Transaction platform user access method and system based on SSLVPN gateway

Abstract

The present application relates to the field of SSL VPN gateway access technology, and specifically to a method and system for user access to a trading platform based on an SSL VPN gateway. The method comprises: a user terminal establishes an encrypted tunnel with an SSL VPN gateway through the SSL / TLS protocol, performs multi-factor authentication on the user, and the user accesses the trading platform after passing the multi-factor authentication; determines the user's credit score, device health score, environmental risk score, and behavioral risk score; when any of the scores does not meet a preset condition, rejects the user request and disconnects the access; otherwise, calculates the user's current trust score in real time; introduces time decay to the trust score and corrects the user's current trust score; defines a three-dimensional permission matrix of the trading platform and obtains permission values; when the permission is insufficient, the user is required to perform the multi-factor authentication twice, and if the authentication passes, the operation permission is granted; otherwise, the user access is disconnected. This improves the user access security of the trading platform.

Description

Technical Field

[0001] The present application relates to the field of SSLVPN gateway access technology, and in particular to a method and system for user access to a trading platform based on an SSLVPN gateway. Background Art

[0002] With the increasing prevalence of digital trading platforms, users need to securely access internal trading systems via the internet to ensure the confidentiality and integrity of trading instructions and data during transmission, as well as the legitimacy of user identities. Traditional network access methods face security risks such as data leakage, identity forgery, and privilege abuse, and they struggle to meet the financial industry's stringent compliance and real-time requirements. This is why user access solutions based on SSLVPN gateways have emerged. Their core approach is to establish secure and reliable remote access channels through SSL / TLS encrypted tunnels, preventing users from directly accessing trading platforms. This addresses the critical issue of secure access to trading platforms from distributed terminals and supports secure access requirements in high-concurrency and highly sensitive business scenarios.

[0003] However, existing SSLVPN user access is mostly based on a static trust model of "one-time authentication, permanent trust", which does not take into account the dynamic changes in user behavior, real-time fluctuations in device security status, dynamic evolution of environmental risks, and differences in operational risk levels. As a result, it is impossible to dynamically adjust permissions based on real-time risks, which can easily lead to security risks such as access of defective devices, abuse of permissions, and missed detection of abnormal operations. It is difficult to meet the trading platform's requirements for high security, fine-grained control, and continuous compliance. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a trading platform user access method and system based on SSLVPN gateway. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for user access to a trading platform based on an SSLVPN gateway, the method comprising the following steps:

[0006] The user terminal establishes an encrypted tunnel with the SSLVPN gateway through the SSL / TLS protocol. During the tunnel establishment process, the user undergoes multi-factor authentication. After passing the authentication, the user can access the trading platform.

[0007] When a user requests an operation on the trading platform:

[0008] Based on the user's historical transaction data on the trading platform, a neural network model is used to determine the user's credit score. The security agent program on the user's terminal collects the underlying status data of the platform's access devices, and uses vulnerability risk assessment to obtain the user's device health score.

[0009] The Bayesian network is used to obtain the user's current environmental risk score based on the user's current trading platform login IP reputation, login location, and access network type. The isolation forest algorithm is used to obtain the user's current behavior risk score based on the deviation between the user's real-time operation data and historical operation data on the trading platform.

[0010] If any of the user's credit score, device health score, environmental risk score, and behavioral risk score does not meet the preset conditions, the user's request will be rejected and access will be disconnected. Otherwise, the user's current trust score will be calculated in real time based on the user's credit score, device health score, environmental risk score, and behavioral risk score.

[0011] Introducing time decay into the current trust score, and adjusting the user's current trust score in real time based on the risk level of the user's request type; defining a three-dimensional permission matrix for the trading platform, including user roles, trust scores, and request types;

[0012] Based on the current user's role, the corrected trust score, and the request type, the permission value of the current user's request in the three-dimensional permission matrix is ​​obtained; if the permission is insufficient, the user is required to perform the multi-factor authentication twice, and the operation permission is granted if the authentication is passed; otherwise, the user's access is disconnected;

[0013] The underlying status data includes the installed patch list and running process hash value of the user access device, identifies unpatched known vulnerabilities based on the device's patch information, and simultaneously matches the malware feature library by running process hash to obtain the device health score based on the degree of vulnerability damage.

[0014] In one embodiment, the multi-factor authentication includes identity credential verification, device fingerprint recognition, and geographic location verification.

[0015] In one embodiment, the historical transaction data includes: the user's daily transaction times, daily authentication failure times, and daily unauthorized access times.

[0016] In one embodiment, obtaining the environmental risk score includes:

[0017] Use MISP threat intelligence to obtain the reputation level of the user's current login IP address and the distance between the user's current IP login location and historically frequently used login locations;

[0018] The conditional probability table is trained through the IP reputation level of historical attack events, the location distance of historical attack events, and the access network type. Combined with the Bayesian network, the posterior probability of the environmental risk of the current user login is determined through the joint tree algorithm to obtain the environmental risk score.

[0019] In one embodiment, obtaining the behavior risk score includes:

[0020] The real-time operation data includes the ratio of the user's current daily transaction amount to the average daily transaction amount over multiple historical days, the number of abnormal daily interface call frequencies, and the number of daily transactions; the historical operation data includes the ratio of the user's daily transaction amount at historical times to the average daily transaction amount over multiple historical days, the number of abnormal daily interface call frequencies, and the number of daily transactions;

[0021] The behavioral risk score of the user's current operation is determined based on the anomaly score output by the isolation forest algorithm.

[0022] In one embodiment, the preset conditions are: the user credit score is less than a preset threshold, the device health score is less than a preset threshold, the environmental risk score is greater than a preset threshold, and the behavioral risk score is greater than a preset threshold.

[0023] In one embodiment, the real-time calculation of the user's current trust score includes:

[0024] Calculate the product of the user's credit score and the device health score, recorded as the first product, calculate the product of the current environmental risk score and the behavioral risk score, recorded as the second product, and determine the sum of the second product and a preset value greater than 0. The user's current trust score is the normalized result of the ratio of the first product to the sum.

[0025] In one embodiment, the real-time modification of the user's current trust score includes:

[0026] Determine a risk coefficient of the user's request type based on the risk level of the user's request type, wherein the risk coefficient ranges from (-1 to 1). The higher the risk level of the user's request type, the closer the risk coefficient is to -1.

[0027] The current revised trust score calculation method is: Where, is the corrected trust score of the user at time t, is the user's trust score at the moment before the current moment t, e is a natural constant, is the preset attenuation coefficient, The time interval between the user's current moment and the last multi-factor authentication. The number of user requests between the current time and the last multi-factor authentication. is the risk factor of the user's i-th request operation.

[0028] In the second aspect, an embodiment of the present application also provides a trading platform user access system based on an SSLVPN gateway, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above methods when executing the computer program.

[0029] This application has at least the following beneficial effects:

[0030] The user terminal of this application establishes an encrypted tunnel with the SSLVPN gateway through the SSL / TLS protocol, and performs multi-factor authentication on the user during the tunnel establishment process. After the authentication is passed, the user accesses the trading platform; it significantly improves the authenticity verification strength of the user identity, effectively avoids illegal access caused by credential theft, and improves transaction security; by determining the user's credit score, device health score, environmental risk score, and behavioral risk score; the four-dimensional dynamic score improves the accuracy of risk assessment, enhances the quantification ability of user behavior credibility, realizes the accurate perception of device security status, enhances the recognition sensitivity of abnormal environment, and improves the early warning ability of attack behavior; combines the user's credit score, device health score, environmental risk score, and behavioral risk score to calculate the user's current trust score in real time; introduces time decay to the current trust score, and corrects the user's current trust score in real time based on the risk level of the user's request type. trust score before; solves the risk of permission retention after session hijacking, and improves the accuracy and security of permission granting; defines the three-dimensional permission matrix of the trading platform, including user roles, trust scores and request types; breaks through the coarse-grained defects of the traditional RBAC model and realizes refined permission control; this application breaks through the limitations of traditional SSLVPN static authentication "one-time authentication, permanent trust", and solves the fixed permission allocation and risk problems caused by static authentication in traditional access methods through dynamic trust evaluation and adaptive permission control. It can cope with dynamic changes in user behavior, device status, and environmental risks, realize dynamic matching of permissions and risks, achieve fine-grained management and control, effectively resist advanced threats such as credential leakage and device hijacking, meet the trading platform's requirements for high security and continuous compliance, block unauthorized access and data leakage risks from the source, and achieve a dynamic balance between user experience and security protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0032] Figure 1A flowchart of the steps of a method for accessing a trading platform user based on an SSLVPN gateway provided in one embodiment of the present application;

[0033] Figure 2 Flowchart for obtaining permissions to request operations on the trading platform. DETAILED DESCRIPTION

[0034] To further illustrate the technical means and effectiveness of this application to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of the SSLVPN gateway-based trading platform user access method and system proposed in this application. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0035] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0036] The following describes in detail the specific solutions of the trading platform user access method and system based on SSLVPN gateway provided by this application with reference to the accompanying drawings.

[0037] See also Figure 1 , which shows a flowchart of a method for user access to a trading platform based on an SSLVPN gateway provided by an embodiment of the present application, the method comprising the following steps:

[0038] S1, the user terminal establishes an encrypted tunnel with the SSLVPN gateway through the SSL / TLS protocol, and performs multi-factor authentication on the user during the tunnel establishment process.

[0039] The user terminal initiates a connection request to the SSLVPN gateway using the SSL / TLS protocol. The SSLVPN gateway first negotiates encryption parameters with the terminal to establish an end-to-end encrypted tunnel. After the tunnel is established, the SSLVPN gateway performs a multi-factor authentication mechanism, which is a traditional static authentication method. In this embodiment, the multi-factor authentication mechanism includes:

[0040] Identity credential verification: Verify the username / password entered by the user and verify the legitimacy of the credentials in conjunction with the OAuth protocol;

[0041] Device fingerprint recognition: Collect device fingerprints, including hardware UUID, operating system hash value, and browser fingerprint, and compare them with the pre-registered list of trusted devices to ensure the uniqueness of the access device;

[0042] Geographic location verification: Verify the login location through the IP geolocation database, and combine the user's historical login trajectory to trigger additional verification for remote login behavior.

[0043] If the multi-factor authentication is passed, the user can access the trading platform. Otherwise, the user is denied access to the trading platform. After accessing the trading platform, the user can request operations in the trading platform.

[0044] S2, based on the user's historical transaction data on the trading platform, uses a neural network model to determine the user's credit score; collects the underlying status data of the platform access device through the security agent program of the user terminal, and uses vulnerability risk assessment to obtain the user's device health score.

[0045] Traditional trading platform access relies on static credentials and one-time authentication, which cannot address the dynamically changing security risks in trading scenarios. A zero-trust architecture requires continuous verification and never trusting, but a single-dimensional authentication factor makes it difficult to comprehensively assess access risks. Therefore, this embodiment integrates multi-dimensional data in real time to build a dynamically changing trust assessment system. This addresses the security blind spot in traditional solutions where successful authentication confers complete trust, protecting against advanced threats such as credential leakage, device hijacking, and session hijacking.

[0046] This embodiment first determines the user's credit score based on the user's historical transaction data on the trading platform using a neural network model, specifically including:

[0047] The user's transaction data from the trading platform over the past 30 days is collected. In this embodiment, this transaction data includes the daily number of abnormal transactions, authentication failures, and unauthorized accesses. The transaction data is normalized using a sigmoid function and converted into numerical features ranging from 0 to 1. The normalized number of abnormal transactions, authentication failures, and unauthorized accesses is then combined into a daily three-dimensional array. The three-dimensional array of the user's daily data from the past 30 days is then arranged chronologically to form a behavior sequence. The behavior sequence is input into a trained long-short-term memory (LSTM) neural network model. The output of the LSTM model is linearly mapped to a range of 0 to 100, which serves as the user's credit score. A higher score indicates higher creditworthiness. The use of an LSTM model to obtain user credit scores is well known in the art, and the specific architecture and training process are not described in detail here. The 30-day period is only one example of this application; implementers can set it based on their actual needs. This embodiment does not limit this.

[0048] Secondly, the security agent program on the user terminal collects the underlying status data of the platform access device and uses vulnerability risk assessment to obtain the user's device health score, specifically:

[0049] The user terminal's security agent collects underlying status data from the user's access device. In this embodiment, this underlying status data includes a list of installed patches and hash values ​​of running processes. This encrypted data is then transmitted to the device security monitoring server to generate a health score. The device security monitoring server compares the device patch information with the NVD library and the CVE list to identify known, unpatched vulnerabilities. It also matches the running process hashes against a malware signature database and assigns a weight based on the vulnerability's severity. In this embodiment, vulnerability severity is assessed using the CVSS 3.1 scale. A CVSS score greater than 9.0 is considered high-risk, a CVSS score between 7.0 and 8.9 is considered medium-risk, and a CVSS score less than 7.0 is considered low-risk. In this embodiment, the device health score is set to a total of 100 points. For each high-risk vulnerability, 30 points are deducted; for each medium-risk vulnerability, 15 points are deducted; and for each low-risk vulnerability, 5 points are deducted. The server outputs the device health score; a higher value indicates a healthier access device.

[0050] S3 uses the Bayesian network to obtain the user's current environmental risk score through the IP reputation, login location, and access network type of the user's current trading platform login. Based on the deviation between the user's real-time operation data and historical operation data on the trading platform, the isolation forest algorithm is used to obtain the user's current operation behavior risk score.

[0051] First, the reputation rating of the user's current IP address logged into the trading platform is obtained by calling the MISP threat intelligence API. The MaxMind GeoIP database is used to determine the city corresponding to the user's current IP address and compare it with the user's historical login locations. The geographic distance is calculated using the Haversine formula. If the geographic distance exceeds 500 kilometers, the remote login risk is flagged. Network layer protocol detection is used to distinguish the type of access network the user is logging into. A Bayesian network is constructed, consisting of three parent nodes: IP reputation rating, geographic distance, and network type, and a child node: environmental risk. The conditional probability table (CPT) is trained using the IP reputation, geographic distance, and network type corresponding to historical attack events. Each time a user accesses the platform, the real-time IP reputation rating, geographic distance, and network type are input into the Bayesian network. The posterior probability of "environmental risk = high / medium / low" is calculated using a junction tree algorithm. This is then converted into a risk score from 0 to 100 via linear mapping, with higher values ​​indicating a more dangerous environment.

[0052] Secondly, collect the real-time operation data of users on the trading platform, construct a multi-dimensional feature vector, and record it as a behavioral feature vector. In this embodiment, the real-time operation data includes the ratio of the current daily transaction amount to the average daily transaction amount in the past 7 days, the number of abnormal daily interface call frequency, and the number of daily transactions. The ratio at the current moment, the number of abnormal daily interface call frequency, and the number of daily transactions are all recorded as features, and all features are combined into the behavioral feature vector at the current moment. Based on the user's behavioral feature vector in the past 90 days, the sliding window statistical method is used to generate the baseline range of each feature, and the baseline range is ,in, is the mean, The Isolation Forest algorithm is used to detect the anomaly score of the real-time behavior feature vector. The anomaly score is linearly mapped to a behavioral risk score from 0 to 100, with larger values ​​indicating more dangerous user behavior. The Isolation Forest algorithm is a well-known technique, and the specific process is not described here. The past 7 days and the past 90 days are examples of this application. Implementers can set them according to actual circumstances, and this embodiment does not impose any restrictions on this.

[0053] It should be noted that user credit scores and device health scores are static trustworthiness. The risk of overstepping authority for long-term compliant users is significantly lower than that for users with a history of violations. The probability of devices that meet the security baseline being infected with malware is much lower than that of devices with unpatched vulnerabilities, thereby enhancing the robustness of the trust assessment system. However, there is a defect that cannot perceive changes in the network environment and abnormal user behavior in real time. Therefore, this embodiment combines the dynamic risk of environmental risk scores and behavioral risk scores to improve the transaction platform's anti-attack capabilities and access efficiency in an open network environment. Static trustworthiness is calculated once after multi-factor authentication is passed, while dynamic risk needs to be calculated in real time after multi-factor authentication is passed.

[0054] S4: When any of the user's credit score, device health score, environmental risk score, and behavioral risk score does not meet the preset conditions, the user's request is rejected and access is disconnected; otherwise, the user's current trust score is calculated in real time based on the user's credit score, device health score, environmental risk score, and behavioral risk score.

[0055] If any of the following four conditions are met: a user's credit score is less than a preset threshold, a device health score is less than a preset threshold, an environmental risk score is greater than a preset threshold, or a behavioral risk score is greater than a preset threshold, the user's transaction request on the trading platform will be rejected and the trading platform's access port will be disconnected. In this embodiment, the preset threshold is 60, which can be set by the implementer based on actual circumstances and is not limited in this embodiment.

[0056] This embodiment combines static credibility and dynamic risk to calculate the user's current trust score in real time, reflecting the inherent security level of the user subject and capturing real-time threat changes during the session, breaking through the limitations of traditional SSLVPN static authentication.

[0057] The calculation formula for the trust score is ,in, is the user's trust score at the current moment t, norm() represents the normalization function, 、 Represents the user's credit score and the user's device health score respectively, 、 They represent the user’s environmental risk score and behavioral risk score at the current moment t, To preset a value greater than 0, avoid the denominator being 0. , the implementer can set it according to the actual situation, and this embodiment does not limit it. Denote it as the first product, Recorded as the second product.

[0058] It should be noted that the trust score is rounded to one decimal place. The larger the trust score, the lower the risk of the user and the more trustworthy the user is, and the greater the permissions can be granted to the user. On the contrary, the higher the risk of the user, the less trustworthy the user is, and the user's permissions need to be reduced.

[0059] S5 introduces time decay to the current trust score, and adjusts the user's current trust score in real time based on the risk level of the user's request type; defines the three-dimensional permission matrix of the trading platform, including user roles, trust scores, and request types.

[0060] Traditional role-based access control (RBAC) uses a static permission allocation model, where user permissions are determined solely by their roles. This model cannot address the real-time changing risks in transaction scenarios. Therefore, this embodiment allocates permissions based on dynamic trust scores, avoiding over-trust or under-trust and achieving a dynamic match between permissions and risks.

[0061] A three-dimensional permission matrix is ​​defined, consisting of user role R, trust score T, and request type O. User roles are a predefined set of roles based on organizational structure and business responsibilities, each associated with different permissions. For example, R = 1 for regular trader, R = 2 for risk manager, and R = 3 for system administrator. The trust score reflects the user's current trustworthiness, ranging from 0 to 1. Request type categorizes requests for different functional modules of the trading platform, with each request associated with a different risk level. For example, O = 1 for data query, O = 2 for order creation, O = 3 for fund transfer, and O = 4 for system configuration. Implementers assign values ​​to the elements in the three-dimensional permission matrix based on the specific circumstances of the trading platform: 1 indicates permission, 0.5 indicates secondary authentication is required, and 0 indicates denial. For example, when R = 1, T = 0.8, and O = 1, the corresponding element in the three-dimensional permission matrix is ​​assigned a value of 1, indicating that regular traders are allowed to perform data queries. The three-dimensional permission matrix is ​​a well-known technique and can be customized by implementers based on their specific circumstances.

[0062] After the user performs the requested operation, the corresponding permission value can be queried through the three-dimensional permission matrix, realizing multi-dimensional refined management and dynamic risk adaptation of the trading platform access control, changing the traditional static authorization model based on fixed roles, and enabling permissions to be dynamically adjusted with real-time changes in user behavior, device status, and environmental risks.

[0063] Furthermore, based on the principle of never trusting and continuously verifying in the zero-trust architecture, the gateway needs to continuously monitor user behavior and environmental changes and adjust permissions, which in this embodiment includes permission reduction, secondary authentication and session termination.

[0064] Permission reduction: The trust score only reflects the user's current risk status and cannot reflect the impact of natural trust loss over time and operation frequency. However, even if the current status is compliant, the risk of unverified sessions for a long time will increase. This embodiment simulates the natural accumulation of session risk by introducing real-time loss of trust scores, forcing the scope of permissions to gradually shrink as trust decreases. This avoids security vulnerabilities such as unauthorized abuse and session hijacking caused by long-term static permissions, ultimately achieving a dynamic balance between user experience and security protection.

[0065] The calculation formula of dynamic trust score which introduces trust score decay over time is expressed as , where is the corrected trust score of the user at time t, is the user's trust score at the moment before the current moment t, e is a natural constant, is the preset attenuation coefficient, which ranges from 0 to 1. , The time interval between the user's current moment and the last multi-factor authentication. The number of user requests between the current time and the last multi-factor authentication. is the risk factor of the user's i-th request operation.

[0066] It should be noted that the risk coefficient ranges from -1 to 1. According to the specific request operation settings, the request operation types set in this embodiment include data query, order creation, fund transfer, and system configuration operations. Among them, the risk level of data query, order creation, fund transfer, and system configuration operations gradually increases. When the risk of the request operation type is higher, the value of the risk coefficient is set closer to -1. The purpose is to reduce the user's trust score and ensure transaction security. When the risk of the request type is lower, the value of the risk coefficient is set closer to 1, which can slow down the rate at which the trust score decays over time. Therefore, this embodiment sets the risk coefficient of data query to 0.8, the risk coefficient of order creation to 0.6, the risk coefficient of fund transfer to -0.8, and the risk coefficient of system configuration operation to -0.9. By converting operational behavior into trust supplements, compliance operations can offset some of the natural decay, maintain a reasonable level of trust, reflect the dynamic adjustment logic that recent operations are more weighted than historical states, balance security and user experience, and avoid repeated authentication.

[0067] S6, based on the current user's role, the corrected trust score, and the request type, obtain the permission value of the current user's request in the three-dimensional permission matrix; when the permission is insufficient, the user is required to perform the multi-factor authentication twice, and the operation permission is granted if the authentication is passed, otherwise the user access is disconnected.

[0068] The permission value of the current user's request in the three-dimensional permission matrix is ​​obtained through the current user's role, the corrected trust score, and the request type. When the permission value is 1, it means that the user has permission and the operation is allowed. When the permission value is 0.5, it means that the permission is insufficient and a secondary multi-factor authentication is required. If the authentication passes, the permission is granted. If the authentication fails, the connection with the trading platform is disconnected. When the permission value is 0, the connection with the trading platform is also disconnected to ensure security. In this way, user access control of the trading platform is achieved. The flowchart of the permission acquisition process for the trading platform request operation is as follows: Figure 2 shown.

[0069] Based on the same inventive concept as the above method, an embodiment of the present application also provides a trading platform user access system based on SSLVPN gateway, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-mentioned trading platform user access methods based on SSLVPN gateway.

[0070] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0071] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0072] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A user access method for a trading platform based on an SSLVPN gateway, characterized in that: The method comprises the following steps: The user terminal establishes an encrypted tunnel with the SSLVPN gateway through the SSL / TLS protocol. During the tunnel establishment process, the user undergoes multi-factor authentication. After passing the authentication, the user can access the trading platform. When a user requests an operation on the trading platform: Based on the user's historical transaction data on the trading platform, a neural network model is used to determine the user's credit score. The security agent program on the user's terminal collects the underlying status data of the platform's access devices, and uses vulnerability risk assessment to obtain the user's device health score. The Bayesian network is used to obtain the user's current environmental risk score based on the user's current trading platform login IP reputation, login location, and access network type. The isolation forest algorithm is used to obtain the user's current behavior risk score based on the deviation between the user's real-time operation data and historical operation data on the trading platform. If any of the user's credit score, device health score, environmental risk score, and behavioral risk score does not meet the preset conditions, the user's request will be rejected and access will be disconnected. Otherwise, the user's current trust score will be calculated in real time based on the user's credit score, device health score, environmental risk score, and behavioral risk score. Introducing time decay into the current trust score, and adjusting the user's current trust score in real time based on the risk level of the user's request type; defining a three-dimensional permission matrix for the trading platform, including user roles, trust scores, and request types; Based on the current user's role, the corrected trust score, and the request type, the permission value of the current user's request in the three-dimensional permission matrix is ​​obtained; if the permission is insufficient, the user is required to perform the multi-factor authentication twice, and the operation permission is granted if the authentication is passed; otherwise, the user's access is disconnected; The underlying status data includes the installed patch list and running process hash value of the user access device, identifies unpatched known vulnerabilities based on the device's patch information, and simultaneously matches the malware feature library by running process hash to obtain the device health score based on the degree of vulnerability damage.

2. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The multi-factor authentication includes identity credential verification, device fingerprint recognition and geographic location verification.

3. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The historical transaction data includes: the user's daily transaction times, daily authentication failure times, and daily unauthorized access times.

4. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The acquisition of the environmental risk score includes: Use MISP threat intelligence to obtain the reputation level of the user's current login IP address and the distance between the user's current IP login location and historically frequently used login locations; The conditional probability table is trained through the IP reputation level of historical attack events, the location distance of historical attack events, and the access network type. Combined with the Bayesian network, the posterior probability of the environmental risk of the current user login is determined through the joint tree algorithm to obtain the environmental risk score.

5. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The acquisition of the behavioral risk score includes: The real-time operation data includes the ratio of the user's current daily transaction amount to the average daily transaction amount over multiple historical days, the number of abnormal daily interface call frequencies, and the number of daily transactions; the historical operation data includes the ratio of the user's daily transaction amount at historical times to the average daily transaction amount over multiple historical days, the number of abnormal daily interface call frequencies, and the number of daily transactions; The behavioral risk score of the user's current operation is determined based on the anomaly score output by the isolation forest algorithm.

6. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The preset conditions are: the user credit score is less than the preset threshold, the device health score is less than the preset threshold, the environmental risk score is greater than the preset threshold, and the behavioral risk score is greater than the preset threshold.

7. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The real-time calculation of the user's current trust score includes: Calculate the product of the user's credit score and the device health score, recorded as the first product, calculate the product of the current environmental risk score and the behavioral risk score, recorded as the second product, and determine the sum of the second product and a preset value greater than 0. The user's current trust score is the normalized result of the ratio of the first product to the sum.

8. The method for accessing a trading platform user based on an SSLVPN gateway according to claim 1, wherein: The real-time correction of the user's current trust score includes: Determine a risk coefficient of the user's request type based on the risk level of the user's request type, wherein the risk coefficient ranges from (-1 to 1). The higher the risk level of the user's request type, the closer the risk coefficient is to -1. The current revised trust score calculation method is: Where, is the corrected trust score of the user at time t, is the user's trust score at the moment before the current moment t, e is a natural constant, is the preset attenuation coefficient, The time interval between the user's current moment and the last multi-factor authentication. The number of user requests between the current time and the last multi-factor authentication. is the risk factor of the user's i-th request operation.

9. A trading platform user access system based on an SSL VPN gateway, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images