Method and system for realizing intelligent dynamic protection based on prediction
By identifying the intent of network attacks and dynamically migrating computing resources, the problem of not being able to identify the intent of attacks and avoiding attacks in the prior art is solved, and the security and defense capabilities of artificial intelligence systems are improved.
Patent Information
- Application Number
- CN202411183761.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-12
AI Technical Summary
Existing artificial intelligence systems cannot identify the intent of cyber attacks and dynamically lay out computing resources to avoid attacks, resulting in insufficient security and defense capabilities.
通过收集日常数据,提取指令、签名、字段和时间戳生成指纹序列集,分析行为操作抽取特征,生成特征向量,构建智能模型识别攻击意图,预测攻击路径,并动态迁移计算资源以避开攻击。
It realizes automatic identification of attack intentions, predict attack paths, dynamically avoid or block attacks, and improves the security and defense capabilities of artificial intelligence systems.
Smart Images

Figure CN118944955A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of network security technology, and in particular to a method and system for realizing intelligent dynamic protection based on prediction. Background Art
[0002] With the support of artificial intelligence technology, science and technology have developed rapidly, opening up new tracks, but also bringing more security issues. Existing artificial intelligence systems can identify and predict network attacks through models, trace the source of attacks, and contain attacks at the source, but they cannot provide intelligent dynamic protection, that is, they cannot identify attack intentions and dynamically deploy computing resources according to attack intentions to avoid attacks at the end.
[0003] Therefore, there is an urgent need for a targeted prediction-based intelligent dynamic protection method and system. Summary of the invention
[0004] The purpose of the present invention is to provide a method and system for realizing intelligent dynamic protection based on prediction, which can identify attack intentions, predict attack paths, and dynamically migrate computing resources according to the predictions, avoid attacks at the end, or deceive attacks at the end, thereby improving the security and dynamic defense capabilities of artificial intelligence systems.
[0005] In a first aspect, the present application provides a method for realizing intelligent dynamic protection based on prediction, the method comprising:
[0006] Collecting daily data, the collection includes scanning log files and system files, extracting instructions, signatures, fields and timestamps contained therein, dividing them according to events, and generating fingerprint sequence sets for instructions, signatures, fields and timestamps belonging to the same event;
[0007] Among them, instruction, signature, field and timestamp are four sub-dimensions that form a subset;
[0008] Analyze the behavior operation of the event and extract features with special meanings or signs therefrom, wherein the features with special meanings or signs include one or more of a character string, a control flow graph, a bytecode sequence, PE file header information, an API sequence, a system call behavior, runtime state information, a process operation, a file operation, and a registry operation;
[0009] Select representative features from the extracted features by using feature selection, alignment and disambiguation, and convert them into a vector form that can be recognized by the model through vector representation to generate a feature vector;
[0010] The feature vector is correspondingly merged with the above-mentioned fingerprint sequence set, and the feature vector belonging to the same event is clustered and analyzed with the subset to obtain the behavior intention of the feature vector. Since the behavior here is a series of multiple operations corresponding to a subset of multiple timestamps, the behavior intention here is the intention of continuous time;
[0011] Build and train an intelligent model, use the trained intelligent model to learn the above behavior intentions, identify whether the above behavior intentions carry attack intentions, and locate the timestamp and location of the attack;
[0012] In the above learning process, a traceability graph of the feature vector is constructed, and an optimized traceability graph is obtained through pruning and reorganization. The change path of the feature vector is predicted based on the optimized traceability graph;
[0013] Infer the attack path based on the predicted change path, and dynamically migrate computing resources on the path to avoid attacks at the end;
[0014] Based on the timestamp and location of the attack, the attack is blocked at that location.
[0015] In a second aspect, the present application provides a prediction-based intelligent dynamic protection system, the system comprising:
[0016] A sequence generation module is used to collect daily data, wherein the collection includes scanning log files and system files, extracting instructions, signatures, fields and timestamps contained therein, and dividing the instructions, signatures, fields and timestamps belonging to the same event into a fingerprint sequence set according to the event.
[0017] Among them, instruction, signature, field and timestamp are four sub-dimensions that form a subset;
[0018] A behavior analysis module, used to analyze the behavior operation of the event and extract features with special meanings or signs therefrom, wherein the features with special meanings or signs include one or more of a string, a control flow graph, a bytecode sequence, PE file header information, an API sequence, a system call behavior, runtime state information, a process operation, a file operation, and a registry operation;
[0019] A vector generation module is used to select representative features from the extracted features by using feature selection, alignment and disambiguation, and convert them into a vector form that can be recognized by the model through vector representation to generate a feature vector;
[0020] The intention analysis module is used to merge the feature vector with the above-mentioned fingerprint sequence set, cluster the feature vectors belonging to the same event with the subset, and obtain the behavior intention of the feature vector. Since the behavior here is a series of multiple operations corresponding to a subset of multiple timestamps, the behavior intention here is the intention of continuous time;
[0021] The model recognition module is used to build and train an intelligent model, use the trained intelligent model to learn the above behavior intentions, identify whether the above behavior intentions carry attack intentions, and locate the timestamp and location of the attack; in the above learning process, construct a traceability graph of the feature vector, obtain an optimized traceability graph through pruning and reorganization, and predict the change path of the feature vector based on the optimized traceability graph;
[0022] The resource migration module is used to infer the attack path based on the predicted change path and dynamically migrate the computing resources on the path to avoid attacks at the end.
[0023] Based on the timestamp and location of the attack, the attack is blocked at that location.
[0024] In a third aspect, the present application provides a prediction-based intelligent dynamic protection system, the system comprising a processor and a memory:
[0025] The memory is used to store program code and transmit the program code to the processor;
[0026] The processor is used to execute any one of the four possible methods of the first aspect according to the instructions in the program code.
[0027] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to be executed by a processor to implement any one of the four possible methods in the first aspect.
[0028] Beneficial Effects
[0029] The present invention provides a method and system for realizing intelligent dynamic protection based on prediction, which generates a fingerprint sequence set by extracting instructions, signatures, fields and timestamps of massive daily data; analyzes corresponding event behaviors, extracts features with special meanings or signs, generates feature vectors, clusters and analyzes the above two generated features to obtain behavioral intentions; constructs an intelligent model, identifies and locates attack intentions by learning behavioral intentions, predicts the change path of feature vectors, dynamically migrates computing resources on the path, and avoids attacks at the end, thereby overcoming the problem that the prior art cannot identify attack intentions and dynamically arrange computing resources according to attack intentions.
[0030] The artificial intelligence security protection method and system of the present invention have the following advantages and effects:
[0031] It is able to independently analyze daily data, automatically identify attacks, recognize attack intentions, predict attack paths, and dynamically migrate computing resources based on predictions to avoid attacks at the end, or deceive attacks at the end, thereby improving the security and dynamic defense capabilities of the artificial intelligence system. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0033] Figure 1 A general flow chart of the method for realizing intelligent dynamic protection based on prediction according to the present invention;
[0034] Figure 2 This is an architecture diagram of the intelligent dynamic protection system implemented based on prediction in the present invention. DETAILED DESCRIPTION
[0035] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings 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.
[0036] Existing technologies are unable to automatically identify network attacks and conduct network defense, and there is a need to improve the security and elastic expansion capabilities of artificial intelligence systems.
[0037] Figure 1 A general flow chart of a method for realizing intelligent dynamic protection based on prediction provided in this application, the method comprising:
[0038] Collecting daily data, the collection includes scanning log files and system files, extracting instructions, signatures, fields and timestamps contained therein, dividing them according to events, and generating fingerprint sequence sets for instructions, signatures, fields and timestamps belonging to the same event;
[0039] Since the generated fingerprint sequences correspond to different timestamps, the fingerprint sequence set here includes fingerprint sequences with multiple different timestamps.
[0040] Among them, instruction, signature, field and timestamp are four sub-dimensions that form a subset;
[0041] Analyze the behavior operation of the event and extract features with special meanings or signs therefrom, wherein the features with special meanings or signs include one or more of a character string, a control flow graph, a bytecode sequence, PE file header information, an API sequence, a system call behavior, runtime state information, a process operation, a file operation, and a registry operation;
[0042] Select representative features from the extracted features by using feature selection, alignment and disambiguation, and convert them into a vector form that can be recognized by the model through vector representation to generate a feature vector;
[0043] The feature vector is correspondingly merged with the above-mentioned fingerprint sequence set, and the feature vector belonging to the same event is clustered and analyzed with the subset to obtain the behavior intention of the feature vector. Since the behavior here is a series of multiple operations corresponding to a subset of multiple timestamps, the behavior intention here is the intention of continuous time;
[0044] Build and train an intelligent model, use the trained intelligent model to learn the above behavior intentions, identify whether the above behavior intentions carry attack intentions, and locate the timestamp and location of the attack;
[0045] The learning process here is the process of using intelligent models to learn continuous-time behavioral intentions.
[0046] In the above learning process, a traceability graph of the feature vector is constructed, and an optimized traceability graph is obtained through pruning and reorganization. The change path of the feature vector is predicted based on the optimized traceability graph;
[0047] Infer the attack path based on the predicted change path, and dynamically migrate computing resources on the path to avoid attacks at the end;
[0048] Based on the timestamp and location of the attack, the attack is blocked at that location.
[0049] The identification of whether there is an attack intention may include four detections: maliciousness detection, malicious group detection, malicious behavior identification, and malicious attack location. The malicious group here refers to whether there are multiple malicious sources and whether they are highly correlated groups.
[0050] The data storage and computing of the above-mentioned embodiments can also be separated in architecture, respectively using plug-in implementation methods, to facilitate elastic expansion for multiple instances and multiple users, and to adjust the architectural layout of data storage and computing according to the predicted change path.
[0051] Adjusting the architecture layout according to the predicted change path here means that when there is an attack on the predicted change path, the resources on the change path are transferred, and other adjacent resources are used as storage and computing carriers. When the attack is eliminated, the original functions of the resources are restored.
[0052] The embodiments may further include, on the one hand, building a fingerprint library by collecting historical attack data, and obtaining newly discovered network attack protocols in recent years to obtain IP fragment messages corresponding to response messages; on the other hand, filtering out traffic that meets predetermined conditions to filter out traffic that meets potential reflection attacks, and then extracting features of the traffic according to the fragment message fields, matching the features with the features of the fingerprint library to determine whether it belongs to a new type of network attack protocol; if not, detecting it according to the above artificial intelligence security protection method and taking corresponding defense strategies.
[0053] In some preferred embodiments, targeted defense measures taken after the attack is located include: data encryption, firewall settings, intrusion blocking, and redirection.
[0054] In some preferred embodiments, there is a mapping relationship between the daily data and the behavioral operations of the event, and the feature vector of the same event can be associated with the fingerprint sequence set based on the mapping relationship.
[0055] In some preferred embodiments, blocking the attack at the location includes putting the node at the location to sleep or redirecting the traffic of the node.
[0056] In this embodiment, in addition to putting the node into sleep mode or redirecting the node's services, a false service spoofing attack may also be set at the node to change the attack path and protect the terminal computing resources.
[0057] At the same time, reflection network attacks often use defects or loopholes in certain protocols to spoof IP. Therefore, it is necessary to analyze the protocol types used in the traffic, especially those protocols that do not authenticate the source IP, such as ICMP, UDP, etc. If these protocols are used in a large number of attack traffic and the source IP address is forged, it may be a reflection network attack. The above approach can also be added to the embodiments of the present application as a factor of fuzzy judgment.
[0058] Figure 2 The architecture diagram of the intelligent dynamic protection system based on prediction provided in this application includes:
[0059] A sequence generation module is used to collect daily data, wherein the collection includes scanning log files and system files, extracting instructions, signatures, fields and timestamps contained therein, and dividing the instructions, signatures, fields and timestamps belonging to the same event into a fingerprint sequence set according to the event.
[0060] Among them, instruction, signature, field and timestamp are four sub-dimensions that form a subset;
[0061] A behavior analysis module, used to analyze the behavior operation of the event and extract features with special meanings or signs therefrom, wherein the features with special meanings or signs include one or more of a string, a control flow graph, a bytecode sequence, PE file header information, an API sequence, a system call behavior, runtime state information, a process operation, a file operation, and a registry operation;
[0062] A vector generation module is used to select representative features from the extracted features by using feature selection, alignment and disambiguation, and convert them into a vector form that can be recognized by the model through vector representation to generate a feature vector;
[0063] The intention analysis module is used to merge the feature vector with the above-mentioned fingerprint sequence set, cluster the feature vectors belonging to the same event with the subset, and obtain the behavior intention of the feature vector. Since the behavior here is a series of multiple operations corresponding to a subset of multiple timestamps, the behavior intention here is the intention of continuous time;
[0064] The model recognition module is used to build and train an intelligent model, use the trained intelligent model to learn the above behavior intentions, identify whether the above behavior intentions carry attack intentions, and locate the timestamp and location of the attack; in the above learning process, construct a traceability graph of the feature vector, obtain an optimized traceability graph through pruning and reorganization, and predict the change path of the feature vector based on the optimized traceability graph;
[0065] The resource migration module is used to infer the attack path based on the predicted change path and dynamically migrate the computing resources on the path to avoid attacks at the end.
[0066] Based on the timestamp and location of the attack, the attack is blocked at that location.
[0067] The present application provides a prediction-based intelligent dynamic protection system, the system comprising: the system comprising a processor and a memory:
[0068] The memory is used to store program code and transmit the program code to the processor;
[0069] The processor is used to execute the method described in any one of all embodiments of the first aspect according to the instructions in the program code.
[0070] The present application provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to be executed by a processor to implement the method described in any one of all the embodiments of the first aspect.
[0071] In a specific implementation, the present invention further provides a computer storage medium, wherein the computer storage medium may store a program, and when the program is executed, the program may include some or all of the steps in each embodiment of the present invention. The storage medium may be a disk, an optical disk, a read-only storage memory (abbreviated as: ROM) or a random access memory (abbreviated as: RAM), etc.
[0072] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution in the embodiments of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention or some parts of the embodiments.
[0073] The same and similar parts between the various embodiments of this specification can be referred to each other. In particular, for the embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiment.
[0074] The above-described embodiments of the present invention do not limit the protection scope of the present invention.
Claims
1. A method for realizing intelligent dynamic protection based on prediction, characterized in that: The method comprises: Collecting daily data, the collection includes scanning log files and system files, extracting instructions, signatures, fields and timestamps contained therein, dividing them according to events, and generating fingerprint sequence sets for instructions, signatures, fields and timestamps belonging to the same event; Among them, instruction, signature, field and timestamp are four sub-dimensions that form a subset; Analyze the behavior operation of the event and extract features with special meanings or signs therefrom, wherein the features with special meanings or signs include one or more of a character string, a control flow graph, a bytecode sequence, PE file header information, an API sequence, a system call behavior, runtime state information, a process operation, a file operation, and a registry operation; Select representative features from the extracted features by using feature selection, alignment and disambiguation, and convert them into a vector form that can be recognized by the model through vector representation to generate a feature vector; The feature vector is correspondingly merged with the above-mentioned fingerprint sequence set, and the feature vector belonging to the same event is clustered and analyzed with the subset to obtain the behavior intention of the feature vector. Since the behavior here is a series of multiple operations corresponding to a subset of multiple timestamps, the behavior intention here is the intention of continuous time; Build and train an intelligent model, use the trained intelligent model to learn the above behavior intentions, identify whether the above behavior intentions carry attack intentions, and locate the timestamp and location of the attack; In the above learning process, a traceability graph of the feature vector is constructed, and an optimized traceability graph is obtained through pruning and reorganization. The change path of the feature vector is predicted based on the optimized traceability graph; Infer the attack path based on the predicted change path, and dynamically migrate computing resources on the path to avoid attacks at the end; Based on the timestamp and location of the attack, the attack is blocked at that location.
2. The method according to claim 1, characterized in that: After the attack is located, targeted defense measures are taken, including: data encryption, firewall settings, intrusion blocking, and redirection.
3. The method according to claim 1, characterized in that: There is a mapping relationship between the daily data and the behavior operations of the events, and the feature vector of the same event can be associated with the fingerprint sequence set according to the mapping relationship.
4. The method according to any one of claims 2 or 3, characterized in that: Blocking the attack at the location includes putting the node at the location to sleep or redirecting the traffic of the node.
5. An intelligent dynamic protection system based on prediction, characterized in that: The system comprises: A sequence generation module is used to collect daily data, wherein the collection includes scanning log files and system files, extracting instructions, signatures, fields and timestamps contained therein, and dividing the instructions, signatures, fields and timestamps belonging to the same event into a fingerprint sequence set according to the event. Among them, instruction, signature, field and timestamp are four sub-dimensions that form a subset; A behavior analysis module, used to analyze the behavior operation of the event and extract features with special meanings or signs therefrom, wherein the features with special meanings or signs include one or more of a string, a control flow graph, a bytecode sequence, PE file header information, an API sequence, a system call behavior, runtime state information, a process operation, a file operation, and a registry operation; A vector generation module is used to select representative features from the extracted features by using feature selection, alignment and disambiguation, and convert them into a vector form that can be recognized by the model through vector representation to generate a feature vector; The intention analysis module is used to merge the feature vector with the above-mentioned fingerprint sequence set, cluster the feature vectors belonging to the same event with the subset, and obtain the behavior intention of the feature vector. Since the behavior here is a series of multiple operations corresponding to a subset of multiple timestamps, the behavior intention here is the intention of continuous time; The model recognition module is used to build and train an intelligent model, use the trained intelligent model to learn the above behavior intentions, identify whether the above behavior intentions carry attack intentions, and locate the timestamp and location of the attack; in the above learning process, construct a traceability graph of the feature vector, obtain an optimized traceability graph through pruning and reorganization, and predict the change path of the feature vector based on the optimized traceability graph; The resource migration module is used to infer the attack path based on the predicted change path and dynamically migrate the computing resources on the path to avoid attacks at the end. Based on the timestamp and location of the attack, the attack is blocked at that location.
6. An intelligent dynamic protection system based on prediction, characterized in that: The system comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the method according to any one of claims 1 to 4 according to the instructions in the program code.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program codes, and the program codes are used to be executed by a processor to implement the method according to any one of claims 1 to 4.
