A sensitive data privacy protection computing method in a cloud environment

Abstract

The application discloses a sensitive data privacy protection computing method in a cloud environment, comprising: collecting traffic data of a power internet of things, and structurally processing the collected traffic data; under a privacy computing technology framework, performing multi-dimensional privacy feature extraction on the structurally processed traffic data, and identifying sensitive information including user addresses, names and contact numbers; wherein, an address adaptive identification algorithm is adopted to identify the user addresses in the traffic data; the identified sensitive information is marked and graded, and monitoring protection is performed according to the grading results. The application combines the traffic data characteristics of the power internet of things, proposes a brand-new address tree adaptive identification technology, can quickly identify sensitive data in a cloud computing environment, deeply integrates privacy computing technology, constructs a multi-level protection system in the whole life cycle of data, and ensures the safety and privacy of sensitive data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power internet of things, and particularly relates to a sensitive data privacy protection computing method in a cloud environment. BACKGROUND

[0002] With the continuous improvement of the digitalization, cloud computing and intelligence of power internet of things, the wide application of intelligent terminals, including smart meters, temperature detection devices, humidity detection devices and power distribution devices, more and more terminal collected data are transmitted in a cloud computing environment, and these data contain sensitive data, such as user address, user name, power consumption and power equipment. That is, with the popularization and development of power internet of things, great convenience is brought to people's life, and the data privacy problem in cloud computing has become the focus of attention.

[0003] These sensitive data need to be properly protected and managed to prevent data leakage and damage, and necessary security measures and technical means need to be taken in the generation, transmission, use, storage and destruction of these sensitive data to ensure the security and privacy of sensitive data. SUMMARY

[0004] The purpose of the present application is to provide a sensitive data privacy protection computing method in a cloud environment, combining the traffic data characteristics of power internet of things, a new address tree adaptive recognition technology is proposed, which can quickly recognize sensitive data in a cloud computing environment, and deeply integrate privacy computing technology, build a multi-level protection system in the whole life cycle of data, and ensure the security and privacy of sensitive data.

[0005] To achieve the above technical purpose, the technical scheme adopted by the present application is:

[0006] The present application discloses a sensitive data privacy protection computing method in a cloud environment, which comprises the following steps:

[0007] S1, collecting traffic data of power internet of things, and structuring the collected traffic data;

[0008] S2, under the framework of privacy computing technology, multi-dimensional privacy feature extraction is performed on the structured traffic data to identify sensitive information including user address, name and contact number; wherein an address adaptive recognition algorithm is used to recognize the user address in the traffic data, and the recognition process comprises the following sub-steps:

[0009] S21, target data word segmentation; specifically, the text information obtained in the traffic data is segmented to obtain address segmentation segments of multiple levels of administrative district levels;

[0010] S22, initialize a standard address library; specifically, import the standard address library, which includes detailed address information of provinces, cities, districts and streets nationwide and corresponding address Hash data;

[0011] S23, address tree global Hash matching; specifically, obtain Hash data corresponding to a target address to be identified; compare the Hash data of the target address with address Hash data of the standard address library one by one, if equal data is found, it is determined that address sensitive data exists in the traffic data and the address information is output, and step S3 is entered, otherwise, step S24 is entered;

[0012] S24, address tree node depth and breadth weight matching; segment the target address to be identified by Hash, based on the standardized address library, the Hash data of the multiple address segments of the target address to be identified is searched in breadth and depth according to the order of administrative region level from large to small, if each address segment can match equal data, it is determined that address sensitive data exists in the traffic data and the address information is output, and step S3 is entered, otherwise, step S25 is entered;

[0013] S25, for different administrative region levels, introduce corresponding matching weights, combine the Hash data of the multiple address segments of the target address to be identified, and calculate the hierarchical sensitive coefficient of the address tree node depth and breadth, if the hierarchical sensitive coefficient of the address tree node depth and breadth exceeds a preset threshold, it is determined that address sensitive data exists in the traffic data of the power Internet of Things and the address information is output, and step S3 is entered, otherwise, it is determined that address sensitive data does not exist in the traffic data of the power Internet of Things, and the process ends;

[0014] S3, mark and classify the identified sensitive information, and monitor and protect according to the classification result.

[0015] Further, in step S1, a bypass mode is adopted to copy the current network traffic on the cloud in real time from the power Internet of Things, connect the north-south network traffic of the unified entrance and exit of the external network, and the data communication traffic between the network inside and outside, that is, the network traffic across the network boundary vertically.

[0016] Further, in step S1, the process of structuring the collected traffic data includes the following steps:

[0017] The collected traffic data is cleaned to remove invalid, expired, repeated and incorrect data;

[0018] The cleaned traffic data is uniformly converted in format;

[0019] Public parameters including storage time and collection time are added to the format-converted traffic data.

[0020] Further, the name recognition process in step S2 includes the following steps:

[0021] According to the surname, the matching of the hundred surnames is performed, and the single-character and double-character Chinese names are combined for screening, so as to preliminarily identify all possible Chinese names in the sentence;

[0022] The preliminarily identified Chinese names are matched with the words in the dictionary, and all possible combinations are outputted;

[0023] The contact telephone number recognition process includes the following steps:

[0024] The input text is converted into a unified format, and the punctuation and spaces contained therein are removed;

[0025] Regular expressions are used to find strings that conform to the mobile phone number format;

[0026] A secure multi-party computation protocol in privacy computing is introduced to perform distributed collaborative verification on the mobile phone number matching rule.

[0027] Further, in step S2, the text information obtained in the traffic data of the power internet of things is segmented, and address segmentation segments of multiple levels of administrative regions are obtained, and if part of the address segmentation segments are empty, the empty values are filled, to obtain:

[0028] S={S1, S2, S3, S4, S5};

[0029] Wherein S is the target address information to be identified, S1 represents the province data to be identified, S2 represents the city data to be identified, S3 represents the district data to be identified, S4 represents the street data to be identified, and S5 represents the detailed address data to be identified;

[0030] The standard address library D is imported:

[0031]

[0032] Wherein, m represents the total number of all complete addresses in the address library; represents the name of the i-th administrative region in the j-th address, D1 represents the province data of the standard address, D2 represents the city data of the standard address, D3 represents the district data of the standard address, D4 represents the street data of the standard address, and D5 represents the detailed address data of the standard address;

[0033] The address data in the standard address library D is hashed to obtain the corresponding address Hash data:

[0034]

[0035] In the formula, Hd Hash information in address information of the standard address library, HD1 represents Hash data of a province, HD2 represents Hash data of a city, HD3 represents Hash data of a district, HD4 represents Hash data of a street, and HD5 represents Hash data of a detailed address.

[0036] Further, in step S23, the process of the address tree global Hash matching includes the following steps:

[0037] S231, Hashing the target address to be identified in the power internet of things traffic to obtain the following three types of data: Hash data S of the province, city, district, street and detailed address h1 = Hash*(S1+S2+S3+S4+S5), Hash data S of the district, street and detailed address h2 = Hash*(S3+S4+S5), Hash data S of the street and detailed address h3 = Hash*(S4+S5);

[0038] S232, comparing S h1 and D h1 , if they are equal, it is found that there is address sensitive data in the traffic data and the address information is a province S1, a city S2, a district S3, a street S4 and a detailed address S5; otherwise, go to step S233;

[0039] S233, comparing S h2 and D h2 , if they are equal, it is found that there is address sensitive data in the traffic data and the address information is a district S3, a street S4 and a detailed address S5, otherwise, go to step S234;

[0040] S234, comparing S h3 and D h3 , if they are equal, it is found that there is address sensitive data in the traffic data and the address information is a street S4 and a detailed address S5;

[0041] In the formula, D h1 is Hash data of a piece of province, city, district, street and detailed address information obtained by traffic in the power internet of things after splicing; D h2 is Hash data of a piece of district, street and detailed address information obtained by traffic in the power internet of things after splicing; and D h3 is Hash data of a piece of street and detailed address information obtained by traffic in the power internet of things after splicing.

[0042] Step S24 further includes:

[0043] S241, Hash the target address to be identified by segment, using the Hash value H of the province S1 s1 A breadth search is performed until the same standard address province data H is searched s1 d1 If the same data cannot be searched, step S25 is entered;

[0044] S242, using the Hash value H of the city S2 s2 A depth search is performed until the same standard address city data H is searched s2 d2 If the same data cannot be searched, step S25 is entered;

[0045] S243, using the Hash value H of the district S3 s3 A depth search is performed until the same standard address district data H is searched s3 d3 If the same data cannot be searched, step S25 is entered;

[0046] S244, using the Hash value H of the street S4 s4 A depth search is performed until the same standard address street data H is searched s4 d4 If the same data cannot be searched, step S25 is entered;

[0047] S245, using the Hash value H of the detailed address S5 s5 A depth search is performed until the same standard address detailed address data H is searched s5 d5 If the same data cannot be searched, step S25 is entered;

[0048] S246, combining H d1 , H d2 , H d3 , H d4 and H d5 , to obtain the identification result of the target address to be identified.

[0049] Step S25 further comprises:

[0050] The address sensitive information to be identified found in the traffic of the power internet of things is segmented to obtain multiple administrative district level address data including province level, city level, district level, street level and detailed address information, and the address data at each level is Hash calculated to obtain a unique Key value H S ; ​​​​​

[0051] The full address information of multiple levels including full provincial address information, full municipal address information, full district information, full street level information and full detail address information is constructed by known standard address library, the full address information is Hash calculated to obtain corresponding full Key value, and a nationwide address Hash address tree is formed, each node of the tree is H d ;

[0052] The address sensitive information to be identified found in the traffic of the power internet of things and the nationwide address Hash address tree are compared in weighted Hash value, and a hierarchical sensitive coefficient is obtained through hierarchical dependence and weighted average

[0053]

[0054] In the formula, is a hierarchical sensitive coefficient, used to represent a sensitive degree comprehensive score of address data; is a dynamic weight of the i-th administrative region, which is dynamically adjusted by a preset weight and a security risk coefficient corresponding to the corresponding administrative region; s i represents a Hash matching similarity of the i-th level administrative region address; represents a progressive hierarchical dependence, wherein k is a superior level number, s κ is a matching degree of the k-th level address; λ represents a hierarchical attenuation compensation factor, which is negatively related to data flow complexity; σ() represents a dynamic threshold function.

[0055] Further, the preset threshold adopts an adaptive sensitive threshold, the adaptive sensitive threshold is a boundary value of a sensitive degree of the address sensitive information to be identified found in the traffic of the power internet of things, and the adaptive sensitive threshold is used for adaptively adjusting the sensitive data judgment level through joint modeling of security risk, flow complexity and user strategy.

[0056] Further, a calculation formula of the adaptive sensitive threshold AST is as follows:

[0057]

[0058] Among them, is a security risk coefficient, related to attack frequency and vulnerability exposure degree; is a data flow complexity, related to unstructured text proportion and address segmentation hierarchical depth; is a user strategy weight, related to user-defined sensitive level; β, γ and δ are weight factors, and β+γ+δ=1;

[0059] If Less than the adaptive sensitive threshold, the address to be identified found in the power internet of things traffic is excluded as sensitive information; if Greater than or equal to the adaptive sensitive threshold, the address to be identified found in the power internet of things traffic is determined as sensitive information.

[0060] Compared with the prior art, the cloud environment sensitive data privacy protection computing method has the following beneficial effects:

[0061] First, the cloud environment sensitive data privacy protection computing method accurately and quickly identifies the address information in the sensitive data in the power internet of things traffic by using the address tree adaptive identification technology in the data life cycle of transmission, use, storage and destruction, and the matching of names and the identification of mobile phone numbers.

[0062] Second, in the sensitive address identification scene of the power internet of things, the cloud environment sensitive data privacy protection computing method improves the determination accuracy through the "dynamic adjustment and layer-by-layer progression" of the "hierarchical sensitive coefficient" algorithm, so as to realize the adaptive protection of "the higher the risk, the stricter the protection, and the more chaotic the data, the more detailed the protection" in the complex network environment.

[0063] Third, in the transmission process, the cloud environment sensitive data privacy protection computing method uses strong encryption technology and a secure transmission protocol to protect the confidentiality, integrity and availability of sensitive data, and introduces homomorphic encryption technology to realize the secret state transmission and calculation of sensitive data; in the use process, the cloud environment sensitive data privacy protection computing method uses a multi-factor authentication mechanism to ensure that only authorized users can access sensitive data, and constructs a privacy computing sandbox based on a trusted execution environment (TEE) to support the secure collaborative analysis of multi-party data; in the storage process, the cloud environment sensitive data privacy protection computing method processes data by using an asymmetric RSA encryption algorithm, and adds noise disturbance to aggregated data by using differential privacy technology to ensure that even if the data is illegally accessed during storage, the real content cannot be obtained, and a strict access control mechanism is established to ensure that only authorized personnel can access and operate sensitive data. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 The cloud environment sensitive data privacy protection computing method flow chart of the present application;

[0065] Figure 2 The flow chart of the traffic data processing;

[0066] Figure 3 The address adaptive identification algorithm flow chart;

[0067] Figure 4 The address tree graph of the full amount of provincial, municipal, district and street detailed address levels. DETAILED DESCRIPTION

[0068] Embodiments of the present application are described in further detail below with reference to the accompanying drawings.

[0069] The application discloses a sensitive data privacy protection computing method in a cloud environment, and the method comprises the following steps:

[0070] S1, collecting traffic data of the power Internet of Things, and performing structured processing on the collected traffic data;

[0071] S2, under the technical framework of privacy computing, performing multi-dimensional privacy feature extraction on the structured traffic data, and identifying sensitive information including user addresses, names and contact numbers; wherein an address adaptive identification algorithm is used to identify the user addresses in the traffic data, and the identification process comprises the following sub-steps:

[0072] S21, target data word segmentation; specifically, the text information obtained in the traffic data is subjected to word segmentation to obtain address segmented segments of multiple levels of administrative regions;

[0073] S22, initializing a standard address library; specifically, a standard address library is imported, and the standard address library comprises detailed address information of provinces, cities and districts nationwide and corresponding address Hash data;

[0074] S23, address tree global Hash matching; specifically, Hash data corresponding to a target address to be identified is obtained; the Hash data of the target address is compared with the address Hash data of the standard address library one by one, if equal data is found, it is determined that address sensitive data exists in the traffic data and the address information is output, and the process proceeds to step S3, otherwise, the process proceeds to step S24;

[0075] S24, address tree node depth and breadth weight matching; the target address segmented segments to be identified are subjected to Hash, based on the standardized address library, the Hash data of the multiple address segmented segments of the target address to be identified are searched in breadth and depth according to the order of administrative region levels from large to small, if each address segmented segment can match equal data, it is determined that address sensitive data exists in the traffic data and the address information is output, and the process proceeds to step S3, otherwise, the process proceeds to step S25;

[0076] S25, for different administrative region levels, a corresponding matching weight is introduced, and a hierarchical sensitive coefficient of the address tree node depth and breadth is calculated based on the Hash data of the multiple address segmented segments of the target address to be identified, if the hierarchical sensitive coefficient of the address tree node depth and breadth exceeds a preset threshold, it is determined that address sensitive data exists in the traffic data of the power Internet of Things and the address information is output, and the process proceeds to step S3, otherwise, it is determined that no address sensitive data exists in the traffic data of the power Internet of Things, and the process ends;

[0077] S3, marking and grading the identified sensitive information, and monitoring and protecting according to the grading results.

[0078] As Figure 1 shown, the sensitive data privacy protection computing method in the cloud environment of the application includes sensitive data identification, sensitive data monitoring and alarm, sensitive data encryption transmission, sensitive data access authorization, and deep integration of privacy computing technology to build a multi-level protection system in the whole life cycle of data. The key of data privacy protection of power internet of things is to take network access power internet terminal network traffic as the starting point, to identify sensitive data from network traffic collection, to monitor sensitive data, and to alarm after discovering sensitive sensitive data; and the sensitive data of power internet management center and each application is transmitted through strong encryption, and if access to sensitive data needs authorization operation, so as to ensure the confidentiality and security of sensitive data in the process of generation, transmission, use and storage, to ensure the safe use of sensitive data of power internet of things, and to promote the sustainable development of power internet of things.

[0079] 1Sensitive data identification:

[0080] The existing external application system in the cloud computing environment of power internet of things calls the business system through API interface, such as electricity bill query, payment record query, etc.; there is also information sharing between internal systems, such as cross-regional electricity information statistics, real-time analysis of power terminal. These information contains sensitive data of power internet of things, which needs to be protected, so as to ensure the normal operation of power system, prevent data from being tampered or damaged, and maintain the safety and stability of power system. The internet access entrance on the cloud needs to be set up as a traffic collection point, and the network traffic collection equipment is set up by mirroring the switch port, which can real-time traffic replication to obtain the network traffic information of the cloud environment, such as Figure 2 .

[0081] 1.1Cloud traffic collection:

[0082] In power internet of things, our network traffic collection system works in bypass mode by default. The core of this bypass mode is that it can real-time copy the current network traffic without intercepting or interrupting the traffic. This non-intrusive way ensures smooth business process and provides reliable support for traffic data collection and analysis. In this way, we can effectively monitor and manage the running state of power internet of things without interfering with normal business operations.

[0083] The traffic replication collection strategy is adopted, and then the asynchronous transmission mechanism is used to efficiently transmit the traffic information to the back-end cluster deployment data structured analysis device to support subsequent data structured analysis and sensitive data identification.

[0084] 1.2 Structured Analysis of Collected Data:

[0085] For data collection of network traffic, we need to choose the appropriate collector based on the data type. Network traffic on the cloud involves diverse data formats, including HTML, XML, MIME, JSON, ASCII, and mixed types, and the collector needs to match them to ensure accurate parsing of the data. In addition, log data from network devices are also supported.

[0086] After collection, the traffic data will go through a series of processing steps. First is data cleaning, which aims to eliminate invalid, outdated, duplicate, and erroneous data to ensure atomic consistency and accuracy of the data. Then comes data conversion, which addresses the problem of different formats of standard parameters in traffic data. We will standardize these parameters and convert them into a unified and more recognizable format.

[0087] Finally, to achieve better traceability and multidimensional query of data, we enrich the data by adding necessary public parameters such as storage time and collection time. These steps together ensure efficient collection, processing, and application of network traffic data in the power Internet of Things.

[0088] 1.3 Sensitive Data Identification:

[0089] Under the framework of privacy computing technology, we extract multi-dimensional privacy features from the collected and analyzed data to identify sensitive information such as user address, name, and contact number. The specific process is as follows:

[0090] 1.3.1 Name Identification: A specific algorithm is used to process the sampled data, which includes two key steps: preprocessing and disambiguation.

[0091] The preprocessing stage aims to identify all possible Chinese names in the sentence. The process first matches the surname according to the common surnames, and then filters the common single and double Chinese names. Due to the complexity of Chinese, there may still be ambiguity problems after preprocessing.

[0092] To solve this problem, we perform the disambiguation step: match the preprocessed Chinese names with the words in the dictionary and output all possible combinations. Finally, we select the best combination based on the gap size between words and the number of divided words to accurately identify the name. Combined with the Private Information Retrieval (PIR) technology in privacy computing, we achieve "query privacy protection" in the name matching process, ensuring that external systems cannot infer user identity information through query operations.

[0093] 1.3.2 Contact phone number identification: The process of identifying mobile phone numbers usually relies on specific rules and algorithms, first data preprocessing, converting the input text into a format suitable for analysis, such as removing punctuation, spaces, etc. Regular expression matching: Use regular expressions to find strings that match the format of a mobile phone number, use regular expressions similar to 1[345789]\d{9} to match mobile phone numbers.

[0094] Introduce secure multi-party computation (MPC) protocols in privacy computing, distributed collaborative verification of mobile phone matching rules, to avoid exposing raw numbers in plaintext in computing nodes.

[0095] 1.3.3 Address identification: The identification of sensitive data in the traffic of the power Internet of Things is a complex task, the target address is not a standard address style, the format is various, the data fusion is chaotic, and a unique adaptive algorithm is needed to quickly and accurately identify.

[0096] The address adaptive identification algorithm adopted by the application is divided into four parts: target data word segmentation, initialization of standard address library, address tree global Hash matching, hierarchical dependence and weighted average of address tree node depth and breadth, such as Figure 3 .

[0097] Based on the address tree adaptive identification technology, the following four parts are used to realize the accurate identification of address sensitive data, and the privacy computing technology is used to realize cross-domain collaborative modeling:

[0098] 1.3.3.1 Target data word segmentation

[0099] Before executing the address tree adaptive algorithm, the text information obtained in the traffic of the power Internet of Things needs to be segmented, and the data filtering is performed through the unique properties of the address in the segmentation process, that is, the address segmentation segment in a string of text: a province, a city, a district, a street, a detailed address, if some address segmentation segments are empty, use null to fill.

[0100] S={S1,S2,S3,S4,S5}

[0101] Where S is the target address information to be identified, S1 represents a province data, S2 represents a city data, S3 represents a district (county) data, S4 represents a street (town) data, and S5 represents a detailed address.

[0102] 1.3.3.2 Initialization of standard address library

[0103] Next, import a standard address library information, which includes detailed address information of provinces, cities, districts and streets across the country, in order to facilitate fast global query and generate address Hash data.

[0104]

[0105] Wherein, m is the total number of all complete addresses in the address library (for example, all possible "province-city-district-street-detailed address" combinations in the country), The name of the i-level administrative area in the jth address (for example, i=1 is the province, i=5 is the detailed address); D is the address information of the standard address library, D1 represents the province data of the standard address, D2 represents the city data of the standard address, D3 represents the district data of the standard address, D4 represents the street data of the standard address, and D5 represents the detailed address data of the standard address. s H is the Hash information in the target address information that needs to be identified in the traffic, d HD is the Hash information in the address information of the standard address library, HD1 represents the Hash data of a province, HD2 represents the Hash data of a city, HD3 represents the Hash data of a district, HD4 represents the Hash data of a street, and HD5 represents the Hash data of a detailed address.

[0106] Through the federal learning technology to dynamically update the features of the standard address library, each participant (such as different regional power nodes) shares the address feature gradient under the privacy computing framework, realizes the collaborative optimization of the global address model, and ensures that the original address data does not leave the local region.

[0107] 1.3.3.3 Address tree global Hash matching

[0108] S is obtained by Hashing the target address, S h1 : Hash of province-city-district-street-detailed address, S h2 : Hash of district-street-detailed address, S h3 : Hash of street-detailed address.

[0109] S h1 = Hash*(S1+S2+S3+S4+S5)

[0110] S h2 = Hash*(S3+S4+S5)

[0111] S h3 = Hash*(S4+S5)

[0112] S h1 and D h1 are compared, and if they are equal, it is found that there is address sensitive data in this piece of traffic data and the address information is a province (S1) a city (S2) a district (S3) a street (S4) a detailed address (S5);

[0113] S h2 and D h2If equal, it is found that there is address sensitive data in this traffic data, and the address information is a certain district (S4) and a certain detailed address (S5);

[0114] S h3 and D h3 If equal, it is found that there is address sensitive data in this traffic data, and the address information is a certain province (a certain street (S4) and a certain detailed address (S5));

[0115] If the address information of the sensitive data of the power internet of things is not identified in the Hash data of the address library through the global Hash of the address tree, the next step is to perform address tree node depth and breadth weight matching. In the cross-cloud node comparison S h and D h , the blind Hash technology in privacy calculation is adopted to ensure that the original address data of the opposite party cannot be reversely deduced between nodes. h1 is the Hash data of a certain province-city-district-street-detailed address information obtained by traffic in the power internet of things after splicing; h2 is the Hash data of a certain district-street-detailed address information obtained by traffic in the power internet of things after splicing; h3 is the Hash data of a certain street-detailed address information obtained by traffic in the power internet of things after splicing.

[0116] 1.3.3.4 Address tree node depth and breadth weight matching

[0117] First, the full address tree is constructed. The known full address library is divided into two steps to construct the tree. First, the province, city, district, street, and detailed address information in the full address library are subjected to Hash operation to obtain a unique Key value. Then, each level of address is drawn and constructed from the root to the leaf node of the tree. Finally, the full address library information is constructed into a multi-level and multi-node address tree. Each node of the tree is H d .

[0118] The address sensitive information to be identified found in the traffic of the power internet of things is segmented to obtain the province, city, district, street, and detailed address information. Then, the Hash calculation is performed on the data of each level to obtain a unique Key value, that is, H S .

[0119] The target address to be identified is segmented and subjected to Hash. First, the Hash value (H s1 ) of a certain province (S1) is used for breadth search. If H s1 is the same as H d1 , depth search is performed. The Hash value (H s2) to find H s2 With H d2 , the same, in turn, until the detailed address of the depth search, this kind of scene is ideal, the target address and address library address matching coefficient is 100%.

[0120] In turn, the breadth of address tree nodes, get all the provincial node weight data, and then depth traversal of city nodes, district nodes, street level nodes, detailed address nodes, in turn, to obtain all the city node weight data, district node weight data, street level node weight data, detailed address node weight data. Through the "hierarchical dependence and weighted average" algorithm to identify the address sensitive degree found in the power of the Internet of Things in the flow of the to-be-identified address, that is, "hierarchical sensitivity coefficient ".

[0121] The address sensitive information (H S ) found in the flow of the power of the Internet of Things and the national address tree (H d ) are compared by Hash weighting, and the hierarchical sensitivity coefficient is obtained by hierarchical dependence and weighted average; wherein, the calculation formula of the hierarchical sensitivity coefficient of the address tree node depth breadth weight is:

[0122]

[0123] The formula means as follows:

[0124] (Hierarchical sensitivity coefficient): represents the sensitivity degree comprehensive score of address data, the range is [0, 1], if Adaptive sensitivity threshold (Adaptive Sensitivity Threshold), it is determined that the address found in the flow of the power of the Internet of Things is sensitive information; if Adaptive sensitivity threshold (Adaptive Sensitivity Threshold), it is determined that the address found in the flow of the power of the Internet of Things is non-sensitive information.

[0125] (Dynamic weight): the weight distribution of the i-th administrative region is dynamically adjusted by the preset weight and the security risk coefficient. When the security risk is high, the weight of the high level (such as the provincial level) is improved, and the weight of the low level (such as the street) is reduced.

[0126] (Hierarchical similarity): the Hash matching similarity of the i-th administrative region address (such as the provincial matching degree s1, the city s2, etc.), which is calculated by Hash value comparison (for example: complete matching is 1, partial matching is 0.8). ​

[0127] (Progressive Hierarchical Dependency): k represents the upper level number (k = 1 for province level, k = 2 for city level, and so on), s k represents the matching degree of the kth level address (s k ∈ [0, 1]), represents taking the square root of the upper matching degree, reducing excessive inhibition.

[0128] λ (Hierarchical Attenuation Compensation Factor): negatively related to data flow complexity (λ = 1 - complexity / 2). The more chaotic the flow (high complexity), the slower the attenuation, the smaller λ, and the deeper the address contribution.

[0129] σ (Dynamic Threshold Function): compresses the score to [0, 1] through the Sigmoid function, and the parameters kk (steepness) and θθ (threshold offset) of the dynamic threshold function are dynamically controlled by user policy and security risk.

[0130] For example, when the power monitoring system detects a suspected address segment, with the following initial data: security risk coefficient = 0.8 (detecting abnormal login behavior); data flow complexity = 0.7 (address fragmentation and containing unstructured text); user policy weight = 0.7 (high sensitivity mode); administrative district level: 3 levels (province, city, street); preset weight: W = [0.6, 0.3, 0.1], then the corresponding hierarchical sensitivity coefficient is:

[0131]

[0132] In the sensitive address identification scenario of the power Internet of Things, the "hierarchical sensitivity coefficient" algorithm improves the judgment accuracy through "dynamic adjustment and progressive" methods: when the system detects a network attack (such as DDoS), the "hierarchical sensitivity coefficient" algorithm automatically increases the matching weight of the provincial address (similar to "focusing on high-risk entrances"), while reducing the strictness of the deep street address (allowing partial fuzzy matching); if the flow data is complex (such as severe address fragmentation), the "hierarchical attenuation compensation" mechanism is used to strengthen the matching contribution of deep addresses (to avoid missing hidden sensitive segments). The system also dynamically adjusts the strictness of the judgment threshold according to the user's security level (low / medium / high) - for example, when the user selects "high sensitivity" mode, even if the address segment has only 60% similarity (such as "Beijing Haidian District ** Street" partial fuzzy), it can also trigger the sensitive marking, thereby achieving "the higher the risk, the stricter the data, and the more chaotic the environment" adaptive protection.

[0133] The application adopts an AST adaptive sensitivity threshold as a boundary value of the sensitivity degree of sensitive information of the to-be-identified address found in the traffic of the power Internet of Things, and realizes adaptive sensitive data determination through joint modeling of security risks, traffic complexity and user policies.

[0134] The adaptive sensitivity threshold formula is as follows:

[0135]

[0136] wherein, is a security risk coefficient, related to attack frequency and vulnerability exposure degree; is data traffic complexity, related to unstructured text proportion and address segmentation hierarchical depth; is a user policy weight, related to user-defined sensitive level; β, γ and δ are weight factors, and β+γ+δ=1. The security risk coefficient (with a value range of 0-1) is an environmental threat level, related to attack frequency and vulnerability exposure degree, and the greater the risk, the greater the value; the data traffic complexity (with a value range of 0-1) is related to unstructured text proportion, address segmentation hierarchical depth and other indicators, and the higher the complexity, the greater the value; the user policy weight (with a value range of 0-1) maps the user-defined sensitive level into a numerical value, for example, mapping low / medium / high in the user-defined sensitive level into corresponding user policy weights 0.3 / 0.5 / 0.7, and the more strict the sensitive mode, the higher the value.

[0137] If Adaptive Sensitivity Threshold: excluding the to-be-identified address found in the traffic of the power Internet of Things as sensitive information; if Adaptive Sensitivity Threshold: determining the to-be-identified address found in the traffic of the power Internet of Things as sensitive information.

[0138] For example, when a DDoS attack is detected and the traffic complexity is low, the user policy is “medium sensitive”, and at this time, the sensitive data is preferentially protected from being leaked. Therefore, the adaptive sensitivity threshold is calculated according to the scene: AST=0.6×0.1+0.2×0.8+0.2×0.5=0.32, and at this time, the hierarchical sensitive coefficient As long as it is greater than or equal to 0.32, it is considered that the to-be-identified address found in the traffic of the power Internet of Things is sensitive information, and less than 0.32, it is excluded that the to-be-identified address found in the traffic of the power Internet of Things is sensitive information. In this example, the security risk coefficient is 0.9; the data traffic complexity is 0.2; the user policy weight is 0.5 (medium sensitive mode); the weight distribution is β=0.6, γ=0.2 and δ=0.2 (satisfying β+γ+δ=1).

[0139] For example, when there is no security threat, but the traffic complexity is high, the user policy is "medium sensitive", and the priority is to avoid false positives. Therefore, the adaptive sensitive threshold is calculated according to the scene: AST = 0.6 * 0.9 + 0.2 * 0.2 + 0.2 * 0.5 = 0.68, and the level sensitive coefficient is If it is greater than or equal to 0.68, it is considered that the address to be identified found in the traffic of the power internet of things is sensitive information, and if it is less than 0.68, it is excluded that the address to be identified found in the traffic of the power internet of things is sensitive information. In this example, the security risk coefficient = 0.1; the data traffic complexity = 0.8; the user policy weight = 0.5 (medium sensitive mode); the weight distribution: β = 0.6, γ = 0.2, δ = 0.2 (satisfying β + γ + δ = 1).

[0140] In the weight calculation stage, a privacy calculation sandbox can also be constructed based on a trusted execution environment (TEE) to securely isolate the generation process of the level sensitive coefficient R and prevent sensitive weight parameters from being leaked during the calculation process.

[0141] 3. Sensitive data monitoring and alarm:

[0142] Through timely and accurate sensitive data identification, sensitive data information in the network traffic under the cloud environment of the power internet of things can be accurately captured, especially data related to personal privacy such as address information, mobile phone number information, and user name information during the process of querying electricity information. The leakage of such information not only may damage the reputation and interests of the enterprise, but also may have a serious impact on the daily operation of the enterprise. Therefore, we must pay great attention to and strictly protect these sensitive information.

[0143] Once such sensitive information is detected under the cloud environment of the power internet of things, a marking and classification mechanism is started to finely label and classify the sensitive data. An emergency disposal process is started, and alarm information is sent to relevant personnel through the email and short message notification mechanism dedicated to the power internet of things. When the alarm is triggered, the leaked data is subjected to "minimum desensitization" through the privacy calculation middleware layer, and differential privacy technology is used to add noise disturbance to the statistical features in the alarm information to avoid the risk of secondary leakage.

[0144] When confirming that sensitive data is being externally transmitted through the network, respond and accurately assess the scope, content and severity of the data leakage to ensure a comprehensive understanding of the situation.

[0145] The security team, legal counsel and management need to be informed of the data leakage event in the first place so that they can take prompt action to reduce potential risks. In order to contain the impact of the leakage, emergency measures such as adjusting firewall rules and suspending affected systems will be taken to minimize the scope of sensitive data leakage.

[0146] In addition, detailed information about the leak is collected, such as the time of the leak, the affected business systems, and possible causes, in order to accurately assess the damage caused by the leak to the enterprise and individuals, including the impact on the company's brand and reputation. Finally, a thorough investigation of the sensitive data leak incident is needed to identify security vulnerabilities and fix them in a timely manner, forming a closed-loop disposal process to ensure the comprehensive protection of sensitive data in the power Internet of Things cloud environment.

[0147] 4. Sensitive data encryption transmission:

[0148] After discovering sensitive data transmission in the power Internet of Things cloud environment, trace the source to find the application source, and need to optimize the transmission mechanism of the application.

[0149] Mainly use symmetric encryption and asymmetric encryption to encrypt sensitive data transmission. From the data source, all-round security protection is carried out. This includes end-to-end encryption transmission between power Internet of Things application calls, Internet of Things centers, sensors, collectors and edge computing centers, to ensure that data is always encrypted in the entire data transmission link, thus ensuring the security of sensitive data transmission.

[0150] According to the characteristics of the power Internet of Things cloud environment, AES and RSA combined encryption method is adopted. This encryption method is optimized to ensure the security of encryption and does not affect the resources and terminal performance of the power Internet of Things cloud environment.

[0151] AES encryption, as a symmetric high-level encryption standard, plays an important role in the power Internet of Things. It adopts block encryption mode, including ECB, CBC, OFB, CTR and CFB five modes, and supports 128, 192, 256 bit key length. This encryption method has fast operation speed and small memory occupation, which is very suitable for encryption demand in the process of power Internet of Things data transmission.

[0152] RSA encryption, as a kind of asymmetric encryption method, provides another layer of security for the power Internet of Things. It contains a pair of public and private keys, with a key length of 1024 bits or more, with high security. Although it will occupy some resources for large data transmission, we mainly use RSA encryption to transmit symmetric keys in this method, thus realizing the perfect combination of security and efficiency, ensuring the secure transmission of keys.

[0153] In the power Internet of Things, from application calls in the cloud environment, terminal data collection to edge computing, we use end-to-end encryption transmission; from the cloud environment to the Internet of Things management center, we realize node-to-node encryption. In the process of data transmission, we use AES-256 mode for encryption by default, and use dynamic key mode for encryption and decryption. When the receiving end receives encrypted data, it uses the private key of RSA to decrypt the AES-Key, obtains the symmetric encryption key, and then uses this key to decrypt the data. In this way, we can ensure the confidentiality of terminal data collection, transmission, use and storage, and effectively prevent the leakage of sensitive data.

[0154] In end-to-end encryption transmission, homomorphic encryption technology is introduced to support cloud-side direct aggregation analysis of ciphertext data (such as power consumption statistics), and to realize the privacy computing feature of "data available but invisible". For cross-institution data collaboration scenarios, secure multi-party computation (MPC) protocol is used to realize multi-party key collaborative management, ensuring that a single node cannot obtain complete key information in encryption and decryption operations.

[0155] 5 Sensitive data access authorization:

[0156] In the power Internet of Things cloud environment, on the one hand, sensitive data in network traffic needs to be identified in a timely manner, and on the other hand, sensitive data needs to be transmitted through encryption to ensure the security of the transmission process. At the same time, in order to ensure the security and integrity of sensitive data, access permissions need to be designed.

[0157] First of all, strong password policies need to be implemented, including long password length and multiple character types, to increase the difficulty of cracking. Second, the Role-Based Access Control (RBAC) model is used to assign different access permissions based on user roles and responsibilities, ensuring that only appropriate users can access sensitive data. In addition, dynamic permission management is introduced to adjust user access permissions in real time based on business needs and data sensitivity. A trusted access control layer based on privacy computing is built: dynamic permission decision-making is implemented in the TEE environment, combined with Attribute-Based Encryption (ABE) technology, to realize "fine-grained data access policy" ciphertext execution. Privacy computing authorization credentials (such as zk-SNARKs) are introduced to support users to complete permission verification without exposing their identity information, meeting the principle of minimizing privacy exposure.

[0158] At the same time, data auditing and monitoring mechanisms are implemented to record all user access behaviors to sensitive data, and set up abnormal behavior alarms to timely discover and handle potential security risks. Finally, security training for employees is strengthened to improve their awareness of sensitive data protection, ensuring the safety of sensitive data in the power Internet of Things cloud environment.

[0159] In conclusion, the application constructs a power Internet of Things data privacy protection system based on a cloud computing environment, realizes the real-time collection and analysis of north-south traffic, the accurate identification of privacy data, the adaptive algorithm of address tree combined with the optimization of cross-domain feature library of federal learning, the encryption transmission, the realization of secret state calculation by introducing homomorphic encryption, the dynamic authorization based on TEE and attribute-based encryption, and the fusion of differential privacy, realizes the "useful but invisible" protection of the whole life cycle of data, discovers sensitive data in the data flow of the cloud environment in a timely manner, provides real-time feedback for data privacy protection, ensures that sensitive data is not illegally intercepted or tampered with during transmission, ensures that only authorized personnel can access related sensitive data in the cloud environment, greatly improves the security of sensitive data, improves the security of sensitive data, reduces the risk of sensitive data leakage, and provides a solid guarantee for the privacy information of the power Internet of Things.

[0160] Although the preferred embodiments of the application have been described, those skilled in the art can make further changes and modifications to the embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to include the preferred embodiments and all changes and modifications falling within the scope of the application.

[0161] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A method for sensitive data privacy protection computing in a cloud environment, characterized in that, The method comprises the following steps: S1, collecting traffic data of the power internet of things, and structurally processing the collected traffic data; S2, under the framework of privacy computing technology, performing multi-dimensional privacy feature extraction on the structurally processed traffic data, and identifying sensitive information including user addresses, names and contact numbers; wherein, an address adaptive identification algorithm is used to identify the user addresses in the traffic data, and the identification process comprises the following sub-steps: S21, target data word segmentation; specifically, the text information obtained in the traffic data is segmented to obtain address segmented segments of multiple levels of administrative regions; S22, initializing a standard address library; specifically, a standard address library is imported, and the standard address library comprises detailed address information and corresponding address Hash data of provinces, cities and districts nationwide; S23, address tree global Hash matching; specifically, Hash data corresponding to the target address to be identified is obtained; the Hash data of the target address is compared with the address Hash data of the standard address library one by one, if equal data is found, it is determined that there is address sensitive data in the traffic data and the address information is output, and the process proceeds to step S3, otherwise, the process proceeds to step S24; S24, address tree node depth and breadth weight matching; the target address to be identified is segmented and Hashed, and based on the standardized address library, the Hash data of the multiple address segmented segments of the target address to be identified is searched in breadth and depth according to the order of administrative region level from large to small, if each address segmented segment can match equal data, it is determined that there is address sensitive data in the traffic data and the address information is output, and the process proceeds to step S3, otherwise, the process proceeds to step S25; S25, for different administrative region levels, the corresponding matching weight is introduced, and the Hash data of the multiple address segmented segments of the target address to be identified is combined to calculate the hierarchical sensitive coefficient of the address tree node depth and breadth, if the hierarchical sensitive coefficient of the address tree node depth and breadth exceeds a preset threshold, it is determined that there is address sensitive data in the traffic data of the power internet of things and the address information is output, and the process proceeds to step S3, otherwise, it is determined that there is no address sensitive data in the traffic data of the power internet of things, and the process ends; S3, marking and grading the identified sensitive information, and monitoring and protecting according to the grading result; Step S25 further comprises: The address sensitive information to be identified found in the traffic of the power internet of things is segmented to obtain multiple administrative district level address data including provincial, municipal, district, street and detail address information, Hash calculation is performed on the address data at each level to obtain a unique Key value H S ; The known standard address library is used to build full address information of multiple levels including full address information of provincial level, full address information of municipal level, full address information of district level, full address information of street level and full address information of detail address information. Hash calculation is performed on the full address information to obtain corresponding full Key values, and a nationwide address Hash address tree is formed, wherein each node of the tree is H d ; The address sensitive information to be identified found in the traffic of the power internet of things and the address tree of the national address Hash are compared in weighted Hash values, and a hierarchical sensitive coefficient is obtained through hierarchical dependence and weighted average In the formula, is a hierarchical sensitivity coefficient, used to represent the sensitivity degree of address data comprehensive score; is the dynamic weight of the i-th level administrative region, which is dynamically adjusted by the preset weight and the corresponding security risk coefficient of the corresponding administrative region; represents the hash matching similarity of the i-th level administrative region address; represents the progressive hierarchical dependence, wherein k is the upper level number, s κ is the matching degree of the k-th level address; λ represents the hierarchical attenuation compensation factor, which is negatively correlated with the data flow complexity; σ() represents the dynamic threshold function.

2. The sensitive data privacy protection computing method in the cloud environment according to claim 1, characterized in that, In step S1, a bypass mode is used to copy the current network traffic on the cloud internet of things in real time, connect the north-south network traffic of the unified entrance and exit of the external network, and communicate data between the network inside and outside, that is, the network traffic across the network boundary vertically.

3. The sensitive data privacy protection computing method in the cloud environment according to claim 1, characterized in that, In step S1, the process of structurally processing the collected traffic data comprises the following steps: Data cleaning is performed on the collected traffic data to remove invalid, expired, duplicate and error data; The cleaned traffic data is uniformly converted in format; Public parameters including storage time and collection time are added to the traffic data after format conversion.

4. The sensitive data privacy protection computing method in the cloud environment according to claim 1, characterized in that, In step S2, the identification process of the name comprises the following steps: According to the surname, the matching of the hundred surnames is carried out, and the single-word and double-word Chinese names are screened, so that all possible Chinese names in the sentence are preliminarily identified; The preliminarily identified Chinese names are matched with the words in the dictionary, and all possible combinations are output; The identification process of the contact number includes the following steps: The input text is converted into a unified format, and the punctuation and spaces contained therein are removed; Regular expressions are used to find strings that match the mobile phone number format; A secure multi-party computation protocol in privacy computing is introduced to cooperatively verify the mobile phone matching rules in a distributed manner.

5. The sensitive data privacy protection computing method in the cloud environment according to claim 1, characterized in that, In step S2, the text information obtained in the traffic data of the power internet of things is segmented, and address segmentation segments of multiple levels of administrative regions are obtained, and if part of the address segmentation segments are empty, the empty values are filled, to obtain: S={S1, S2, S3, S4, S5}; Where S is the target address information to be identified, S1 represents the province data to be identified, S2 represents the city data to be identified, S3 represents the district data to be identified, S4 represents the street data to be identified, and S5 represents the detailed address data to be identified; The standard address library D is imported: Wherein, m represents the total number of all complete addresses in the address database; D1 represents the province data of the standard address, D2 represents the city data of the standard address, D3 represents the district data of the standard address, D4 represents the street data of the standard address, and D5 represents the detailed address data of the standard address. The address data in the standard address library D is hashed to obtain the corresponding address Hash data: In the formula, H d Hash information in the address information of the standard address library, HD1 represents Hash data of a province, HD2 represents Hash data of a city, HD3 represents Hash data of a district, HD4 represents Hash data of a street, and HD5 represents Hash data of a detailed address.

6. The sensitive data privacy protection computing method in the cloud environment according to claim 5, characterized in that, In step S23, the process of global Hash matching of the address tree includes the following steps: S231, Hash the target address to be identified in the power internet of things traffic to obtain the following three types of data: Hash data S of province, city, district, and detailed address h1 = Hash * (S1 + S2 + S3 + S4 + S5), Hash data S of province, city, district, and detailed address h2 = Hash * (S3 + S4 + S5), Hash data S of district and detailed address h3 = Hash * (S4 + S5) S232, go to S h1 and D h1 are compared, if equal, it is found that there is address sensitive data in this piece of traffic data and the address information is a province S1, a city S2, a district S3, a street S4 and a detailed address S5; otherwise, go to step S233; S233, S h2 and D h2 are compared, if equal, it is found that there is address sensitive data in this piece of traffic data and the address information is a certain district S3, a certain street S4 and a certain detailed address S5, otherwise, go to step S234; S234, S h3 and D h3 comparison, if equal, found that this traffic data exists address sensitive data and address information for a street S4 some detailed address S5; In the formula, D h1 Hash data after splicing of certain street detailed address information of a province, city or district obtained by traffic in the power Internet of Things; D h2 Hash data after splicing of certain district street detailed address information obtained by traffic in the power Internet of Things; D h3 Hash data after splicing of certain street detailed address information obtained by traffic in the power Internet of Things.

7. The sensitive data privacy protection computing method in the cloud environment according to claim 6, characterized in that, Step S24 further includes: S241, the target address to be identified is segmented into words and hashed, using the hash value H from S1. s1 Perform a breadth-first search until a match is found with H. s1 Same standard address, provincial data H d1 Proceed to step S242. If no matching data can be found, proceed to step S25. S242, using the Hash value H of the city S2 s2 A deep search is performed until the same data H is searched s2 with the same standard address city S2 d2 , and the process goes to step S243, and if the same data cannot be searched, the process goes to step S25; S243, using the Hash value H of the area S3 s3 a deep search is performed until the same data H s3 of the same standard address area is found d3 , and the process goes to step S244, and if the same data cannot be found, the process goes to step S25; S244, using the Hash value H of the street S4 s4 A deep search is performed until the same standard address street data H s4 is searched for d4 Step S245 is entered if the same data cannot be searched for, and Step S25 is entered. S245, using the hash value H of the detailed address S5. s5 Perform a depth-first search until a match is found with H. s5 Same standard address detailed address data H d5 Proceed to step S246. If no matching data can be found, proceed to step S25. S246, integrate H d1 , H d2 , H d3 , H d4 and H d5 , get the identification result of the target address to be identified.

8. The sensitive data privacy protection computing method in a cloud environment according to claim 1, characterized in that, The preset threshold value is an adaptive sensitive threshold value, which is a boundary value of the sensitive degree of the address to be identified found in the traffic of the power internet of things. The adaptive sensitive threshold value is used for adaptive adjustment of the sensitive data judgment level through joint modeling of security risks, traffic complexity and user strategies.

9. The sensitive data privacy protection computing method in the cloud environment according to claim 8, characterized in that, The calculation formula of the adaptive sensitive threshold value AST is: wherein, is a security risk coefficient, related to attack frequency and vulnerability exposure degree; is a data flow complexity, related to the proportion of unstructured text and address segmentation level depth; is a user policy weight, related to user-defined sensitivity level; β, γ and δ are weight factors, and β+γ+δ=1; If Less than the adaptive sensitive threshold, the to-be-identified address found in the traffic of the power internet of things is excluded as sensitive information; if Greater than or equal to the adaptive sensitive threshold, the to-be-identified address found in the traffic of the power internet of things is determined as sensitive information.

Images