An anti-escape detection method
By simulating sample escape and abnormal detection of multiple sandbox environments and adjusting the sandbox environment, the problem of malware escape detection is solved and detection accuracy and reliability are improved.
Patent Information
- Application Number
- CN202310344675.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-03-31
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2043-03-31
AI Technical Summary
In the prior art, malware has the ability to escape sandbox detection, resulting in poor detection accuracy and cannot meet current network security needs.
By performing sample escape simulation on multiple sandbox environments, detecting whether there are abnormalities in hardware information, environment information, time information and interaction information, and adjusting the sandbox environment according to the abnormal situation, obtaining behavior files to determine behavior abnormality, and locate escape behavior. Update thresholds to improve detection accuracy.
Improve detection accuracy, prevent detection escape of malicious samples, and ensure the reliability of detection.
Smart Images

Figure CN116305091B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of security detection technology, and more specifically, to an anti-escape detection method. Background Art
[0002] With the increase of network threats and the emergence of APT attacks, traditional malicious code detection can no longer meet the requirements of the current severe network security environment. Sandbox technology, as a relatively safe dynamic analysis environment, can automatically analyze unknown samples and record the various behaviors and operations of samples in the sandbox environment, so as to determine the true intention of the samples. At the same time, the operation of malicious samples in the sandbox will not affect the real environment, so it is a relatively ideal analysis environment.
[0003] With the development of malware, some malicious samples have the ability to escape sandbox detection, resulting in poor detection accuracy and failure to meet existing network security needs.
[0004] Therefore, how to improve detection accuracy is a technical problem that needs to be solved. Summary of the invention
[0005] The present invention provides an anti-escape detection method to solve the technical problem of low detection accuracy in the prior art. The method comprises:
[0006] Perform sample escape simulation on multiple sandbox environments, and detect whether there are any anomalies in the sandbox's hardware information, environment information, time information, and interaction information. If there are any anomalies, make corresponding adjustments until the sample escapes simulation;
[0007] After the sample escape simulation passes, obtain the sample to be tested, run it multiple times in multiple sandbox environments, and obtain the behavior file;
[0008] Obtain distance information based on the behavior file, determine the degree of abnormal behavior between sandbox environments based on the distance information, and determine whether there is abnormal behavior based on the relationship between the degree of abnormal behavior and the threshold. If there is abnormal behavior, obtain the system call sequence corresponding to the abnormal behavior, and locate the escape behavior based on the system call sequence;
[0009] The evaluation parameters of each sandbox are obtained, a threshold update array is established according to the evaluation parameters, and the threshold is updated according to the threshold update array.
[0010] In some embodiments of the present application, if there is an abnormality, corresponding adjustments are made, including:
[0011] If there is an anomaly in the hardware information, the hardware information of the real host is obtained, the difference hardware is determined based on the hardware information of the real host and the hardware information of the sandbox, and the sandbox hardware information is adjusted according to the difference hardware and its corresponding cost.
[0012] In some embodiments of the present application, if there is an abnormality, corresponding adjustments are made, including:
[0013] If there is an abnormality in the environmental information, the real environment characteristics and the sandbox environment characteristics are obtained, and the real environment characteristics and the sandbox environment characteristics are compared to determine the characteristics of the sandbox environment, obtain the environmental characteristics required for the sandbox operation, and adjust the sandbox environment characteristics according to the characteristics of the sandbox environment and the environmental characteristics required for the sandbox operation.
[0014] In some embodiments of the present application, if there is an abnormality, corresponding adjustments are made, including:
[0015] If there is an anomaly in the time information, check whether there is a dormant function or a loop function in the sample, determine the start time of the sandbox detection based on the dormant function or the loop function, obtain the resources consumed by different sandbox detection durations, obtain the effective detection time of multiple malicious samples, determine the balance amount based on the resources consumed by different sandbox detection durations and the effective detection time of multiple malicious samples, determine the duration of the sandbox detection based on the balance amount, and perform detection based on the start time and duration of the sandbox detection.
[0016] In some embodiments of the present application, a balance is determined based on resources consumed by different sandbox detection durations and effective detection times of multiple malicious samples, and the duration of sandbox detection is determined based on the balance, including:
[0017] Determine multiple balancing amounts based on resources consumed by different sandbox detection durations, multiple malicious sample effective detection times, and a first preset weight, take the maximum value among the multiple balancing amounts as the required balancing amount, and substitute the balancing amount into a conversion formula to obtain the duration of the sandbox detection;
[0018] The conversion formula is, T=L*M, where T is the duration of the sandbox detection, M is the balance amount, and L is the conversion coefficient. The first preset weight and conversion coefficient corresponding to different sandboxes are different.
[0019] In some embodiments of the present application, if there is an abnormality, corresponding adjustments are made, which also include:
[0020] If there is an abnormality in the interaction information, the interaction information of the real host is obtained, and the interaction information of the sandbox is set according to the interaction information of the real host.
[0021] In some embodiments of the present application, distance information is obtained according to the behavior file, and the degree of abnormal behavior between sandbox environments is determined based on the distance information, including:
[0022] The distance information includes horizontal distance and vertical distance. The horizontal distance is the distance between behavior files in different environments, and the vertical distance is the distance between behavior files in the same environment. The degree of behavior abnormality is obtained based on the difference between the horizontal distance and the vertical distance.
[0023] In some embodiments of the present application, judging whether abnormal behavior exists based on the relationship between the abnormality degree of behavior and the threshold value includes:
[0024] If the degree of behavioral abnormality exceeds the threshold, abnormal behavior exists;
[0025] If the degree of abnormal behavior does not exceed the threshold, there is no abnormal behavior.
[0026] In some embodiments of the present application, a threshold update array is established according to the evaluation parameters, including:
[0027] Evaluation parameters include accuracy, detection rate, and error rate;
[0028] The position order is determined based on the accuracy, detection rate and error rate, and a threshold update array is constructed based on the accuracy, detection rate, error rate and position order.
[0029] In some embodiments of the present application, updating the threshold according to the threshold array includes:
[0030] Update the thresholds based on the threshold update array and the local factors corresponding to the position order.
[0031] By applying the above technical solution, a sample escape simulation is performed on multiple sandbox environments, and the hardware information, environment information, time information and interaction information of the sandbox are detected in turn to see if there are any anomalies. If there are any anomalies, corresponding adjustments are made until the sample escape simulation is passed; after the sample escape simulation is passed, the sample to be detected is obtained, and the sample to be detected is run multiple times in multiple sandbox environments respectively, and a behavior file is obtained; distance information is obtained according to the behavior file, and the degree of abnormal behavior between the sandbox environments is determined based on the distance information. Based on the relationship between the degree of abnormal behavior and the threshold, it is determined whether there is abnormal behavior. If there is abnormal behavior, the system call sequence corresponding to the abnormal behavior is obtained, and the escape behavior is located according to the system call sequence; the evaluation parameters of each sandbox are obtained, and a threshold update array is established according to the evaluation parameters, and the threshold is updated according to the threshold update array. The present application detects defects in the sandbox through escape simulation and adjusts the sandbox, thereby deceiving the environment detection of malicious samples. According to the behavior files in multiple sandboxes and multiple behavior files in one sandbox, abnormal behavior is determined and the system call sequence is obtained to locate the escape behavior. The detection accuracy is improved, the malicious samples are prevented from escaping detection, and the reliability of detection is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.
[0033] Figure 1 A schematic flow chart of an anti-escape detection method proposed in an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0034] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0035] The present application embodiment provides an anti-escape detection method, such as Figure 1 As shown, the method comprises the following steps:
[0036] Step S101, perform sample escape simulation on multiple sandbox environments, and sequentially detect whether there are any anomalies in the sandbox's hardware information, environment information, time information, and interaction information. If there are any anomalies, make corresponding adjustments until the sample escapes simulation. The escape simulation is only an initial detection to adjust the sandbox and facilitate subsequent behavior analysis.
[0037] In this embodiment, multiple sandbox environments include Cuckoo, Noriben, Ether, etc. Sample escape simulation is to simulate the escape behavior of malicious samples in the sandbox, so as to adjust the sandbox. The hardware information, environment information, time information and interaction information of the sandbox are detected in sequence to see if there are any anomalies. The anomaly is the condition for the malicious sample to determine that the environment is a sandbox analysis environment. If the malicious sample is determined to be a sandbox analysis environment after analysis, it will not perform malicious behavior.
[0038] In order to improve the anti-escape effect, in some embodiments of the present application, if there is an abnormality, corresponding adjustments are made, including: if there is an abnormality in the hardware information, the hardware information of the real host is obtained, the difference hardware is determined based on the hardware information of the real host and the hardware information of the sandbox, and the sandbox hardware information is adjusted according to the difference hardware and its corresponding cost.
[0039] In this embodiment, when configuring the sandbox analysis environment, the malicious sample analyst is likely to allocate only one processor core to the sandbox according to the default configuration. This situation should not exist in the case of a real host. Generally, the processor has multiple cores in the real situation, which is related to the current hardware configuration. Malware can check the number of cores of the CPU processor to determine whether it is a virtual detection environment. Based on the hardware information of the real host and the hardware information of the sandbox, the difference hardware is determined. The difference hardware is the hardware information that exists in the real host, but does not exist in the sandbox, and the malware can discover the hardware information of this vulnerability. According to the difference hardware and its corresponding cost, the sandbox hardware information is adjusted to ensure that reasonable hardware information is adjusted under economically reasonable conditions to confuse the detection of malicious samples.
[0040] In order to improve the anti-escape effect, in some embodiments of the present application, if there is an abnormality, corresponding adjustments are made, including: if there is an abnormality in the environmental information, real environment characteristics and sandbox environment characteristics are obtained, and the real environment characteristics and sandbox environment characteristics are compared to determine the characteristics of the sandbox environment, and the environmental characteristics required for the sandbox operation are obtained. The sandbox environment characteristics are adjusted according to the characteristics of the sandbox environment and the environmental characteristics required for the sandbox operation.
[0041] In this embodiment, sandbox detection is to run malicious samples in a special virtual environment. The running environment has a single and obvious environmental feature, which is easy to be detected by malicious samples. Therefore, the fewer these features, the better. The features that exist in the sandbox environment are features that do not exist in the real environment, but exist in the sandbox environment. Among the features that exist in the sandbox environment, the environmental features necessary for sandbox operation are discarded, and the remaining features are deleted.
[0042] In order to improve the anti-escape effect, in some embodiments of the present application, if there is an anomaly, corresponding adjustments are made, including: if there is an anomaly in the time information, whether there is a sleep function or a loop function in the detection sample, determining the start time of the sandbox detection according to the sleep function or the loop function, obtaining the resources consumed by different sandbox detection durations, obtaining the effective detection time of multiple malicious samples, determining the balance amount based on the resources consumed by different sandbox detection durations and the effective detection time of multiple malicious samples, determining the duration of the sandbox detection based on the balance amount, and performing detection based on the start time of the sandbox detection and the duration of the sandbox detection.
[0043] In this embodiment, the time for the sandbox to detect malicious samples is often short, so once the malicious sample does not perform any operation in the sandbox detection time window, the sandbox will stop detecting and exit. Malicious samples often use sleep functions or loop functions to sleep or loop meaningless code to shield detection. The start time of the sandbox detection is determined according to the sleep function or loop function, and no detection is performed before this time. The effective detection time of the malicious sample is the time from the start of the sample's action until the abnormal behavior is detected. It is necessary to find a balance between the time consumption of resources and the detection time to improve reliability.
[0044] In order to further improve the anti-escape effect, in some embodiments of the present application, a balance amount is determined based on the resources consumed by different sandbox detection durations and the effective detection time of multiple malicious samples, and the duration of sandbox detection is determined based on the balance amount, including: determining multiple balance amounts based on the resources consumed by different sandbox detection durations, the effective detection time of multiple malicious samples and a first preset weight, taking the maximum value among the multiple balance amounts as the required balance amount, substituting the balance amount into a conversion formula to obtain the duration of sandbox detection; the conversion formula is, T=L*M, wherein T is the duration of sandbox detection, M is the balance amount, and L is the conversion coefficient, and the first preset weight and conversion coefficient corresponding to different sandboxes are different.
[0045] In this embodiment, M=first preset weight 1*consumed resources+first preset weight 2*effective detection time.
[0046] In some embodiments of the present application, if there is an abnormality, a corresponding adjustment is made, which also includes: if there is an abnormality in the interaction information, the interaction information of the real host is obtained, and the interaction information of the sandbox is set according to the interaction information of the real host.
[0047] In this embodiment, the interaction information includes mouse clicks, pop-up windows, mouse movements, etc. For example, a malicious sample will determine whether it is in a sandbox environment by detecting whether there is a mouse click operation. The sandbox simulates host interaction information shielding detection.
[0048] Step S102, after the sample escape simulation is passed, the sample to be detected is obtained, and the sample to be detected is run multiple times in multiple sandbox environments to obtain a behavior file.
[0049] In this embodiment, when the environment information, hardware information, time information and interaction information of the malicious sample are all normal, the sandbox formal detection is carried out. The behavior file includes file operation behavior, network communication behavior, process operation behavior, registry operation behavior, etc.
[0050] Step S103, obtain distance information according to the behavior file, determine the behavior abnormality between sandbox environments based on the distance information, and determine whether there is abnormal behavior based on the relationship between the behavior abnormality and the threshold. If there is abnormal behavior, obtain the system call sequence corresponding to the abnormal behavior, and locate the escape behavior according to the system call sequence.
[0051] In this embodiment, distance information is obtained based on multiple behavior files, and the degree of behavior abnormality represents the possibility of abnormal behavior. It is a conventional technology in the field to locate the escape behavior based on the system call sequence, which will not be repeated here.
[0052] In some embodiments of the present application, distance information is obtained according to the behavior file, and the degree of behavioral abnormality between sandbox environments is determined based on the distance information, including: the distance information includes horizontal distance and vertical distance, the horizontal distance is the distance between behavior files in different environments, and the vertical distance is the distance between behavior files in the same environment, and the degree of behavioral abnormality is obtained based on the difference between the horizontal distance and the vertical distance.
[0053] In some embodiments of the present application, whether there is abnormal behavior is determined based on the relationship between the behavior abnormality and the threshold, including: if the behavior abnormality exceeds the threshold, abnormal behavior exists; if the behavior abnormality does not exceed the threshold, abnormal behavior does not exist.
[0054] In this embodiment, the distance is calculated using the Jaccard algorithm, and the horizontal distance calculation formula is as follows:
[0055]
[0056] For the sets X and Y of behavior features in the behavior files generated in the two environments, calculate the proportion of the same behavior features in the two sets to all the behavior features in the two sets. A result of 1 indicates that the two sets are completely different sets, and a result of 0 indicates that the two sets are exactly the same sets.
[0057] use To represent the Jaccard distance of the behavior files generated in environment i and environment j, the abnormality S is obtained by calculating the difference between the Jaccard distance of the behavior files in the same environment and the Jaccard distance of the behavior files in different environments. The abnormality S indicates whether there is an escape behavior in the environment of different sandboxes. The value range is 0-1. The formula is as follows:
[0058]
[0059] If the abnormality exceeds the specified threshold, it is judged as escape behavior.
[0060] Step S104, obtaining evaluation parameters of each sandbox, establishing a threshold update array according to the evaluation parameters, and updating the threshold according to the threshold update array.
[0061] In this embodiment, in order to improve the detection accuracy, the threshold is updated in real time to ensure the accuracy of the escape behavior judgment.
[0062] In some embodiments of the present application, a threshold update array is established based on evaluation parameters, including: the evaluation parameters include accuracy, detection rate and error rate; the position order is determined based on the size of the accuracy rate, detection rate and error rate, and the threshold update array is constructed based on the accuracy rate, detection rate, error rate and position order.
[0063] In this embodiment, the accuracy, detection rate and error rate of the system are set to P1, P2 and P3 respectively. The accuracy rate is the ratio of the number of escape samples judged correctly to the sum of the number of escape samples judged correctly and the number of escape samples judged incorrectly. The detection rate is the ratio of the number of escape samples judged correctly to the sum of the number of escape samples judged as normal samples. The error rate is the ratio of the number of escape samples judged incorrectly to the number of normal samples.
[0064] The position order is determined based on the accuracy, detection rate, and error rate, specifically:
[0065] Assimilate the three, calculate the product of the three and the corresponding constants, compare the size values, the larger the value, the higher the position. Set the corresponding constants of the three to be K1, K2, and K3 respectively; the products of the three are P1*K1, P2*K2, and P3*K3 respectively. For example, if the three gradually decrease, then P1*K1>P2*K2>P3*K3, and the accuracy, detection rate, and error rate are the first, second, and third respectively. The threshold update array is (P1, P2, P3).
[0066] In some embodiments of the present application, updating the threshold value according to the threshold value includes: updating the array according to the threshold value and updating the threshold value according to the local factor corresponding to the position order.
[0067] In this embodiment, the local factors corresponding to the position order are (α1, α2, α3), and (α1P1, α2P2, α3P3) are obtained, and the influence amount Q=α1P1+α2P2+α3P3 is obtained. The original threshold (abnormality threshold in step S103) is updated according to the influence amount.
[0068] Set the influence value array Q0 (Q1, Q2, Q3, Q4), where Q1, Q2, Q3, Q4 are all preset values, and Q1<Q2<Q3<Q4;
[0069] Set the original threshold to E, set the update coefficient to F, and preset the update coefficient array F0 (F1, F2, F3, F4), F1, F2, F3, F4 are all preset values, and 0.7<F1<F2<F3<F4<1.3;
[0070] According to the relationship between the influence amount and each preset influence amount, an update coefficient is determined and updated;
[0071] If Q<Q1, determine the first preset update coefficient F1 as the update coefficient, and the updated threshold value is E*F1;
[0072] If Q1≤Q<Q2, determine the second preset update coefficient F1 as the update coefficient, and the updated threshold value is E*F2;
[0073] If Q2≤Q<Q3, determine the third preset update coefficient F3 as the update coefficient, and the updated threshold value is E*F3;
[0074] If Q3≤Q<Q4, the fourth preset update coefficient F4 is determined as the update coefficient, and the updated threshold value is E*F4.
[0075] By applying the above technical solution, a sample escape simulation is performed on multiple sandbox environments, and the hardware information, environment information, time information and interaction information of the sandbox are detected in turn to see if there are any anomalies. If there are any anomalies, corresponding adjustments are made until the sample escape simulation is passed; after the sample escape simulation is passed, the sample to be detected is obtained, and the sample to be detected is run multiple times in multiple sandbox environments respectively, and a behavior file is obtained; distance information is obtained according to the behavior file, and the degree of abnormal behavior between the sandbox environments is determined based on the distance information. Based on the relationship between the degree of abnormal behavior and the threshold, it is determined whether there is abnormal behavior. If there is abnormal behavior, the system call sequence corresponding to the abnormal behavior is obtained, and the escape behavior is located according to the system call sequence; the evaluation parameters of each sandbox are obtained, and a threshold update array is established according to the evaluation parameters, and the threshold is updated according to the threshold update array. The present application detects defects in the sandbox through escape simulation and adjusts the sandbox, thereby deceiving the environment detection of malicious samples. According to the behavior files in multiple sandboxes and multiple behavior files in one sandbox, abnormal behavior is determined and the system call sequence is obtained to locate the escape behavior. The detection accuracy is improved, the malicious samples are prevented from escaping detection, and the reliability of detection is guaranteed.
[0076] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present invention can be implemented by hardware, or by software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present invention.
[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements 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 application.
Claims
1. An anti-escape detection method, characterized in that: The method comprises: Perform sample escape simulation on multiple sandbox environments, and detect whether there are any anomalies in the sandbox's hardware information, environment information, time information, and interaction information. If there are any anomalies, make corresponding adjustments until the sample escapes simulation; After the sample escape simulation passes, obtain the sample to be tested, run it multiple times in multiple sandbox environments, and obtain the behavior file; Obtain distance information based on the behavior file, determine the degree of abnormal behavior between sandbox environments based on the distance information, and determine whether there is abnormal behavior based on the relationship between the degree of abnormal behavior and the threshold. If there is abnormal behavior, obtain the system call sequence corresponding to the abnormal behavior, and locate the escape behavior based on the system call sequence; Obtain evaluation parameters of each sandbox, establish a threshold update array according to the evaluation parameters, and update the threshold according to the threshold update array; If there are any abnormalities, corresponding adjustments will be made, including: If there is an anomaly in the hardware information, the hardware information of the real host is obtained, the difference hardware is determined based on the hardware information of the real host and the hardware information of the sandbox, and the sandbox hardware information is adjusted according to the difference hardware and its corresponding cost; Corresponding adjustments will be made, including: If there is an abnormality in the environmental information, the real environment characteristics and the sandbox environment characteristics are obtained, and the real environment characteristics and the sandbox environment characteristics are compared to determine the characteristics of the sandbox environment, obtain the environmental characteristics necessary for the sandbox operation, and adjust the sandbox environment characteristics according to the characteristics of the sandbox environment and the environmental characteristics necessary for the sandbox operation; Corresponding adjustments will be made, including: If there is an anomaly in the time information, detect whether there is a dormant function or a loop function in the sample, determine the start time of the sandbox detection according to the dormant function or the loop function, obtain the resources consumed by different sandbox detection durations, obtain the effective detection time of multiple malicious samples, determine the balance amount based on the resources consumed by different sandbox detection durations and the effective detection time of multiple malicious samples, determine the duration of the sandbox detection based on the balance amount, and perform detection based on the start time and duration of the sandbox detection; The balance is determined based on the resources consumed by the sandbox at different detection times and the effective detection time of multiple malicious samples. The duration of the sandbox detection is determined based on the balance, including: Determine multiple balancing amounts based on resources consumed by different sandbox detection durations, multiple malicious sample effective detection times, and a first preset weight, take the maximum value among the multiple balancing amounts as the required balancing amount, and substitute the balancing amount into a conversion formula to obtain the duration of the sandbox detection; The conversion formula is, T=L*M, where T is the duration of the sandbox detection, M is the balance amount, and L is the conversion coefficient. The first preset weight and conversion coefficient corresponding to different sandboxes are different; If there are any abnormalities, corresponding adjustments will be made, including: If there is an abnormality in the interaction information, the interaction information of the real host is obtained, and the interaction information of the sandbox is set according to the interaction information of the real host; Distance information is obtained according to the behavior file, and the degree of behavioral abnormality between sandbox environments is determined based on the distance information, including: The distance information includes horizontal distance and vertical distance. The horizontal distance is the distance between behavior files in different environments, and the vertical distance is the distance between behavior files in the same environment. The behavior abnormality is obtained based on the difference between the horizontal distance and the vertical distance. Establish a threshold update array based on the evaluation parameters, including: Evaluation parameters include accuracy, detection rate, and error rate; Determine the position order based on the accuracy, detection rate and error rate, and build a threshold update array based on the accuracy, detection rate, error rate and position order; Assimilate the accuracy, detection rate and error rate, calculate the product of the three with the corresponding constants, compare the values, and the larger the value, the higher the position. Update the threshold array according to the threshold, including: Update the thresholds based on the threshold update array and the local factors corresponding to the position order.
2. The method according to claim 1, characterized in that Determine whether there is abnormal behavior based on the relationship between the abnormality of the behavior and the threshold, including: If the degree of behavioral abnormality exceeds the threshold, abnormal behavior exists; If the degree of abnormal behavior does not exceed the threshold, there is no abnormal behavior.
