AI security defense method and device based on heterogeneous multi-core processor and storage medium

Abstract

The invention discloses an AI security defense method and device based on a heterogeneous multi-core processor and a storage medium, and relates to the technical field of AI security, in order to improve the execution speed of complex algorithms such as a verification function, a hash function and a deep learning model, a security core of a CPU is internally provided with a logic circuit for realizing various algorithms through hardware, and a corresponding API interface is provided outwards through a driving program. Not only is security reinforcement performed on the software level, but also the security and response speed of the system are improved through hardware acceleration and a self-adaptive security strategy.

Description

Technical Field

[0001] The present invention relates to the field of AI security technology, and in particular to an AI security defense method, device and storage medium based on a heterogeneous multi-core processor. Background Art

[0002] In recent years, with the rapid development of artificial intelligence, the application scenarios of artificial intelligence have gradually moved from the server side to the edge side, and the demand for edge computing has been increasing. Among them, deep learning algorithms represented by convolutional neural networks are increasingly deployed and applied on heterogeneous processors at the edge side. However, the popularization of artificial intelligence cannot avoid solving AI security issues. Most of the existing AI security defense methods rely on software-level security defense, which takes up a lot of system resources and has low execution efficiency. Summary of the invention

[0003] In order to overcome the deficiencies of the above technologies, the present invention provides an AI security defense method, device and storage medium based on a heterogeneous multi-core processor, which improves the security and response speed of the system.

[0004] The technical solution adopted by the present invention to overcome the technical problems is:

[0005] An AI security defense method based on a heterogeneous multi-core processor, comprising:

[0006] S1. When a user downloads a new AI application, the operating system receives and responds to the download request and reviews it;

[0007] S2. After the operating system passes the review, the CPU allocates a program storage area and a model data storage area for the AI ​​application, and the CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area;

[0008] S3. The CPU security core obtains the preset audit mechanism from the cloud database, and the CPU security core performs a security audit on the AI ​​application based on the preset audit mechanism;

[0009] S4. After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation;

[0010] S5. The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm for data verification;

[0011] S6. The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models. When an anomaly is found, the CPU's security core handles it.

[0012] The security core of S7.CPU synchronizes with the cloud database in real time and updates security policies and algorithms.

[0013] Furthermore, in step S1, after the operating system receives and responds to the download request, the CPU initiates a download review request to the security core, and the security core reviews the submitted request according to preset rules.

[0014] Furthermore, the preset rules include: reviewing whether the user who submits the request has the authority to install the AI ​​application, reviewing whether the source of the AI ​​application is credible, and reviewing whether the AI ​​application complies with security policies.

[0015] Furthermore, in step S2, after the CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area, the CPU calls the verification interface of the security core to verify whether the AI ​​application and model parameters downloaded locally are consistent with the content provided by the download source.

[0016] Furthermore, the audit mechanism preset in step S3 includes: virus scanning and malicious code detection.

[0017] Furthermore, in step S4, the method for setting memory protection for the AI ​​application storage area and the model data storage area after installation is as follows: setting the memory protection attribute of the model data storage area to read-only, and setting the memory protection attribute of the application AI application storage area to readable and executable.

[0018] Furthermore, the method also includes executing in step S5: the security core of the CPU provides an unalterable audit trail for data through blockchain technology.

[0019] Furthermore, the method for the CPU security core to perform processing in step S6 includes: isolating the application, notifying the user, and automatically updating the security policy.

[0020] On the other hand, the present invention also relates to an AI security defense device based on a heterogeneous multi-core processor, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; wherein:

[0021] The memory is used to store computer programs;

[0022] The processor is configured to execute, by running a program stored in the memory:

[0023] When a user downloads a new AI application, the operating system receives and responds to the download request and performs an audit. After the operating system passes the audit, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area. The CPU's security core obtains the preset audit mechanism from the cloud database, and the CPU's security core performs a security audit on the AI ​​application based on the preset audit mechanism.

[0024] After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation;

[0025] The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm to verify the data;

[0026] The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models. When an anomaly is found, the CPU's security core handles it.

[0027] The CPU's security core synchronizes with the cloud database in real time to update security policies and algorithms.

[0028] On the other hand, the present invention also relates to a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program implements:

[0029] When a user downloads a new AI application, the operating system receives and responds to the download request and performs an audit. After the operating system passes the audit, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area. The CPU's security core obtains the preset audit mechanism from the cloud database, and the CPU's security core performs a security audit on the AI ​​application based on the preset audit mechanism.

[0030] After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation;

[0031] The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm to verify the data;

[0032] The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models. When an anomaly is found, the CPU's security core handles it.

[0033] The CPU's security core synchronizes with the cloud database in real time to update security policies and algorithms.

[0034] The beneficial effects of the present invention are as follows: in order to improve the execution speed of complex algorithms such as verification functions, hash functions, and deep learning models, the CPU security core has built-in hardware to implement the logic circuits of various algorithms, and provides corresponding API interfaces to the outside through the driver. Not only is security reinforcement performed at the software level, but also the security and response speed of the system are improved through hardware acceleration and adaptive security strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0036] The following is combined with Figure 1 The preferred embodiments of the present invention are described in detail so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0037] In the description of the present invention, it should be noted that the embodiments described in the present invention are only part of the embodiments of the present invention, rather than all of the embodiments; based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0038] The terms "first", "second", etc. in the specification and claims of this article and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of this article described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0039] Example 1

[0040] An AI security defense method based on a heterogeneous multi-core processor, comprising:

[0041] S1. When a user downloads a new AI application, the operating system receives and responds to the download request and reviews it.

[0042] S2. After the operating system is reviewed and approved, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area.

[0043] S3. The CPU security core obtains the preset audit mechanism from the cloud database, and the CPU security core performs a security audit on the AI ​​application based on the preset audit mechanism.

[0044] S4. After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation.

[0045] S5. The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm to verify the data to prevent the data from being maliciously tampered with.

[0046] S6. The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models to identify and respond to abnormal behaviors. When an abnormality is detected, the CPU's security core processes it and takes immediate action.

[0047] S7. In order to improve the robustness of the system, the CPU's security core synchronizes with the cloud database in real time and automatically updates security policies and algorithms based on the latest security threat intelligence without manual intervention.

[0048] It includes seven security mechanisms: request review, data verification, content review, memory protection, data tamper-proofing, behavior analysis, and security updates. The heterogeneous multi-core chip consists of three heterogeneous cores: CPU core, NPU core, and security core. In order to improve the execution speed of complex algorithms such as verification functions, hash functions, and deep learning models, the CPU security core has built-in hardware to implement the logic circuits of various algorithms, and provides the corresponding API interface to the outside through the driver. Not only is security hardening performed at the software level, but the system's security and response speed are also improved through hardware acceleration and adaptive security strategies.

[0049] In this embodiment, after the operating system receives and responds to the download request in step S1, the CPU initiates a download review request to the security core, and the security core reviews the submitted request according to preset rules.

[0050] Furthermore, the preset rules include: reviewing whether the user who submits the request has the authority to install the AI ​​application, reviewing whether the source of the AI ​​application is credible, and reviewing whether the AI ​​application complies with security policies.

[0051] In this embodiment, in step S2, after the CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area, the CPU calls the verification interface of the security core to verify whether the AI ​​application and model parameters downloaded locally are consistent with the content provided by the download source.

[0052] In this embodiment, the audit mechanism preset in step S3 includes: virus scanning, malicious code detection. In this embodiment, the method of setting memory protection for the AI ​​application storage area and the model data storage area after installation in step S4 is: setting the memory protection attribute of the model data storage area to read-only, and setting the memory protection attribute of the application AI application storage area to readable and executable, to prevent the accidental writing of illegal data.

[0053] In this embodiment, it also includes executing in step S5: the security core of the CPU provides an unalterable audit trail for data through blockchain technology.

[0054] In this embodiment, the method for processing by the security core of the CPU in step S6 includes: isolating the application, notifying the user, and automatically updating the security policy.

[0055] Example 2

[0056] An AI security defense device based on a heterogeneous multi-core processor comprises: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; wherein:

[0057] The memory is used to store computer programs;

[0058] The processor is configured to execute, by running a program stored in the memory:

[0059] When a user downloads a new AI application, the operating system receives and responds to the download request and performs an audit. After the operating system passes the audit, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area. The CPU's security core obtains the preset audit mechanism from the cloud database, and the CPU's security core performs a security audit on the AI ​​application based on the preset audit mechanism.

[0060] After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation.

[0061] The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm for data verification.

[0062] The CPU's security core monitors the behavior of AI applications in real time, analyzes the behavior patterns of AI applications through deep learning models, and processes any anomalies that are detected.

[0063] The CPU's security core synchronizes with the cloud database in real time to update security policies and algorithms.

[0064] Example 3

[0065] A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the computer program implements:

[0066] When a user downloads a new AI application, the operating system receives and responds to the download request and performs an audit. After the operating system passes the audit, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area. The CPU's security core obtains the preset audit mechanism from the cloud database, and the CPU's security core performs a security audit on the AI ​​application based on the preset audit mechanism.

[0067] After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation.

[0068] The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm for data verification.

[0069] The CPU's security core monitors the behavior of AI applications in real time, analyzes the behavior patterns of AI applications through deep learning models, and processes any anomalies that are detected.

[0070] The CPU's security core synchronizes with the cloud database in real time to update security policies and algorithms.

[0071] It should be understood that in the various embodiments of this document, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this document.

[0072] It should also be understood that in the embodiments of this article, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0073] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this article.

[0074] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0075] In the several embodiments provided herein, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0076] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this article.

[0077] In addition, each functional unit in each embodiment of this invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional unit.

[0078] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this article is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of this article. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-On l yMemory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0079] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An AI security defense method based on heterogeneous multi-core processors, characterized in that: include: S1. When a user downloads a new AI application, the operating system receives and responds to the download request and reviews it; S2. After the operating system passes the review, the CPU allocates a program storage area and a model data storage area for the AI ​​application, and the CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area; S3. The CPU security core obtains the preset audit mechanism from the cloud database, and the CPU security core performs a security audit on the AI ​​application based on the preset audit mechanism; S4. After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation; S5. The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm for data verification; S6. The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models. When an anomaly is found, the CPU's security core handles it. The security core of S7.CPU synchronizes with the cloud database in real time and updates security policies and algorithms.

2. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 1 is characterized in that: In step S1, after the operating system receives and responds to the download request, the CPU initiates a download review request to the security core, and the security core reviews the submitted request according to preset rules.

3. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 2 is characterized in that: The preset rules include: reviewing whether the user who submits the request has the authority to install the AI ​​application, reviewing whether the source of the AI ​​application is credible, and reviewing whether the AI ​​application complies with security policies.

4. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 1, characterized in that: In step S2, after the CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area, the CPU calls the verification interface of the security core to verify whether the AI ​​application and model parameters downloaded to the local computer are consistent with the content provided by the download source.

5. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 1 is characterized in that: The audit mechanism preset in step S3 includes: virus scanning and malicious code detection.

6. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 1 is characterized in that: The method for setting memory protection for the AI ​​application storage area and the model data storage area after installation in step S4 is: setting the memory protection attribute of the model data storage area to read-only, and setting the memory protection attribute of the AI ​​application storage area to readable and executable.

7. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 1 is characterized in that: It also includes executing in step S5: the security core of the CPU provides an unalterable audit trail for data through blockchain technology.

8. The AI ​​security defense method based on heterogeneous multi-core processor according to claim 1 is characterized in that: The method for the CPU security core to perform processing in step S6 includes: isolating the application, notifying the user, and automatically updating the security policy.

9. An AI security defense device based on a heterogeneous multi-core processor, characterized in that: include: A processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; wherein: The memory is used to store computer programs; The processor is configured to execute, by running a program stored in the memory: When a user downloads a new AI application, the operating system receives and responds to the download request and performs an audit. After the operating system passes the audit, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area. The CPU's security core obtains the preset audit mechanism from the cloud database, and the CPU's security core performs a security audit on the AI ​​application based on the preset audit mechanism. After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation; The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm to verify the data; The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models. When an anomaly is found, the CPU's security core handles it. The CPU's security core synchronizes with the cloud database in real time to update security policies and algorithms.

10. A computer-readable storage medium, characterized in that: The computer readable storage medium stores a computer program, which, when executed by a processor, implements: When a user downloads a new AI application, the operating system receives and responds to the download request and performs an audit. After the operating system passes the audit, the CPU allocates a program storage area and a model data storage area for the AI ​​application. The CPU downloads the AI ​​application to the corresponding program storage area and writes the model parameters to the model data storage area. The CPU's security core obtains the preset audit mechanism from the cloud database, and the CPU's security core performs a security audit on the AI ​​application based on the preset audit mechanism. After the security audit is passed, the CPU's security core allows the CPU to install the AI ​​application and sets memory protection for the AI ​​application storage area and model data storage area after installation; The CPU's security core monitors the data in the AI ​​application storage area and the model data storage area in real time and periodically calls its own hash function algorithm to verify the data; The CPU's security core monitors the behavior of AI applications in real time and analyzes their behavior patterns through deep learning models. When an anomaly is found, the CPU's security core handles it. The CPU's security core synchronizes with the cloud database in real time to update security policies and algorithms.

Images