Terminal data security processing method and system

Abstract

The invention provides a terminal data security processing method and system. The method comprises the following steps: collecting multi-source heterogeneous data generated by an intelligent terminal and carrying out cross-modal collaborative labeling; carrying out localization preprocessing operation on the labeled multi-source heterogeneous data according to the edge side of the cloud server to obtain preprocessed data and target features, and carrying out dynamic encryption on the preprocessed data to obtain encrypted data; wherein the target feature is used for distinguishing the type and the security level of the encrypted data; and carrying out isolated transmission on the encrypted data according to the type and the security level based on the network slice, carrying out risk early warning processing on the encrypted data based on a target neural network of the cloud server under the condition that the intelligent terminal passes block chain identity authentication, obtaining an early warning result, generating an intelligent decision instruction according to the early warning result, and sending the intelligent decision instruction to the cloud server. And the operation state of the intelligent terminal is adjusted. According to the method disclosed by the invention, the real-time performance and the security of data processing of the intelligent terminal in a complex dynamic scene are improved.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a terminal data security processing method and system. Background Technology

[0002] 5G (5th Generation Mobile Communication Technology) technology, with its three core advantages of ultra-high bandwidth (eMBB), ultra-low latency (uRLLC), and massive machine-type communication (mMTC), provides revolutionary data transmission capabilities for scenarios such as industrial IoT and vehicle networking. However, its real-time interactive characteristics of multi-source heterogeneous data have also introduced security challenges such as the dynamic expansion of the attack surface and the surge in privacy leakage risks.

[0003] The relevant technologies rely on a centralized data collection mode that transmits data directly from the terminal to the cloud, as well as static encryption or post-event backtracking analysis mechanisms. The real-time performance and security of derivative data collection are low, making it difficult to meet the needs of intelligent data processing and secure collaboration in complex dynamic scenarios. Summary of the Invention

[0004] This invention provides a terminal data security processing method and system to address the shortcomings of existing technologies that employ centralized data acquisition modes and static encryption or post-event backtracking analysis mechanisms, resulting in low real-time performance and security of derivative data acquisition. The method described in this invention improves the real-time performance and security of intelligent terminal data processing in complex dynamic scenarios.

[0005] This invention provides a terminal data security processing method, comprising: Collect multi-source heterogeneous data generated by smart terminals, and perform cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data; The labeled multi-source heterogeneous data is preprocessed locally on the edge side of the cloud server to obtain preprocessed data and target features. The preprocessed data is then dynamically encrypted to obtain encrypted data. The target features are used to distinguish the type and security level of the encrypted data. Based on network slicing, the encrypted data is isolated and transmitted according to the type and security level. When the smart terminal is authenticated through blockchain, the encrypted data is processed for risk warning based on the target neural network of the cloud server to obtain a warning result. Based on the warning result, an intelligent decision instruction is generated to adjust the operating status of the smart terminal. Different types of data correspond to different network slices. The warning result includes at least one of abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues.

[0006] According to a terminal data security processing method provided by the present invention, the step of dynamically encrypting the preprocessed data to obtain encrypted data includes: The preprocessed data is encrypted using a lightweight symmetric encryption algorithm at the edge, and the session key is updated using a quantum key distribution mechanism to obtain the encrypted data.

[0007] According to a terminal data security processing method provided by the present invention, the session key is used to trigger a key reset mechanism when a device abnormality or communication quality abnormality is detected.

[0008] According to a terminal data security processing method provided by the present invention, the target neural network is a Transformer network or a graph convolutional network.

[0009] According to a terminal data security processing method provided by the present invention, after generating an intelligent decision instruction based on the early warning result, the method further includes: The edge-side learning model is federated and trained based on the intelligent decision-making instructions and the local data of the intelligent terminal. The model parameters of the trained learning model are then aggregated with the model parameters of the target neural network to obtain a new target neural network.

[0010] The present invention also provides a terminal data security processing system, comprising: The data acquisition module is used to collect multi-source heterogeneous data generated by smart terminals and perform cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data. The preprocessing and encryption module is used to perform localized preprocessing operations on the labeled multi-source heterogeneous data based on the edge side of the cloud server to obtain preprocessed data and target features, and to dynamically encrypt the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data; The transmission and decision-making module is used to isolate and transmit the encrypted data based on network slices according to the type and security level, and when the smart terminal is authenticated through blockchain, it performs risk warning processing on the encrypted data based on the target neural network of the cloud server to obtain a warning result, and generates intelligent decision instructions based on the warning result to adjust the operating status of the smart terminal; different types of data correspond to different network slices; the warning result includes at least one of abnormal behavior type, threat level, scope of impact, confidence level and explanatory clues.

[0011] According to a terminal data security processing system provided by the present invention, after generating intelligent decision instructions based on the early warning results, the system further includes: The optimization module is used to perform federated learning training based on the edge-side learning model according to the intelligent decision instructions and the local data of the intelligent terminal, and to aggregate the model parameters of the trained learning model with the model parameters of the target neural network to obtain a new target neural network.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the terminal data security processing method described above.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the terminal data security processing method as described above.

[0014] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the terminal data security processing method as described above.

[0015] The terminal data security processing method and system provided by this invention performs localized preprocessing operations on labeled multi-source heterogeneous data at the edge of a cloud server, and dynamically encrypts the preprocessed data to obtain encrypted data. Then, based on network slicing, the encrypted data is isolated and transmitted according to type and security level. With the smart terminal authenticated by blockchain, the encrypted data is subjected to risk warning processing based on the target neural network of the cloud server. Finally, intelligent decision instructions are generated based on the warning results to adjust the operating status of the smart terminal, thereby improving the real-time performance and security of smart terminal data processing in complex dynamic scenarios. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is one of the flowcharts illustrating the terminal data security processing method provided by the present invention.

[0018] Figure 2 This is the second flowchart of the terminal data security processing method provided by the present invention.

[0019] Figure 3 This is the third flowchart of the terminal data security processing method provided by the present invention.

[0020] Figure 4 This is the fourth flowchart of the terminal data security processing method provided by the present invention.

[0021] Figure 5 This is a schematic diagram of the terminal data security processing system provided by the present invention.

[0022] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] The following is combined with Figures 1-5 The present invention describes a terminal data security processing method and system.

[0025] Figure 1 This is one of the flowcharts illustrating the terminal data security processing method provided by the present invention, such as... Figure 1 As shown, the method includes the following: Step 110: Collect multi-source heterogeneous data generated by the smart terminal, and perform cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data.

[0026] In this step, multi-source heterogeneous data has the characteristics of high frequency, multi-dimensionality and high dynamics.

[0027] In this step, smart terminals can be applied to application areas with extremely high requirements for data real-time performance, security, and intelligent processing, such as smart cities, industrial IoT, vehicle-to-everything (V2X), smart security, and smart manufacturing.

[0028] Specifically, smart terminals can belong to the vehicle-to-everything (V2X) environment, or to the following scenarios: (1) In smart factories, the monitoring of the operating status of production equipment, anomaly detection, and predictive maintenance are realized; (2) Used in smart cities for crowd density monitoring, gathering risk identification, and public safety early warning; (3) Smart grid and other high-concurrency, high-real-time, and high-security data interaction systems such as telemedicine.

[0029] For example, smart terminals belong to 5G network terminals, such as IoT devices, vehicle units, and mobile terminals.

[0030] In this embodiment, multimodal sensing units, such as image acquisition units, audio acquisition modules, temperature and humidity sensors, gas sensors, and motion detection sensors such as accelerometers / gyroscopes, can be deployed on smart terminals to realize the perception of multi-source heterogeneous data of physical environment and behavioral state; the acquired multi-source heterogeneous data has the characteristics of high frequency, high dimension, high dynamics, and strong noise, which puts forward higher efficiency and safety requirements for subsequent processing.

[0031] In this embodiment, a unified data dictionary and time synchronization mechanism (such as NTP / PTP) can be used to time-align data with different sampling frequencies, and cross-modal collaborative annotation can be achieved through metadata tags (terminal ID, location, time window, data type, integrity check value, etc.) to improve subsequent processing and traceability capabilities.

[0032] For example, multi-source heterogeneous data includes at least two of the following: image data, audio data, environmental data, and motion sensing data. Smart terminals deployed in vehicles (such as in-vehicle cameras, GPS modules, and voice sensors) can collect various heterogeneous data in real time, including vehicle speed, location information, in-vehicle voice, and road condition images, and upload the raw or semi-structured data to edge servers via 5G networks.

[0033] Step 120: Perform localized preprocessing operations on the labeled multi-source heterogeneous data at the edge of the cloud server to obtain preprocessed data and target features, and dynamically encrypt the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data.

[0034] In this step, edge computing nodes deployed at the network edge perform localized preprocessing on the data uploaded by the terminal, such as noise filtering, format standardization, outlier removal, time window segmentation, key feature extraction and vectorization, to reduce the transmission bandwidth burden and improve data processing efficiency; at the same time, target features are extracted to distinguish different types of data and their corresponding security levels.

[0035] In this embodiment, the target features can be key features of different types of data (such as multi-scale texture / target box of images, MFCC / voiceprint features of speech, statistics and frequency domain energy of sensor data, etc.), and can be obtained through feature extraction tools such as machine learning models and neural networks for subsequent security level identification and network slicing strategy selection.

[0036] In this embodiment, the preprocessed data is dynamically encrypted. A lightweight symmetric encryption algorithm at the edge can be used to encrypt the data, and a time-driven and event-driven key update mechanism can be used to achieve dynamic encryption.

[0037] For example, time-driven methods may force the session key to be changed at fixed intervals (e.g., 1 to 5 minutes); event-driven methods may immediately trigger key reset and session renegotiation when abnormal behavior is detected (e.g., base station handover frequency > 5 times / minute, abnormal login, unauthorized access) or QKD error rate > 1%.

[0038] In this embodiment, the key distribution method can be PKI / KMS collaborative distribution or quantum key distribution (QKD) combined with encrypted storage; wherein, the session key can be hosted by KMS and generated and encapsulated within the hardware security module (HSM) to prevent plaintext from appearing in the process space.

[0039] In this embodiment, the processed data is compressed and encoded before transmission (e.g., HEVC, FLAC, differential encoding, or sparse representation compression is selected according to data type) to further reduce bandwidth consumption; sensitive fields can be perturbed or desensitized for storage (e.g., ID mapping, bucketing / truncation) to improve privacy and security without affecting the model's discrimination ability.

[0040] It should be noted that edge nodes can cache feature vectors and encrypted data separately, supporting breakpoint resumption and replay protection (timestamp + random number + MAC) to improve packet loss and replay resistance.

[0041] Step 130: Based on network slicing, the encrypted data is isolated and transmitted according to type and security level. With the smart terminal authenticated via blockchain, the encrypted data undergoes risk warning processing based on the target neural network of the cloud server to obtain warning results. Intelligent decision-making instructions are then generated based on these results to adjust the smart terminal's operating status. Different types of data correspond to different network slices. Warning results include at least one of the following: abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues. In this step, network slicing of the 5G core network is used at the transport layer to dynamically allocate transmission channels based on different data types and security levels, achieving virtual isolation and trusted switching of communication links. For latency-sensitive control data, URLLC slices can be prioritized; for high-throughput video streams, eMBB slices can be allocated; and for massive amounts of low-speed reported monitoring data, mMTC slices can be allocated. Furthermore, priority preemption and elastic bandwidth scaling can be implemented based on congestion levels.

[0042] In this step, a distributed device identity authentication system is built by introducing blockchain technology based on a consortium blockchain architecture. Device access information, access logs and operation permissions are recorded on the chain in an immutable manner to achieve trusted verification of device identity and traceability of access behavior. Events such as device certificate revocation, permission change and abnormal alarm can be automatically triggered through on-chain contracts.

[0043] In this embodiment, at the authentication layer, role-based access control (RBAC) and the principle of least privilege are combined to perform fine-grained management of access permissions for edge and cloud resources, supporting fine-grained authorization based on data type, source, processing purpose, and sensitivity level; at the same time, key encryption, decryption, and signature operations are isolated through remote measurement and trusted execution environment (TEE) to reduce the risk of key leakage.

[0044] In this embodiment, the on-chain record may include: unique device identifier, certificate digest, access time, session ID, slice number, access action and result digest, etc., which facilitates rapid location and evidence collection in post-audit and compliance checks.

[0045] In this embodiment, a target neural network (such as a deep neural network (DNN), a Transformer architecture, or a graph neural network (GNN)) is deployed on a cloud server or cloud platform to perform deep semantic modeling, behavioral pattern recognition, and threat level assessment on uplink encrypted data; cross-modal alignment and spatiotemporal feature fusion are performed on multi-source data to achieve a multi-dimensional risk profile from the device layer, network layer, to the business layer.

[0046] In this embodiment, the encrypted data is uploaded to the cloud via a trusted channel. When the cloud server supports homomorphic encryption algorithms or secure multi-party computation, it can directly perform feature comparison, similarity calculation, or risk scoring on the ciphertext without decrypting the encrypted data. In scenarios that do not support homomorphic computation, the cloud can complete decryption and subsequent processing in a controlled and isolated environment according to the key policy.

[0047] In this embodiment, the cloud-based intelligent analysis module can output, but is not limited to, the following: abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues (contribution of key features); and generate intelligent decision-making instructions based on the warning results and send them to the edge and terminal for risk handling, parameter adaptation, and linkage control, such as dynamically adjusting the sampling frequency, camera resolution / frame rate, reporting cycle, retransmission threshold, slice priority, and transmission power, in order to achieve a balance between security and latency / bandwidth costs.

[0048] It should be noted that, in order to improve the efficiency of the closed loop, cloud-based policies can be rolled out as hot-updateable edge rules or lightweight models, which can take effect immediately locally, thus achieving a rapid closed loop of "detection-evaluation-handling-feedback".

[0049] The terminal data security processing method provided in this invention performs localized preprocessing operations on labeled multi-source heterogeneous data at the edge of a cloud server, and dynamically encrypts the preprocessed data to obtain encrypted data. Then, based on network slicing, the encrypted data is isolated and transmitted according to type and security level. With the smart terminal authenticated by blockchain, the encrypted data is processed for risk warning based on the target neural network of the cloud server. Finally, intelligent decision instructions are generated based on the warning results to adjust the operating status of the smart terminal, thereby improving the real-time performance and security of smart terminal data processing in complex dynamic scenarios.

[0050] In some embodiments, dynamically encrypting the preprocessed data to obtain encrypted data includes: encrypting the preprocessed data using a lightweight symmetric encryption algorithm at the edge and updating the session key using a quantum key distribution mechanism to obtain encrypted data.

[0051] In this embodiment, an improved lightweight symmetric encryption algorithm (e.g., based on optimized AES) and a time window key update mechanism are introduced at the edge, along with quantum key distribution (QKD) technology, to dynamically encrypt the data and ensure data security even in high-speed or frequently switching 5G communication scenarios.

[0052] In this embodiment, by submitting device identity information to a blockchain identity authentication device (consortium blockchain) for legitimacy authentication and access control, and combining role-based access control (RBAC) with audit logs, it is possible to prevent malicious devices from accessing the system and to make access behavior traceable.

[0053] It should be noted that cloud servers can decrypt encrypted data using the encryption key and then perform subsequent processing on the decrypted data; in addition, when a cloud server supports homomorphic encryption algorithms, it can directly perform subsequent processing on the encrypted data without decryption.

[0054] In some embodiments, the session key is used to trigger a key reset mechanism in the event of a detected device malfunction or communication quality anomaly.

[0055] Specifically, a key reset is triggered immediately when abnormal behavior or communication quality abnormalities are detected (e.g., base station handover frequency > 5 times / minute, QKD bit error rate > 1%).

[0056] The terminal data security processing method provided in this invention integrates the technical path of "multi-source perception - edge preprocessing and dynamic encryption - trusted transmission and identity authentication - cloud intelligent analysis and closed-loop control", which can take into account real-time performance, reliability and security compliance in complex dynamic networks and high-concurrency business scenarios.

[0057] Figure 2 This is the second flowchart of the terminal data security processing method provided by the present invention. Figure 2 In the illustrated embodiment, the method includes the following steps: Step 201: Terminal sensing and data collection; Step 202: Edge node formatting process; Step 203: Improved AES encryption; Step 204: Use quantum key distribution; Step 205, secure channel transmission, which means transmitting the terminal's sensed and collected data through a secure channel.

[0058] In some embodiments, the target neural network is a Transformer network or a graph convolutional network.

[0059] In this embodiment, Transformer networks or graph convolutional networks can be used for behavioral patterns, abnormal feature extraction, and classification.

[0060] In this embodiment, encrypted data is uploaded to the cloud via a trusted channel, and the deployed Transformer network or graph convolutional network performs pattern recognition and risk prediction on vehicle behavior. When abnormal driving behavior or potential collision risk is detected, the system will immediately generate a warning message.

[0061] In this embodiment, the Transformer network is capable of parallelization and long sequence modeling, making it more suitable for real-time response to dynamic encrypted streaming data; the graph convolutional network is capable of topology association mining and sparsity optimization, making it more suitable for structured analysis of encrypted graph data.

[0062] The terminal data security processing method provided in this invention, by setting the target neural network as a Transformer network or a graph convolutional network, realizes a smart protection system of "encryption-early warning-decision-execution" through quantum encrypted transmission, TEE secure decryption, Transformer / GCN dynamic risk modeling and decision command closed-loop control.

[0063] In some embodiments, after generating intelligent decision instructions based on the warning results, the terminal data security processing method further includes: performing federated learning training on the edge-based learning model based on the intelligent decision instructions and local data of the intelligent terminal, and aggregating the model parameters of the trained learning model with the model parameters of the target neural network to obtain a new target neural network.

[0064] In this embodiment, a federated learning structure is used to collaboratively update the model parameters of the target neural network online using local data, enabling collaborative training and updating without transferring the original data.

[0065] In this embodiment, the cloud server will feed back the analysis results (intelligent decision instructions) to the edge side and assist intelligent terminal devices in making real-time responses, such as behavior adjustments, early warning prompts, or dynamic optimization of communication parameters.

[0066] In this embodiment, by introducing a federated learning architecture, edge devices can continuously optimize and adaptively adjust the model through local model training and cloud-based model parameter aggregation without uploading raw data. This mechanism can reduce the risk of privacy leakage, improve the model's generalization ability, and enhance the system's adaptability and scalability in different scenarios.

[0067] It should be noted that the above embodiments construct a multi-layered collaborative system centered on "terminal-edge-cloud," integrating six key modules: perception, computing, transmission, security, intelligent analysis, and distributed optimization. This achieves a closed-loop processing mechanism across the entire chain, from data source perception and acquisition, transmission encryption, and identity authentication, to analysis, decision-making, and intelligent feedback. This method not only possesses excellent real-time performance, security, and intelligence, but also exhibits high flexibility and versatility. It can be widely applied to complex and dynamic application scenarios such as smart cities, connected vehicles, industrial IoT, intelligent security, and edge intelligence, and has significant theoretical and engineering value for ensuring data security and stable system operation in the 5G era.

[0068] Figure 3 This is the third flowchart of the terminal data security processing method provided by the present invention. Figure 3 In the illustrated embodiment, the method includes the following steps: Step 301: Cloud-based model training (achieved through batch uploading of historical data); Step 302, Deep Neural Network; Step 303: Behavioral Analysis and Decision Output; Step 304: Central control review; Step 305: Issue strategies and optimization suggestions; Step 306: Federated learning and edge collaboration.

[0069] The terminal data security processing method provided in this invention uses an edge-side learning model to perform federated learning training based on intelligent decision-making instructions and local data of the intelligent terminal. The model parameters of the trained learning model are then aggregated with the model parameters of the target neural network to obtain a new target neural network. This method enables adaptive optimization and continuous evolution of the target neural network on the cloud server while protecting the privacy of the original data. The addition of central control review further improves the reliability and real-time performance of intelligent decision-making instructions.

[0070] Figure 4 This is the fourth flowchart of the terminal data security processing method provided by the present invention. Figure 4 In the illustrated embodiment, the method includes the following steps: Step 401: Multimodal data perception (achieved through cameras / sensors / vehicle-mounted equipment). Step 402, Edge preprocessing node (including formatting, filtering, and feature extraction); Step 403: Dynamic Encryption and Quantum Key; Step 404: Blockchain Identity Authentication and Control; Step 405: Cloud-based DNN intelligent analysis; Step 406: Federated learning drives model optimization feedback mechanism; Step 407: The review mechanism ensures model security.

[0071] The terminal data security processing system provided by the present invention is described below. The terminal data security processing system described below can be referred to in correspondence with the terminal data security processing method described above.

[0072] Figure 5 This is a schematic diagram of the terminal data security processing system provided by the present invention, as shown below. Figure 5 As shown, the terminal data security processing system includes: The data acquisition module 510 is used to acquire multi-source heterogeneous data generated by smart terminals and perform cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data. The preprocessing and encryption module 520 is used to perform localized preprocessing operations on the labeled multi-source heterogeneous data based on the edge side of the cloud server to obtain preprocessed data and target features, and to dynamically encrypt the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data. The transmission and decision module 530 is used to isolate and transmit encrypted data based on network slices according to type and security level. When the smart terminal is authenticated by blockchain, the module performs risk warning processing on the encrypted data based on the target neural network of the cloud server to obtain warning results. Based on the warning results, the module generates intelligent decision instructions to adjust the operating status of the smart terminal. Different types of data correspond to different network slices. The warning results include at least one of the following: abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues.

[0073] The terminal data security processing system provided in this invention performs localized preprocessing operations on labeled multi-source heterogeneous data at the edge of a cloud server, and dynamically encrypts the preprocessed data to obtain encrypted data. Then, based on network slicing, the encrypted data is isolated and transmitted according to type and security level. With the smart terminal authenticated by blockchain, the system performs risk warning processing on the encrypted data based on the target neural network of the cloud server. Finally, based on the warning results, intelligent decision instructions are generated to adjust the operating status of the smart terminal, thereby improving the real-time performance and security of smart terminal data processing in complex dynamic scenarios.

[0074] In some embodiments, after generating intelligent decision instructions based on the early warning results, the terminal data security processing system further includes an optimization module.

[0075] The optimization module is used to perform federated learning training on the edge-based learning model based on intelligent decision instructions and local data from the intelligent terminal, and to aggregate the model parameters of the trained learning model with the model parameters of the target neural network to obtain a new target neural network.

[0076] The terminal data security processing system provided in this embodiment of the invention uses an edge-side learning model to perform federated learning training based on intelligent decision-making instructions and local data of the intelligent terminal. The model parameters of the trained learning model are then aggregated with the model parameters of the target neural network to obtain a new target neural network. This system enables the target neural network on the cloud server to adaptively optimize and continuously evolve while protecting the privacy of the original data, thereby further improving the reliability and real-time performance of intelligent decision-making instructions.

[0077] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6As shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communications bus 640, wherein the processor 610, the communications interface 620, and the memory 630 communicate with each other through the communications bus 640. The processor 610 can call logical instructions in the memory 630 to execute a terminal data security processing method. This method includes: collecting multi-source heterogeneous data generated by the smart terminal and performing cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data; performing localized preprocessing operations on the annotated multi-source heterogeneous data based on the edge side of the cloud server to obtain preprocessed data and target features, and dynamically encrypting the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data; isolating and transmitting the encrypted data according to the type and security level based on network slicing; and, when the smart terminal is authenticated via blockchain, performing risk warning processing on the encrypted data based on the target neural network of the cloud server to obtain warning results, and generating intelligent decision instructions based on the warning results to adjust the operating state of the smart terminal; different types of data correspond to different network slices; the warning results include at least one of abnormal behavior type, threat level, impact range, confidence level, and explanatory clues.

[0078] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the terminal data security processing method provided by the above methods. The method includes: collecting multi-source heterogeneous data generated by a smart terminal and performing cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data; performing localized preprocessing operations on the annotated multi-source heterogeneous data at the edge of a cloud server to obtain preprocessed data and target features, and dynamically encrypting the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data; isolating and transmitting the encrypted data according to the type and security level based on network slicing, and performing risk warning processing on the encrypted data based on the target neural network of the cloud server when the smart terminal is authenticated through blockchain identity, obtaining a warning result, and generating intelligent decision instructions based on the warning result to adjust the operating state of the smart terminal; different types of data correspond to different network slices; the warning result includes at least one of abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues.

[0080] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the terminal data security processing method provided by the above methods. The method includes: collecting multi-source heterogeneous data generated by a smart terminal and performing cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data; performing localized preprocessing operations on the annotated multi-source heterogeneous data based on the edge side of a cloud server to obtain preprocessed data and target features, and dynamically encrypting the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data; isolating and transmitting the encrypted data according to the type and security level based on network slicing, and performing risk warning processing on the encrypted data based on the target neural network of the cloud server when the smart terminal is authenticated through blockchain identity, obtaining a warning result, and generating intelligent decision instructions based on the warning result to adjust the operating state of the smart terminal; different types of data correspond to different network slices; the warning result includes at least one of abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues.

[0081] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0082] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A terminal data security processing method, characterized in that, include: Collect multi-source heterogeneous data generated by smart terminals, and perform cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data; The labeled multi-source heterogeneous data is preprocessed locally on the edge side of the cloud server to obtain preprocessed data and target features. The preprocessed data is then dynamically encrypted to obtain encrypted data. The target features are used to distinguish the type and security level of the encrypted data. Based on network slicing, the encrypted data is isolated and transmitted according to the type and security level. When the smart terminal is authenticated through blockchain, the encrypted data is processed for risk warning based on the target neural network of the cloud server to obtain a warning result. Based on the warning result, an intelligent decision instruction is generated to adjust the operating status of the smart terminal. Different types of data correspond to different network slices. The warning result includes at least one of abnormal behavior type, threat level, scope of impact, confidence level, and explanatory clues.

2. The terminal data security processing method according to claim 1, characterized in that, The step of dynamically encrypting the preprocessed data to obtain encrypted data includes: The preprocessed data is encrypted using a lightweight symmetric encryption algorithm at the edge, and the session key is updated using a quantum key distribution mechanism to obtain the encrypted data.

3. The terminal data security processing method according to claim 2, characterized in that, The session key is used to trigger a key reset mechanism when a device malfunction or communication quality malfunction is detected.

4. The terminal data security processing method according to claim 1, characterized in that, The target neural network is a Transformer network or a graph convolutional network.

5. The terminal data security processing method according to claim 1, characterized in that, After generating intelligent decision instructions based on the early warning results, the method further includes: The edge-side learning model is federated and trained based on the intelligent decision-making instructions and the local data of the intelligent terminal. The model parameters of the trained learning model are then aggregated with the model parameters of the target neural network to obtain a new target neural network.

6. A terminal data security processing system, characterized in that, include: The data acquisition module is used to collect multi-source heterogeneous data generated by smart terminals and perform cross-modal collaborative annotation on the multi-source heterogeneous data to obtain annotated multi-source heterogeneous data. The preprocessing and encryption module is used to perform localized preprocessing operations on the labeled multi-source heterogeneous data based on the edge side of the cloud server to obtain preprocessed data and target features, and to dynamically encrypt the preprocessed data to obtain encrypted data; wherein, the target features are used to distinguish the type and security level of the encrypted data; The transmission and decision-making module is used to isolate and transmit the encrypted data based on network slices according to the type and security level, and when the smart terminal is authenticated through blockchain, it performs risk warning processing on the encrypted data based on the target neural network of the cloud server to obtain a warning result, and generates intelligent decision instructions based on the warning result to adjust the operating status of the smart terminal; different types of data correspond to different network slices; the warning result includes at least one of abnormal behavior type, threat level, scope of impact, confidence level and explanatory clues.

7. The terminal data security processing system according to claim 6, characterized in that, After generating intelligent decision instructions based on the early warning results, the system further includes: The optimization module is used to perform federated learning training based on the edge-side learning model according to the intelligent decision instructions and the local data of the intelligent terminal, and to aggregate the model parameters of the trained learning model with the model parameters of the target neural network to obtain a new target neural network.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the terminal data security processing method as described in any one of claims 1 to 5.

9. A non-transitory 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 terminal data security processing method as described in any one of claims 1 to 5.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the terminal data security processing method as described in any one of claims 1 to 5.

Images