Network data security processing method and platform
By classifying, modeling, and encrypting medical big data, and combining latency detection and anomaly event classification neural networks, the security risks in the network transmission of medical big data have been solved, and the security and integrity of data transmission have been achieved.
Patent Information
- Application Number
- CN202411059970.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-05
AI Technical Summary
The lack of effective security monitoring in the network transmission of traditional Chinese medicine big data in current technology has resulted in the failure to effectively address the problems of leakage and tampering of personal health data.
By classifying and modeling personal health data, using encryption algorithms to extract features and encrypt access requests, and performing latency detection and reverse evaluation at the data transmission and reception ends, combined with anomaly event classification neural networks for intelligent classification, the system can identify and protect against network attacks.
It improves the security of medical big data network transmission, reduces data transmission volume, reduces data indicativeness, and enhances the ability to identify abnormal events through intelligent classification networks, thus ensuring the security and integrity of data transmission.
Smart Images

Figure CN118631585B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a network data security processing method and platform, belonging to the security management technology of medical big data. Background Art
[0002] With the explosive growth of medical big data, personal health information is being collected, stored, used, and processed on a large scale. The leakage of this massive amount of personal health information could lead not only to economic losses but also to social problems. Therefore, it is imperative to monitor the security of medical data network transmission, identify potential security risks, and prevent the leakage and tampering of personal health data. Summary of the Invention
[0003] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a network data security processing method and platform to conduct security supervision on the network transmission of personal health information in medical big data, determine whether there are security risks, and avoid the occurrence of problems such as leakage and tampering of personal health data.
[0004] Technical solution: To achieve the above purpose, the technical solution adopted by the present invention is:
[0005] A method for securely processing network data comprises the following steps:
[0006] Step 1: Divide personal health data into Category, the The health data of an individual in all categories is represented as , , Represents the total number of individuals; establish data models for different categories of personal health data, and build a list of personal health data models. Individual The personal health data of each category is represented by the data model: , For the data model The input features of the categories represent the dimensions of an individual's health data. The larger the dimension, the more detailed and comprehensive the information represented. Generally speaking, it should include basic personal information (gender, age, blood type, etc.), chronic disease information, allergy information, surgical information, etc.
[0007] Step 2: Extract features of access requests and obtain information about 、 Target sequence , using encryption algorithm I to encrypt the target sequence Encrypt to get encrypted data ; Build an access data packet. The header of the access data packet contains control information and verification information. The user information of the access data packet contains encrypted data. , using the sending time As filling data, if the sending time in the filling data If it is incomplete, the complete sending time will be added to the reserved data. ;
[0008] Step 3: Parse the access data packet and extract the encrypted data and access time , if the access time If the delay is within the delay threshold, then: first encrypt the data Decrypt to obtain the target sequence ; Then according to the target sequence Query the personal health database and construct the target feature sequence ; Then query the list of personal health data models, according to the target sequence Retrieve the corresponding data model and combine it with the target feature sequence , build the target data model sequence ; Use encryption algorithm II to encrypt the target data model sequence Encrypt to get encrypted data ; Construct a feedback data packet. The header of the feedback data packet contains control information and verification information. The user information of the feedback data packet contains encrypted data. , use feedback time As filling data, if the feedback time in the filling data If it is incomplete, the complete feedback time will be added to the reserved data. ;
[0009] Step 4: Parse the feedback data packet and extract the encrypted data and feedback time , if the feedback time If the delay is within the delay threshold, then: first encrypt the data Decrypt to obtain the target data model sequence ; Then from the target data model sequence Extract feedback target sequence , if the feedback target sequence With target sequence If the target sequence is consistent, Query the list of personal health data models and retrieve the corresponding data model sequence for the target data model Parse and get the target feature sequence , using the target feature sequence Respond to access requests.
[0010] Because the types of personal health data are relatively limited, access to personal health data can generally be categorized as a query for a specific piece of personal information or a query for a specific category of information about a specific individual. Based on this, we store personal health data along two dimensions: the individual and the category of health data. The core of accessing personal health data can be seen as accessing from these two dimensions, then populating the retrieved data into the framework constructed by the access request to generate a response. Therefore, when sending an access request, this case does not employ a full assembly approach. Instead, it first extracts features from the access request to obtain a two-dimensional number for querying the personal health database. Based on the indication of the two-dimensional number, the data receiver feeds the corresponding data back to the data sender, which then assembles the data to form a response to the access request. To achieve this process, certain identical or corresponding configurations must be implemented on both the data requester and the data receiver to ensure the secure and efficient transmission of personal health data. It should be noted that the two-dimensional number provided in this case is not necessarily unique and fixed; it can also be a formula representing a range of two-dimensional numbers. The encryption algorithm I used in this method is an encryption algorithm for the target sequence, and the encryption algorithm II is an encryption algorithm for the target data model sequence. Both can adopt general encryption algorithms in the existing technology, such as symmetric encryption algorithms (AES, DES, 3DES), asymmetric encryption algorithms (RSA, DSA, ECC), linear hash algorithms (MD5, SHA1, HMAC), hybrid encryption, etc.; the two encryption algorithms can be the same or different.
[0011] In this case, the data requester did not directly send an access request. Instead, it first performed feature extraction on the access request, extracting the individual number and the corresponding group of personal health data category numbers. After combining all the group numbers into a target sequence, the access request was repackaged to form an access data packet. To enhance the security of the access data packet, the target sequence was encrypted, and a delay detection scheme was employed for preliminary security protection. After parsing the access request, the data receiver needed to decrypt the encrypted data, construct a target feature sequence based on the decrypted target sequence, query the list of personal health data models, find the data model for each target feature, and re-express the target feature using the target feature as input to the data model. Finally, the target data model sequence was constructed, and a feedback data packet was constructed using a similar method as the access data packet. This feedback data packet was then sent to the data requester. After parsing the feedback data packet, the data requester needed to decrypt the encrypted data and extract the target sequence. After verifying that the response matched the sent target sequence, the data requester would query the list of personal health data models to find the data model for each target feature. The target feature sequence was then parsed from the data model to respond to the access request.
[0012] In this case, the target sequence does not directly point to the data model, but rather indirectly through the list of personal health data models. This facilitates targeted adjustments to the data model, such as additions, deletions, and modifications. It should be noted that each type of personal health data does not require a single data model. A separate data model library can be established, and a data model from this library can be selected based on rules to represent a specific target feature.
[0013] Specifically, the Step 3 includes an attack evaluation step, which is performed according to the set frequency, specifically: using the same encryption algorithm I as Step 2 to encrypt the target sequence Encrypt to get encrypted data , using encrypted data Evaluate whether the user information of the access data packet carries attack data; Step 4 includes an attack evaluation step, which is performed according to the set frequency; specifically, the target data model sequence is encrypted using the same encryption algorithm II as Step 3. Encrypt to get encrypted data , using encrypted data Evaluate whether the user information in the feedback packet carries attack data. In this case, because the effective information transmitted is limited and clear, the packet size, required traffic flow, and other information can be estimated within a relatively precise range. If it exceeds this range, it can be preliminarily determined that the packet has been attacked. To further determine whether the packet has been attacked, this case employs a reverse evaluation method, which monitors the security of data transmission. The decoded target sequence is repackaged using the same method. If the packet has not been attacked or lost, the repackaged packet should be consistent with the received packet. This consistency can be evaluated using Euclidean distance. Because reverse evaluation requires time and computing power, the frequency of reverse evaluation can be set, or the conditions for reverse evaluation can be set, such as when a sudden change in traffic fluctuation occurs. Considering time constraints, reverse evaluation can be performed outside the communication link of the access request, and the decision to terminate the communication link can be made based on the evaluation results. A safer approach is to perform reverse evaluation on the communication link of the access request and terminate the communication link before the evaluation results are determined. However, this method will suffer from a certain loss of timeliness.
[0014] Specifically, if the access time If the delay exceeds the delay threshold range, or if the attack assessment step finds that the user information of the access data packet carries attack data, anomaly detection is started at the destination node (the receiving node of the access data packet), and the backtracking analysis of the access data packet starting from the destination node is started according to the authority; if the feedback time If the delay of the feedback data packet exceeds the delay threshold range, or if the attack assessment step finds that the user information of the feedback data packet carries attack data, anomaly detection is initiated at the source node (the receiving node of the feedback data packet), and backtracking analysis of the feedback data packet starting from the source node is initiated according to the authority; the anomaly detection detects whether there is abnormal traffic or abnormal communication within a set time range before and after the data packet is received, based on the data packet size, source address and destination address as characteristics. If abnormal traffic or abnormal communication exists, it is determined that an abnormal event has occurred; the backtracking analysis performs anomaly detection on the nodes through which the data packet flows along the data packet transmission link.
[0015] The abnormal events are implemented through the network or other technical means, aiming to cause abnormalities in the information system or potential harm to the current operation of the information system; abnormal events include but are not limited to denial of service attack events, backdoor attack events, vulnerability attack events, network scanning and eavesdropping events, phishing events, interference events, etc. Denial of service attack events are a type of attack that consumes the resources of the target server through a large number of requests, making it unable to process normal requests, resulting in normal users being unable to access services; backdoor attack events involve attackers exploiting backdoors or vulnerabilities in the system to access or control the target system without authorization; vulnerability attack events exploit configuration defects, protocol defects or program defects in the information system to attack, resulting in system abnormalities or potential harm; network scanning and eavesdropping events obtain sensitive information or interfere with normal communications by scanning the network or eavesdropping on communication content; phishing events send emails or messages by forging trusted sources to trick users into clicking malicious links or downloading malicious attachments in order to steal personal information or implant Trojans; interference events interfere with normal network services or communications through various means, affecting the normal operation of the system.
[0016] Specifically, the abnormal event classification neural network is used to classify abnormal events, which involves two steps: training and testing. During the training process, abnormal event attack packets are first used to generate abnormal packets. A feature extraction module is then used to extract behavioral features of the abnormal packets. The classification module is then used to classify the abnormal events of the attack packets based on the behavioral features, obtaining a predicted abnormal event label. A loss function is then used to supervise the training of the classification module and optimize its parameters. During the testing process, the feature extraction module is first used to extract behavioral features of the abnormal packets. The trained classification module is then used to classify the behavioral features and predict the abnormal event label. The abnormal event classification neural network is an intelligent classification network that is refined through training. During the training phase, the classification model is used to predict the type of abnormal event based on the attacked packets. The loss between the predicted results and the actual results is compared to evaluate the training success of the classification model. If the difference is large, the classification model parameters are adjusted to improve the classification model's accuracy. During the testing phase, packets with unknown attack status or attack type are input to predict whether the packet has been attacked and the attack type. New attack types can be incorporated into the training samples for learning.
[0017] Specifically, the behavioral characteristics include data size, source address, destination address, traffic flow, latency, and user data. Attacked data packets often exhibit sudden changes in traffic flow, new executed instructions, and latency, so these are considered behavioral characteristics. Furthermore, attacks on data packets typically have specific targets, so describing the relevant data sources is also crucial.
[0018] A network data security processing platform includes a personal health database, a data model database, a personal health data model list, an encryptor I, a decryptor I, an encryptor II, a decryptor II, a data packet organization module and a delay judgment module; the personal health database divides personal health data into Category, the The health data of an individual in all categories is represented as , , Represents the total number of individuals; the data model database is used to store data models of different categories of personal health data, Individual The personal health data of each category is represented by the data model: ; The personal health data model list is used to characterize the mapping relationship between the data model ID and the data model storage address; the encryptor I and decoder I are used to encrypt and decrypt the target sequence; the encryptor II and decryptor II are used to encrypt and decrypt the target data model sequence; the data packet organization module is used to construct and parse the access data packet and the feedback data packet; the delay judgment module is used to judge whether the delay of the access time and the feedback time exceeds the delay threshold range.
[0019] Specifically, the network data security processing platform includes an access client and a feedback client connected via a network; the personal health database is stored in the feedback client, and a data model database and a personal health data model list are stored in both the access client and the feedback client. Encryptor I, decryptor I, encryptor II, decryptor II, a data packet organization module and a delay judgment module are set in both the access client and the feedback client.
[0020] This section only provides a relatively simple configuration method. In this configuration, a local server should be configured on the access client or feedback client as needed. Setting relevant functions on the local server can reduce the client's storage and computing burdens. To expand the network range, a cloud server can also be set up, moving some client functions to the cloud server and using calls to assemble or parse data. However, the increase in nodes increases the risk of data attacks, making additional protection of the cloud server particularly important.
[0021] Specifically, it also includes an anomaly detection module and a backtracking analysis; the anomaly detection module detects whether there is abnormal traffic or abnormal communication within a set time range before and after the data packet is received, using the data packet size, source address and destination address as characteristics. If there is abnormal traffic or abnormal communication, it is determined that an abnormal event has occurred; the backtracking analysis module performs anomaly detection on the nodes through which the data packet passes along the data packet transmission link.
[0022] Specifically, it also includes an abnormal event classification neural network, which includes a data acquisition module, a feature extraction module, a classification module, a supervised training module and a testing module; the acquisition module is used to collect abnormal data packets and network status information within a set time before and after the occurrence moment; the feature extraction module is used to extract the behavioral characteristics of abnormal data packets; the classification module classifies the abnormal events of the attack data packets according to the behavioral characteristics to obtain abnormal event prediction labels; the supervised training module calculates the loss value between the abnormal event prediction label and the abnormal event true label through a loss function, and optimizes the parameters of the supervised classification module according to the loss value to obtain the most effective parameters, and then obtains a trained classification model; the testing module uses the feature extraction module to extract the behavioral characteristics of the abnormal data packets, uses the trained classification module to classify the behavioral characteristics, and predicts the label of the abnormal event.
[0023] Specifically, the behavior characteristics include data size characteristics, source address characteristics, destination address characteristics, traffic characteristics, latency characteristics, and user data characteristics.
[0024] Beneficial effects: Compared with the existing technology, the network data security processing method and platform provided by the present invention, in combination with the characteristics of personal health data, does not adopt the method of directly transmitting access requests for network transmission of data. Instead, it extracts features of the access requests according to the particularity of personal health data to obtain a two-dimensional number, which reduces the amount of data transmission and also reduces the indicative nature of the data; at the same time, preliminary security supervision based on delay is performed at both the sending and receiving ends of the data, thereby improving the security of network data; the present invention analyzes the event behavior accompanying the data packet and judges whether an abnormal event has occurred, thereby further improving the security of network data transmission; the present invention also introduces a neural network to intelligently classify network attack types, making the transmission of medical and health data more intelligent. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 Schematic diagram of the process of the present invention;
[0026] Figure 2 Schematic diagram of the structure of the platform of the present invention. DETAILED DESCRIPTION
[0027] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
[0028] like Figure 1 The figure shows a method for processing network data security, which is described in detail below in conjunction with each step.
[0029] Step 1: Build a personal health database
[0030] Personal health data is divided into Category, the The health data of an individual in all categories is represented as , , Represents the total number of individuals; establish data models for different categories of personal health data, and construct a personal health data model list to represent the mapping relationship between the data model ID and the data model storage address. Individual The personal health data of each category is represented by the data model: , For the data model The input features of .
[0031] Step 2: Access the client to assemble the access data packet
[0032] After the access client generates an access request, first, the access request is feature extracted to obtain information about 、 Target sequence ; Then, the target sequence is encrypted using the encryption algorithm I Encrypt to get encrypted data ; Next, construct the access data packet. The header of the access data packet contains control information and verification information. The user information of the access data packet contains encrypted data. , using the sending time As filling data, if the sending time in the filling data If it is incomplete, the complete sending time will be added to the reserved data. .
[0033] The access client sends the assembled access data packet to the feedback client via the Internet.
[0034] Step 3: Feedback client to parse the access data packet
[0035] After receiving the access data packet, the feedback client needs to parse the access data packet and extract the encrypted data and access time , determine the access time Is the delay within the delay threshold range: If the access time If the delay exceeds the delay threshold range, it is judged that the access data packet is attacked, and the abnormality detection work is started on the feedback client. If an abnormal event is found, it is backtracked and analyzed according to the authority; otherwise, the extracted encrypted data , use decryption algorithm I to perform decryption operation to obtain the target sequence .
[0036] Here, a reverse evaluation step can be added according to the set frequency or condition, specifically: using encryption algorithm I to encrypt the target sequence Encrypt to get encrypted data , using encrypted data Evaluate whether the user information of the access data packet carries attack data; if the attack assessment step finds that the user information of the access data packet carries attack data, initiate anomaly detection on the feedback client and initiate retrospective analysis of the access data packet starting from the feedback client based on the permissions.
[0037] Step 4: Feedback client assembles feedback data packet
[0038] After the feedback client confirms that it can perform subsequent operations, it first Query the personal health database and construct the target feature sequence ; Then, query the list of personal health data models according to the target sequence Retrieve the corresponding data model and combine it with the target feature sequence , build the target data model sequence ; Then, use encryption algorithm II to encrypt the target data model sequence Encrypt to get encrypted data ; Then, construct a feedback data packet, the header of which contains control information and verification information, and the user information of the feedback data packet contains encrypted data , use feedback time As filling data, if the feedback time in the filling data If it is incomplete, the complete feedback time will be added to the reserved data. .
[0039] The feedback client sends the assembled access data packet to the access client via the Internet.
[0040] Step 5. Access the client to parse the feedback data packet
[0041] After receiving the feedback data packet, the access client needs to parse the feedback data packet and extract the encrypted data and feedback time , judge the feedback time Is the delay within the delay threshold range: If the feedback time If the delay exceeds the delay threshold range, it is judged that the feedback data packet is attacked, and the abnormality detection work is started on the access client. If an abnormal event is found, it is backtracked and analyzed according to the authority; otherwise, the extracted encrypted data , use decryption algorithm II to perform decryption operation to obtain the target data model sequence .
[0042] Here, a reverse evaluation step can be added according to the set frequency or condition, specifically: using encryption algorithm II to encrypt the target data model sequence
[0043] Encrypt to get encrypted data , using encrypted data Evaluate whether the user information in the feedback data packet carries attack data; if the attack evaluation step finds that the user information in the feedback data packet carries attack data, initiate anomaly detection on the access client and initiate retrospective analysis of the feedback data packet starting from the scope client based on the permissions.
[0044] Step 6: Access the client to obtain feedback on the access request
[0045] After confirming that the subsequent operation can be performed, the access client first reads the target data model sequence Extract feedback target sequence , if the feedback target sequence With target sequence If the target sequence is consistent, Query the list of personal health data models and retrieve the corresponding data model sequence for the target data model Parse and get the target feature sequence , using the target feature sequence Organize feedback on access requests.
[0046] Step 7: Anomaly Detection
[0047] The anomaly detection in this case includes delay detection, reverse evaluation, anomaly detection, backtracking analysis, and abnormal event classification. Delay detection is a process that must be performed at the time of receiving the request data packet and the feedback data. Reverse evaluation is a detection performed on the access client or feedback client according to the set frequency or set conditions. Anomaly detection, backtracking analysis, and abnormal event classification can be set up on the client, local server, or cloud server as needed.
[0048] The anomaly detection detects whether there is abnormal traffic or abnormal communication within a set time range before and after the data packet is received, based on the data packet size, source address and destination address as characteristics. If there is abnormal traffic or abnormal communication, it is determined that an abnormal event has occurred.
[0049] The backtracking analysis is to perform anomaly detection on the nodes through which the data packet passes along the data packet transmission link.
[0050] The abnormal event classification uses an abnormal event classification neural network to classify abnormal events, which mainly includes two parts: a training process and a testing process. During the training process, an abnormal event attack data packet is first used to obtain an abnormal data packet, and then the behavioral features of the abnormal data packet are extracted using a feature extraction module. Then, the abnormal event of the attack data packet is classified using a classification module based on the behavioral features to obtain an abnormal event prediction label. The classification module is supervised and trained using a loss function to optimize the parameters of the classification module. During the testing process, the behavioral features of the abnormal data packet are first extracted using a feature extraction module, and then the trained classification module is used to classify the behavioral features and predict the label of the abnormal event. The behavioral features mainly include data size features, source address features, destination address features, traffic features, latency features, and user data features.
[0051] A network data security processing platform for implementing the above method mainly includes a personal health database, a data model database, a personal health data model list, an encryptor I, a decryptor I, an encryptor II, a decryptor II, a data packet organization module and a delay judgment module; the personal health database divides personal health data into Category, the The health data of an individual in all categories is represented as , , Represents the total number of individuals; the data model database is used to store data models of different categories of personal health data, Individual The personal health data of each category is represented by the data model: ; The personal health data model list is used to characterize the mapping relationship between the data model ID and the data model storage address; the encryptor I and decoder I are used to encrypt and decrypt the target sequence; the encryptor II and decryptor II are used to encrypt and decrypt the target data model sequence; the data packet organization module is used to construct and parse the access data packet and the feedback data packet; the delay judgment module is used to judge whether the delay of the access time and the feedback time exceeds the delay threshold range.
[0052] The network data security processing platform in this case, one configuration method is to include an access client and a feedback client connected via a network; the personal health database is stored in the feedback client, and a data model database and a personal health data model list are stored in both the access client and the feedback client; an encryptor I, a decryptor I, an encryptor II, a decryptor II, a data packet organization module and a delay judgment module are set in both the access client and the feedback client.
[0053] Another configuration method of the network data security processing platform in this case is to include an access client and a feedback client, the access client is registered on the local server of the access end, and the feedback client is registered on the local server of the feedback end, and the local server of the access end and the local server of the feedback end are connected through a network; the personal health database is stored on the local server of the feedback end, and the local server of the access end and the local server of the feedback end both store a data model database and a personal health data model list, and an encryptor I, a decryptor I, an encryptor II, a decryptor II, a data packet organization module and a delay judgment module are set on the access client and the feedback client.
[0054] Another configuration method of the network data security processing platform in this case is to include an access client and a feedback client, the feedback client is registered on the local server of the feedback end, and the access client and the local server of the feedback end are connected to the network through the cloud server; the personal health database is stored on the local server of the feedback end, and the local server of the feedback end and the cloud server both store the data model database and the personal health data model list, and the access client and the feedback client are both set with an encryptor I, a decryptor I, an encryptor II, a decryptor II, a data packet organization module and a delay judgment module; the access client parses the target data model sequence by calling the data model database and the personal health data model list in the cloud server.
[0055] The three configuration methods listed above can be selected based on the amount of data and the size of the network range. Other solutions can also be adopted according to actual needs.
[0056] In order to detect abnormal events, an anomaly detection module, backtracking analysis and abnormal event classification neural network are configured in the network data security processing platform.
[0057] The anomaly detection module detects whether there is abnormal traffic or abnormal communication within a set time range before and after the data packet is received, based on the data packet size, source address and destination address as characteristics. If there is abnormal traffic or abnormal communication, it is determined that an abnormal event has occurred.
[0058] The backtracking analysis module performs anomaly detection on the nodes through which the data packet passes along the data packet transmission link.
[0059] The abnormal event classification neural network includes a data acquisition module, a feature extraction module, a classification module, a supervised training module and a testing module; the acquisition module is used to collect abnormal data packets and network status information within a set time before and after the occurrence moment; the feature extraction module is used to extract the behavioral characteristics of the abnormal data packets; the classification module classifies the abnormal events of the attack data packets according to the behavioral characteristics to obtain abnormal event prediction labels; the supervised training module calculates the loss value between the abnormal event prediction label and the abnormal event true label through a loss function, and optimizes the parameters of the supervised classification module according to the loss value to obtain the most effective parameters, and then obtains a trained classification model; the testing module uses the feature extraction module to extract the behavioral characteristics of the abnormal data packets, uses the trained classification module to classify the behavioral characteristics, and predicts the labels of the abnormal events; the behavioral characteristics include data size characteristics, source address characteristics, destination address characteristics, traffic characteristics, delay characteristics, and user data characteristics.
[0060] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any form, and any technical solutions obtained by equivalent replacement or equivalent transformation fall within the scope of protection of the present invention.
Claims
1. A network data security processing method, characterized in that: The steps include: Step 1: Divide personal health data into categories, The health data of an individual in all categories is represented as , , Represents the total number of individuals; establish data models for different categories of personal health data, and construct a list of personal health data models. Personal The personal health data of categories are represented by the data model as follows: , For data model Input features of Step 2: Extract features from access requests and obtain information about , Target sequence , using encryption algorithm I to encrypt the target sequence Encrypt to get encrypted data ; Build an access data packet. The header of the access data packet contains control information and verification information. The user information of the access data packet contains encrypted data. , use the sending time As padding data, if the sending time in the padding data If it is incomplete, the complete sending time is added to the reserved data. ; Step 3. Parse the access data packet and extract the encrypted data and access time If the access time If the delay is within the delay threshold, then: first encrypt the data Decrypt to obtain the target sequence ; Then according to the target sequence Query the personal health database and construct the target feature sequence ; Then query the list of personal health data models, according to the target sequence Retrieve the corresponding data model and combine it with the target feature sequence , build the target data model sequence ; Use encryption algorithm II to encrypt the target data model sequence Encrypt to get encrypted data ; Construct a feedback data packet. The header of the feedback data packet contains control information and verification information. The user information of the feedback data packet contains encrypted data. , use feedback time As filling data, if the feedback time in the filling data If it is incomplete, the complete feedback time will be added to the reserved data. ; Step 4. Parse the feedback data packet and extract the encrypted data and feedback time , if the feedback time If the delay is within the delay threshold, then: first encrypt the data Decrypt to obtain the target data model sequence ; Then from the target data model sequence Extract feedback target sequence , if the feedback target sequence With target sequence If they are consistent, then according to the target sequence Query the list of personal health data models and retrieve the corresponding data model for the target data model sequence Parse and get the target feature sequence , using the target feature sequence Respond to access requests.
2. The network data security processing method according to claim 1, characterized in that: Step 3 includes an attack evaluation step, which is performed according to a set frequency. Specifically, the target sequence is encrypted using the same encryption algorithm I as Step 2. Encrypt to get encrypted data , using encrypted data Evaluate whether the user information of the access data packet carries attack data; Step 4 includes an attack evaluation step, which is performed according to the set frequency; specifically, the target data model sequence is encrypted using the same encryption algorithm II as Step 3. Encrypt to get encrypted data , using encrypted data Evaluate whether the user information in the feedback data packet carries attack data.
3. The network data security processing method according to claim 2, characterized in that: If the access time If the delay exceeds the delay threshold range, or if the attack assessment step finds that the user information of the access data packet carries attack data, anomaly detection is started at the destination node, and the backtracking analysis of the access data packet from the destination node is started according to the authority; if the feedback time If the delay exceeds the delay threshold range, or if the attack assessment step finds that the user information of the feedback data packet carries attack data, anomaly detection is started at the source node, and backtracking analysis of the feedback data packet from the source node is started according to the authority; The anomaly detection detects whether there is abnormal traffic or abnormal communication within a set time range before and after the data packet is received, based on the data packet size, source address and destination address as characteristics. If there is abnormal traffic or abnormal communication, it is determined that an abnormal event has occurred; the backtracking analysis performs anomaly detection on the nodes through which the data packet passes along the data packet transmission link.
4. The network data security processing method according to claim 3 is characterized in that: Abnormal events are classified using an abnormal event classification neural network, which includes two parts: training and testing. During the training process, the abnormal event attack data packet is first used to obtain the abnormal data packet, and then the feature extraction module is used to extract the behavioral features of the abnormal data packet. Then, the classification module is used to classify the abnormal events of the attack data packet according to the behavioral features to obtain the abnormal event prediction label. The loss function is used to supervise the training of the classification module to optimize the parameters of the classification module. During the testing process, the feature extraction module is first used to extract the behavioral features of the abnormal data packet, and then the trained classification module is used to classify the behavioral features to predict the label of the abnormal event.
5. The network data security processing method according to claim 4, characterized in that: The behavior characteristics include data size characteristics, source address characteristics, destination address characteristics, traffic characteristics, delay characteristics, and user data characteristics.
6. A network data security processing platform, characterized by: It includes a personal health database, a data model database, a personal health data model list, an encryptor I, a decryptor I, an encryptor II, a decryptor II, a data packet organization module and a delay judgment module; The personal health database divides personal health data into categories, The health data of an individual in all categories is represented as , , represents the total number of individuals; The data model database is used to store data models of different categories of personal health data. Personal The personal health data of categories are represented by the data model as follows: ; The personal health data model list is used to represent the mapping relationship between the data model ID and the data model storage address; The encryptor 1 and the decoder 1 are used to encrypt and decrypt the target sequence; The encryptor II and the decryptor II are used to encrypt and decrypt the target data model sequence; The data packet organizing module is used to construct and parse access data packets and feedback data packets; The delay judgment module is used to judge whether the delay of access time and feedback time exceeds the delay threshold range.
7. The network data security processing platform according to claim 6, characterized in that: It includes an access client and a feedback client connected through a network; the personal health database is stored in the feedback client, and a data model database and a personal health data model list are stored in the access client and the feedback client. Encryptor I, decryptor I, encryptor II, decryptor II, data packet organization module and delay judgment module are set in the access client and the feedback client.
8. The network data security processing platform according to claim 6, characterized in that: It also includes an anomaly detection module and a backtracking analysis; the anomaly detection module detects whether there is abnormal traffic or abnormal communication within a set time range before and after the data packet is received, based on the data packet size, source address and destination address as features, and if there is abnormal traffic or abnormal communication, it is determined that an abnormal event has occurred; The backtracking analysis module performs anomaly detection on the nodes through which the data packet passes along the data packet transmission link.
9. The network data security processing platform according to claim 8, characterized in that: It also includes an abnormal event classification neural network, which includes a data acquisition module, a feature extraction module, a classification module, a supervised training module and a test module; the acquisition module is used to collect abnormal data packets and network status information within a set time before and after the occurrence moment; the feature extraction module is used to extract the behavioral characteristics of the abnormal data packets; the classification module classifies the abnormal events of the attack data packets according to the behavioral characteristics to obtain abnormal event prediction labels; the supervised training module calculates the loss value between the abnormal event prediction label and the abnormal event real label through a loss function, and optimizes the parameters of the supervised classification module according to the loss value to obtain the most effective parameters, and then obtains a trained classification model; the test module uses the feature extraction module to extract the behavioral characteristics of the abnormal data packets, uses the trained classification module to classify the behavioral characteristics, and predicts the labels of the abnormal events.
10. The network data security processing platform according to claim 9, characterized in that: The behavior characteristics include data size characteristics, source address characteristics, destination address characteristics, traffic characteristics, delay characteristics, and user data characteristics.
