector is directly marked as no further analysis, the reverse mechanism processing is skipped, and the subsequent decision is entered.

[0100] Example: Based on the activation value of the output of the discriminator layer 2 full connection, combined with the micro-jitter frequency of the sliding track, an abnormal feature vector is generated, including the activation value and the micro-jitter frequency feature, and the abnormal feature vector is organized into an abnormal feature space in the cloud server.

[0101] Based on the output of the abnormal feature space, K-means clustering is used to separate real and false behaviors, and the new device's adversarial detection result and blacklist database record are generated, specifically: if the abnormal probability is higher than 0.3, the K-means clustering sets the cluster number k to 2, based on the Euclidean distance, using the Scikit-learn library, the iteration number is 100, the projection of the user behavior identification vector in the abnormal feature space is divided into real class and false class, if the user behavior identification vector is classified as false class, the new device's binding request is rejected, and the abnormal feature is recorded to the blacklist database, if the user behavior identification vector is classified as real class, the new device is confirmed to pass the adversarial detection, and the binding is allowed.

[0102] Further explanation, through the adversarial generation network and the reverse mechanism, the counterfeit behavior (such as the simulated sliding track) is accurately detected, the risk of false binding request passing the verification is significantly reduced, the security of the binding process is enhanced, and it is suitable for complex Internet of Things scenes.

[0103] The embodiment also provides an Internet of Things device binding system based on AI behavior data, comprising:

[0104] The acquisition module acquires multi-modal behavior data, performs preprocessing and feature extraction, and obtains a set of spatio-temporal interaction features;

[0105] The construction module defines the Internet of Things device as a node, defines the set of spatio-temporal interaction features as a node attribute, calculates the weight of the edge between the nodes through the behavior correlation, and completes the construction of the behavior correlation topology graph;

[0106] The judgment module verifies the behavior consistency of the new device through the behavior correlation topology graph using the trained GNN to obtain the legality probability of the new device;

[0107] The verification module sets a trigger threshold, collects the implicit bioelectrical impedance data of the new device when the legitimacy probability of the new device is less than the trigger threshold, performs cross-modal feature fusion, generates a user behavior identification vector, and performs final verification to obtain a final verification result of the new device.

[0108] The binding module detects and peels off a false binding request by using a reverse mechanism according to the user behavior identification vector, and finally completes the binding of the Internet of Things device.

[0109] The embodiment also provides a computer device suitable for the AI behavior data-based Internet of Things device binding method, which includes a memory and a processor.

[0110] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to perform wired or wireless communication with an external terminal. The wireless communication can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, a trackball or a touchpad arranged on the shell of the computer device, or an external keyboard, a touchpad or a mouse, etc.

[0111] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the method for binding an Internet of Things device based on AI behavior data as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic storage, a flash memory, a magnetic disk, or an optical disk.

[0112] To sum up, the present application: by collecting multi-modal behavior data and generating a set of space-time interaction features, multi-dimensional behavior representation of user-device interaction is realized, providing high-resolution input for behavior consistency analysis. Compared with traditional single-modal data, multi-modal data enriches the behavior dimension and enhances the behavior distinguishability, and then the high-distinguishability user behavior identification vector is generated by fusing the implicit bioelectrical impedance feature vector and the touch feature vector through the attention mechanism of the multi-modal neural network, further improving the verification accuracy; in addition, by constructing a dynamic behavior association topology graph, dynamic modeling of the space-time interaction relationship between devices is realized, and the trained graph neural network is used for behavior consistency analysis based on the dynamic behavior association topology graph, deep behavior relationship mining is realized, and the limitation of static topology structure on dynamic interaction mode mining is solved.

[0113] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A method for binding IoT devices based on AI behavioral data, characterized in that: include, Multimodal behavioral data is collected, preprocessed, and features are extracted to obtain a spatiotemporal interaction feature set. The multimodal behavioral data includes initial interaction data of a new device and interaction data of a bound device. IoT devices are defined as nodes, and spatiotemporal interaction feature sets are defined as node attributes. By calculating the behavioral correlation, the weights of the edges between nodes are obtained, and the behavioral correlation topology graph is constructed. By using the behavior association topology graph and the trained GNN, the behavioral consistency of the new device is verified, and the probability of the new device's legitimacy is obtained. Set a trigger threshold. When the probability of a new device's legitimacy is less than the trigger threshold, collect the implicit bioelectrical impedance data of the new device, perform cross-modal feature fusion, generate a user behavior identifier vector, and perform final verification to obtain the final verification result of the new device. Based on the user behavior identifier vector, a reverse mechanism is used to detect and strip away fake binding requests, ultimately completing the binding of IoT devices.

2. The IoT device binding method based on AI behavioral data as described in claim 1, characterized in that: The obtained spatiotemporal interaction feature set refers to the spatiotemporal interaction feature set obtained by performing time series alignment and noise removal on multimodal behavior data collected from bound devices and new devices, followed by feature extraction and standardization.

3. The IoT device binding method based on AI behavioral data as described in claim 2, characterized in that: IoT devices are defined as nodes, and spatiotemporal interaction feature sets are defined as node attributes. By calculating behavioral correlations, the weights of edges between nodes are obtained, and the behavioral correlation topology graph is constructed. The specific steps include the following: Each IoT device is represented as a node in a behavioral association topology graph, with spatiotemporal interaction feature sets serving as node attributes; The similarity of behavior between devices is quantified by calculating the Pearson correlation coefficient, and the spatial distance between each pair of IoT devices in the behavioral association topology graph is calculated using Euclidean algorithm. For each pair of nodes in the behavior association topology graph, the edge weights of the behavior association topology graph are calculated based on behavior similarity and spatial distance, thus completing the construction of the behavior association topology graph.

4. The IoT device binding method based on AI behavioral data as described in claim 3, characterized in that: By using the spatiotemporal interaction feature set and behavioral association topology graph, the behavioral consistency of the new device is verified using the trained GNN to obtain the legality probability of the new device. The specific steps include the following: A historical behavior association topology graph is constructed based on historical behavior data, and a graph neural network is trained. The weight matrix and bias matrix of the graph neural network are optimized by supervised learning to obtain the trained graph neural network. The behavior association topology graph is input into the trained graph neural network. The graph neural network aggregates the behavioral relationships between devices through graph convolution, evaluates the consistency of the behavior of the new device with the bound devices, and outputs the probability of the new device's legitimacy.

5. The IoT device binding method based on AI behavioral data as described in claim 4, characterized in that: The setting of a trigger threshold involves collecting implicit bioelectrical impedance data from a new device when its probability of legitimacy is less than the threshold, performing cross-modal feature fusion, and generating a user behavior identifier vector. This process includes the following steps: Based on confidence requirements, a trigger threshold is set. When the probability of a new device's legitimacy is less than the trigger threshold, implicit bioelectrical impedance data and touch behavior data are collected. The multimodal neural network is trained based on historical real behavior data. Historical implicit bioelectrical impedance data and historical touch behavior data are input into the multimodal neural network. The attention weight matrix and attention bias vector of the multimodal neural network are optimized in a supervised learning manner to obtain the trained multimodal neural network. The historical real behavior data includes historical implicit bioelectrical impedance data and historical touch behavior data. Implicit bioelectrical impedance data and touch behavior data are input into a trained multimodal neural network. The multimodal neural network assigns weights through an attention mechanism to obtain user behavior identifier vectors.

6. The IoT device binding method based on AI behavioral data as described in claim 5, characterized in that: Obtaining the final verification result of the new device specifically includes the following steps. A similarity threshold is set based on historical user identifier vectors, and the cosine similarity between user behavior identifier vectors and historical user identifier vectors is calculated. If the cosine similarity is higher than the similarity threshold, it is marked as valid; otherwise, it is marked as invalid.

7. The IoT device binding method based on AI behavioral data as described in claim 6, characterized in that: The final completion of IoT device binding refers to using user behavior identifier vectors as input and training an adversarial generative network to detect the anomalies of the user behavior identifier vectors. If the detected behavior is false, the binding is rejected and the abnormal features are recorded. If the detection is passed and the verification result is legal, the binding is completed.

8. An IoT device binding system based on AI behavioral data, based on the IoT device binding method based on AI behavioral data according to any one of claims 1 to 7, characterized in that: include, The acquisition module collects multimodal behavioral data, performs preprocessing and feature extraction, and obtains a spatiotemporal interaction feature set. The module defines IoT devices as nodes and spatiotemporal interaction feature sets as node attributes. By calculating behavioral correlations, the weights of edges between nodes are obtained, thus completing the construction of the behavioral correlation topology graph. The judgment module uses the behavior association topology graph and the trained GNN to verify the consistency of the behavior of the new device and obtain the probability of the new device's legitimacy. The verification module sets a trigger threshold. When the probability of a new device's legitimacy is less than the trigger threshold, it collects the implicit bioelectrical impedance data of the new device, performs cross-modal feature fusion, generates a user behavior identifier vector, and performs final verification to obtain the final verification result of the new device. The binding module uses a reverse mechanism to detect and strip away fake binding requests based on user behavior identifier vectors, ultimately completing the binding of IoT devices.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the IoT device binding method based on AI behavioral data as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the IoT device binding method based on AI behavioral data as described in any one of claims 1 to 7.

Images