Data security protection method and system for edge reliable elastic network
By installing edge devices on server network equipment, collecting and analyzing data log information from edge computing nodes, and using Laplace matrix eigenvector clustering to assess security status, the problem of data information collection and anomaly identification in edge computing networks is solved, thereby improving network security and response speed.
Patent Information
- Application Number
- CN202411252268.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-09
AI Technical Summary
In existing technologies, how to effectively collect and analyze data information from edge computing nodes and identify abnormal data to improve network security is an urgent problem to be solved.
By installing edge devices on server network equipment, raw data from each edge computing node is collected, data log information is analyzed, and the eigenvector clustering method of the Laplace matrix is used to evaluate task response efficiency labels, determine the security status of network devices, and implement enhanced protection when insecure.
It enables accurate security assessment and dynamic adjustment of routing strategies for edge computing networks, improving the security and response speed of the network environment.
Smart Images

Figure CN118972151A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of edge computing, and in particular to a data security protection method and system for an edge reliable elastic network. Background Art
[0002] Research has found that data information security and network security have received widespread attention and importance in many fields. At the same time, edge reliable and elastic network security devices have also been widely used. For example, the edge reliable and elastic network security handheld device is a device that utilizes a new type of data sharing software or sharing platform. Through this software, secure information sharing between various edge reliable and elastic network security devices can be achieved.
[0003] The shared software or shared platform can help various user terminals (such as the above-mentioned handheld devices) access and form a secure and flexible cloud security network.
[0004] Edge computing is a computing model that sinks services such as data processing and content distribution to edge devices (edge reliable and resilient network security devices) close to the data source, thereby reducing the burden on the cloud (i.e., server network devices), improving response speed and reducing network transmission costs.
[0005] In the current server network environment, edge devices are widely used for network security monitoring and management.
[0006] Edge devices usually have powerful computing and data processing capabilities, and can perform comprehensive scanning and data collection on each edge node on the server network equipment.
[0007] An edge computing node (such as a shared software or shared platform or edge evaluation module installed on an edge reliable and resilient network security device) refers to an independent computing node located close to the user and has data computing and communication transmission capabilities.
[0008] When a user accesses network information, the request will first reach the edge computing node, which will forward it to the edge device after confirming it is safe. These edge nodes will retain a large amount of access data information, that is, the original data.
[0009] Research has found that this comprehensive scanning and data analysis method based on edge devices can help to timely discover and prevent network security risks and improve the security of server network equipment; however, how to collect data information and analyze related abnormal data from it, so as to quickly and reasonably analyze network security, is an urgent problem that technicians in this field need to solve. Summary of the invention
[0010] The purpose of the present invention is to provide a data security protection method and system for an edge reliable elastic network, which solves the above-mentioned technical problems pointed out in the prior art.
[0011] The present invention provides a data security protection method for an edge reliable elastic network, comprising the following steps: Install multiple edge devices (i.e., edge reliable elastic network security devices) on the server network device (referred to as cloud server), and the edge devices scan all edge computing nodes (or edge evaluation modules) on the server network device; The edge device obtains original data from each edge computing node of the server network device, obtains data log information of each edge computing node through the original data, and determines the security result of the server network device by evaluating the data log information; if the evaluated security result is unsafe, an alarm message is sent to the server network device and an emergency response operation of enhanced protection processing is implemented.
[0012] Accordingly, the present invention also proposes a data security protection system for edge reliable elastic network, comprising: an edge node scanning module, an evaluation server network device module, and a subsequent processing module; The scanning edge node module is used to install an edge device on the server network device, and the edge device scans all edge computing nodes of the server network device; The server network device evaluation module is used for the edge device to obtain original data from each edge computing node of the server network device, obtain data log information of each edge computing node through the original data, and determine the security result of the server network device by evaluating the data log information; The subsequent processing module is used to send an alarm message to the server network device if the security result of the assessment is unsafe.
[0013] Accordingly, the present application also relates to a storage medium, which includes: Memory for storing computer programs; A communication interface, used to connect the memory to the processor; A processor is used to execute a computer program to implement the steps of the data security protection method for an edge reliable elastic network as described in any of the above items.
[0014] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages: It can be seen from the analysis of the data security protection method of the above-mentioned edge reliable elastic network provided by the present invention that, in specific applications, the edge node is scanned by installing the edge device on the server network device, and the data log information is analyzed through the original data of each edge node, and the data log information includes task access request information, task response information and performance information; the daily task description feature value Fn of the edge node is calculated according to the task access time of the task access request information, the task response time of the task response information and the bandwidth utilization rate of the performance information, and the task response efficiency status of the edge node on the nth day can be obtained; Furthermore, firstly, by collecting the data of the daily task description eigenvalue Fn of the current edge computing node, and performing eigenvector clustering of the Laplace matrix according to the data of the daily task description eigenvalue Fn within a certain period of time, a task response efficiency label is obtained. The task response efficiency label includes four levels, namely, a high-efficiency task response efficiency label, a medium-efficiency task response efficiency label, a low-efficiency task response efficiency label, and a severely low-efficiency task response efficiency label; the assessment security result of the current edge computing node is judged according to the task response efficiency label; Furthermore, through the task response efficiency label finally obtained by the daily task description characteristic value Fn of the current edge computing node mentioned above, the daily task description characteristic values Fn of multiple edge computing nodes are collected, and the daily task description characteristic values Fn within a specific time are clustered to obtain the task response efficiency label of each edge node, and the evaluation safety result of each edge node is determined. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0016] Figure 1 A flow chart of a data security protection method for an edge reliable elastic network of the present invention; Figure 2 It is an evaluation flow chart of a task response efficiency label of a data security protection method for an edge reliable elastic network of the present invention; Figure 3 An evaluation flow chart of calculating the task response efficiency label of the current edge computing node in a data security protection method for an edge reliable elastic network of the present invention; Figure 4 An evaluation flow chart of calculating the task response efficiency label of each edge node in a data security protection method for an edge reliable elastic network of the present invention; Figure 5 A schematic diagram of a first evaluation list of a data security protection method for an edge reliable elastic network of the present invention; Figure 6 A schematic diagram of a second evaluation list of a data security protection method for an edge reliable elastic network of the present invention; Figure 7 A process for allocating a corresponding number of edge computing nodes to a first target edge node set of a data security protection method for an edge reliable elastic network of the present invention; Figure 8 A flowchart of allocating subtasks to the same edge computing node of a data security protection method for an edge reliable elastic network of the present invention; Fig. 9 A flowchart of allocating a corresponding number of edge computing nodes to all target edge node sets of a data security protection method for an edge reliable elastic network of the present invention; Fig.10 A flow chart of a data security protection system for an edge reliable elastic network provided in Embodiment 2 of the present invention; Fig.11 A storage medium provided in Embodiment 3 of the present invention.
[0017] Label: scanning edge node module 10, evaluation server network device module 20, subsequent processing module 30; processor 1110; communication interface 1120; memory 1130; computer storage medium 1140. DETAILED DESCRIPTION
[0018] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.
[0019] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making any creative work shall fall within the scope of protection of the present invention.
[0020] The present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings.
[0021] Embodiment 1 like Figure 1 As shown, the present invention proposes a data security protection method for an edge reliable elastic network, comprising the following steps: S10: installing an edge device on the server network device, wherein the edge device scans all edge nodes of the server network device; It should be noted that edge devices are installed on server network devices. Edge devices usually have powerful computing and data processing capabilities, and can also collect data information, provide identity authentication, and perform data analysis. At the same time, edge nodes refer to nodes that are very close to users and are not on the backbone network (the node is an independent computing node with data computing capabilities and communication data transmission capabilities, and can be an independent computing unit). When users access information in the network, the request will first reach the edge node, and then be gradually forwarded to the core node by the edge node; thus, these edge nodes will have a large amount of access data information (i.e., original information); S20: The edge device obtains original data from each edge node of the server network device, obtains data log information of each edge node through the original data, and determines the security result of the server network device by evaluating the data log information; if the evaluated security result is unsafe, an alarm message is sent to the server network device and an emergency response operation of enhanced protection processing is implemented.
[0022] It should be noted that the edge device collects raw data from each edge node, analyzes the raw data, and extracts data log information from the raw data of each edge node; determines the characteristic value through the data log information, and then performs cluster analysis on the quantified characteristic values, thereby finally judging whether the current security result is safe.
[0023] The data log information collected by the embodiment of the present application includes at least four aspects of information, including access request information (or task access request information), response information (or task response information) and performance information; specific research found that the task access request information mainly controls: task access time, access source IP address, access method (GET, POST, etc.), request parameters, request header information; task response information: task response time, response status code, response size; performance information: CPU / memory / bandwidth utilization, throughput, etc.; However, the embodiment of the present application mainly focuses on the information related to task response efficiency in some of the above-mentioned task access request information, task response information and performance information, such as the average task response time, the average task access time of daily tasks, the bandwidth utilization of edge nodes at the current moment, etc.; The purpose of the present invention is to overcome the shortcomings of the above-mentioned existing technologies and propose a data security protection method and system for an edge reliable elastic network, which mainly focuses on comprehensive and accurate analysis of security assessments and targeted redistribution of data traffic to edge computing nodes, thereby adapting to changes in the network environment by dynamically adjusting routing strategies.
[0024] Specifically, Figure 2As shown, in step S20, the edge device obtains original data from each edge node of the server network device, obtains data log information of each edge node through the original data, and determines the security result of the server network device by evaluating the data log information. The specific operation steps are as follows: S21: Collect data from each edge node through the edge device to obtain raw data; analyze the raw data to obtain data log information of each edge node, where the data log information includes: task access request information, task response information, and performance information; S22: Obtaining a task description feature value Fn of the edge node corresponding to each task according to the task access request information of the current edge computing node and each task response information and performance information; That is, by obtaining the access tasks received by the current edge computing node, and then obtaining the task access request information (task access request information includes task access time and access mode type) and each task response information (task response information includes task response time) and performance information (edge node bandwidth utilization, response delay, and throughput) to obtain the edge node task description feature value corresponding to each task; Calculate the daily task description feature value Fn of the nth day, taking into account the daily task volume of the edge node, the response time and access time of each task, the average response time and average access time of the daily tasks, and the edge node bandwidth utilization at the current moment; the formula is as follows: ; Where, Fn: daily task description feature value of day n; h: total number of tasks processed daily; i: the i-th task running on that day; Zi: task response time of the i-th task; Ki: task access time of the i-th task; D: average task response time of daily tasks; C: Average task access time of daily tasks; b: bandwidth utilization of edge nodes at the current moment; α, β, γ: weight coefficients, used to adjust the impact of various parameters on the eigenvalues of task description; It should be noted that, from the above formula, we can know that task response time (Zi) and task access time (Ki): these two parameters reflect the specific performance of each task; average response time (D) and average access time (C) of daily tasks: these parameters are used to standardize the response time and access time of each task for comparison between different days; bandwidth utilization (b): real-time reflection of the bandwidth utilization of the current edge computing node; weight coefficient (α, β, γ): used to adjust the influence weight of each parameter in the task description feature value; through the above formula, the task response efficiency of the edge node on the nth day can be comprehensively reflected, taking into account the specific situation of task response and access and the influence of bandwidth utilization; Explanation: Task access request information (task access time, access method type) and each task response information (task response information includes task response time) and performance information (edge node bandwidth utilization, response latency, and throughput); S23: Collect data of daily task description feature values Fn of the current edge computing node, perform Laplace matrix eigenvector clustering on the daily task description feature values Fn corresponding to the current edge computing node within a preset time period, implement clustering output of the daily task description feature values of the current edge computing node, and finally output the task response efficiency label of the daily task of the current edge computing node; It should be noted that clustering is performed using the eigenvectors of the similarity matrix (or Laplace matrix) of the data. In the above step S23, it is actually a preparation for the subsequent output of the task response efficiency label of the daily task of the current edge computing node; of course, the following step S24 is entirely for the evaluation of the task response efficiency achieved on a specific time node day (for example, a specific time node day is often susceptible to network influence, resulting in unsafe or unreliable problems of the edge node, for example: a cloud service server of an application website or application system is susceptible to external security attacks and influences before and after special times or special holidays and on the day, and is prone to problems at this specific time node); Therefore, step S23 is to cluster and label the task description feature value (Fn) of a current specific edge node through the task response efficiency status in different time periods by clustering, so as to predict the stability of the task response efficiency status of the above-mentioned current edge computing node in different time periods; the above-mentioned task response efficiency label can be divided into four levels, namely, high-efficiency task response efficiency label, medium-level task response efficiency label, low-level task response efficiency label and severely inefficient task response efficiency label.
[0025] S24: Collect data on daily task description feature values Fn of multiple edge computing nodes, perform Laplace matrix eigenvector clustering on the task description feature values Fn of the current multiple edge computing nodes at a specific time node day, achieve clustering output of task description feature values of current different edge nodes, and finally output task response efficiency labels of multiple edge computing nodes to achieve evaluation output of multiple edge computing nodes; It should be noted that step S24 is based on step S23, and the evaluation scope is expanded to multiple edge computing nodes, and the daily task description feature values (Fn) of multiple edge computing nodes are collected at a specific time node; and the daily task description feature values (Fn) of multiple edge computing nodes are collected at a specific time node using the same method as S23 (i.e., eigenvector clustering based on the Laplace matrix) for cluster analysis; Therefore, step S24 is to cluster and label the task response efficiency status of the task description feature values (Fn) of multiple edge computing nodes at a specific time node, thereby assigning a cluster label (i.e., task response efficiency label) to the task response efficiency of each edge node at a specific time node. The final output can compare the performance of different edge nodes at the same time node and identify the security assessment result of each edge node.
[0026] The above-mentioned edge device collects raw data from each edge node, analyzes the raw data, and extracts data log information from the raw data of each edge node; determines the characteristic value through the data log information, and then performs cluster analysis on the quantified characteristic value, that is, step S24 will eventually classify the edge computing nodes in the form of two evaluation lists in a labeled manner, so as to ultimately determine whether the security results of the current cloud server are safe (or to ultimately determine whether the security results of the numerous edge computing nodes in the current cloud server are safe); if most of the edge computing nodes in the current server network device (referred to as the cloud server) (the proportion is greater than m, m=65%) are classified as low task response efficiency labels and severely inefficient task response efficiency labels, it can be simply and directly assessed that the current cloud server is in an unsafe state.
[0027] In the specific technical solution, Figure 3 As shown; in step S23, the data of the daily task description feature value Fn of the current edge computing node is collected, and the daily task description feature value Fn corresponding to the current edge computing node within the preset time period is clustered by the eigenvector of the Laplace matrix, so as to realize the clustering output of the daily task description feature value of the current edge computing node, and finally output the task response efficiency label of the daily task of the current edge computing node. The specific operation steps are as follows: S231: Collect data of daily task description feature values Fn (take the current edge computing node as the target edge node, and collect the daily task description feature values Fn of the target edge node on different dates); S232: Calculate the similarity matrix W between the data points (daily task description eigenvalues Fn) of the current edge computing node, use the Gaussian kernel function to perform the calculation, and construct the similarity matrix; The Gaussian kernel function calculation formula is as follows: ; Here, σ is the parameter of the Gaussian kernel.
[0028] When calculating the similarity matrix W between two data points (the daily task description feature values Fn of the current edge computing node on two different dates), Fi and Fj represent any two data points. For the daily task description feature value Fn of the edge node, the data points here are Fi and Fj; Fi is the daily task description feature value of the current edge computing node on the i-th day; Fj is the daily task description feature value of the current edge computing node on the j-th day.
[0029] Therefore, Fi and Fj represent the daily task description feature values of the same edge node on different dates; S233: constructing a Laplace matrix L using the similarity matrix W; First, calculate the degree matrix D; ; The Laplace matrix L is calculated from the above degree matrix D: L = D − W; S234: Calculate the eigenvectors corresponding to the first k smallest eigenvalues of the Laplace matrix L to form a matrix U; S235: Take the rows of the eigenvector matrix U as new data points (all new data points are collectively referred to as first data points here, and the first data points are new eigenvalues generated by vectorization for expressing task response efficiency), use K-means to cluster the new data points, and output the clustering results of the current new data points (i.e., obtain the eigenvector clustering).
[0030] S236: Map the clustering results of the new data points to the original data, output the clustering results and assign a clustering label (i.e., task response efficiency label) to the daily task description feature value Fn of each current edge computing node (or target edge node); in this way, a clustering label can be assigned to the daily task description feature value Fn of each current edge computing node (or target edge node).
[0031] Generally speaking, if the collected raw data is all the data within one month corresponding to the current edge computing node, cluster analysis and evaluation can be performed to output the task response efficiency label of the current edge computing node within one month; generally speaking, task response efficiency labels can be divided into four levels, namely, high-efficiency task response efficiency labels, medium-level task response efficiency labels, low-level task response efficiency labels, and severely inefficient task response efficiency labels.
[0032] Explanation: The eigenvectors corresponding to the first k smallest eigenvalues calculated form a matrix U. The matrix U is considered as a new feature matrix, in which each row represents a data point. These new data points are clustered using the K-means clustering algorithm; K-means is an unsupervised learning method used to divide data points into K clusters.
[0033] Finally, the clustering result output is realized: after the clustering is completed, each feature vector will be assigned to a specific cluster and obtain a cluster label.
[0034] By mapping these cluster labels back to the original data, a cluster label (i.e., task response efficiency label) can be assigned to the daily task description feature value Fn of each edge node (or target edge node).
[0035] It should be noted that, during the execution of the above steps S231-S226, when calculating the similarity matrix W between data points, Fi and Fj represent any two data points.
[0036] In the above specific application scenario, the daily task description feature value Fn of the edge node is used; therefore, Fi and Fj represent the daily task description feature values of the same edge node on different dates.
[0037] In the specific technical solution, Figure 4 As shown; in step S24, the data of daily task description feature values Fn of multiple edge computing nodes are collected, and then Fn is used in combination with the historical efficiency average value FT to calculate the daily task description feature value Fn'; then the eigenvector clustering of the Laplace matrix is performed on the task description feature values Fn' of the current multiple edge computing nodes at a specific time node day, so as to achieve clustering output of the task description feature values of the current different edge nodes, and finally output the task response efficiency labels of multiple edge computing nodes, so as to achieve the evaluation output of multiple edge computing nodes; During the execution of step S24, the scenario considers the specific classification labels and response efficiency performance of different edge nodes in terms of task execution response efficiency at a specific time node. Therefore, when constructing the similarity matrix, the similarity matrix of different edge nodes at the same time is selected. The specific operation steps are as follows: S241: Collect data on daily task description feature values Fn of multiple edge computing nodes at a specific time node t (and the corresponding historical time moment t can be obtained from the current time node t); at the same time, collect the historical efficiency average FT of the efficiency values of all edge nodes at the same current historical time moment t.
[0038] It should be noted that the time feature t and the historical efficiency average are added to the dataset as additional feature dimensions to consider the impact of the time factor on the task response efficiency.
[0039] Calculate the daily task description feature value Fn' corresponding to the current edge computing node at a specific time node t; Fn'=Fn-historical efficiency average FT; S242: Calculate the similarity matrix W' between the daily task description feature values Fn' corresponding to the current specific time node day of any two edge nodes, use the Gaussian kernel function to perform the calculation, and construct a similarity matrix; The Gaussian kernel function calculation formula is as follows: ; in; Indicates the current The daily task description feature value corresponding to the edge node at the current specific time node day; It is represented as the daily task description feature value corresponding to the current p-th edge node at the current specific time node day; The above formula can be used to calculate the similarity matrix W' between the daily task description feature values corresponding to any two edge nodes at the current specific time node day; S243: constructing a Laplace matrix L' using the similarity matrix W'; First, calculate the degree matrix D'; ; The Laplace matrix L' calculated from the above degree matrix D' is: L'=D'−W'; S244: Calculate the eigenvectors corresponding to the first k' smallest eigenvalues of the Laplace matrix L' to form a matrix U'; S245: Take the rows of the eigenvector matrix U' as new data points (all new data points are collectively referred to as second data points; the second data points are new eigenvalues generated by vectorization for expressing task response efficiency), cluster the new data points using K-means, and output the clustering results.
[0040] S246: Map the clustering results back to the original data, output the clustering results and assign a clustering label (i.e., task response efficiency label) to the daily task description feature value Fn' of each current edge computing node; confirm the output of the assessment safety result through the clustering label (i.e., task response efficiency label) obtained by each current edge computing node.
[0041] It should be noted that, by assigning a clustering label to the daily task description feature value Fn of the current edge computing node through step S23, a task response efficiency label of an edge node (i.e., an efficient task response efficiency label, a medium task response efficiency label, a low task response efficiency label, and a severely inefficient task response efficiency label) can be determined; by clustering the eigenvectors of the Laplace matrix of the task description feature values Fn' of multiple edge computing nodes at a specific time node day through step S24, the task response efficiency labels of multiple edge computing nodes are obtained, thereby analyzing the task response efficiency label of each edge node and determining the security result of the server network device; In the preferred technical solution, clustering and outputting the task description feature values of different edge nodes at present is realized, and finally outputting the task response efficiency labels of multiple edge computing nodes (the data log information of each edge node is obtained through the original data, and the security result of the server network device is determined by evaluating the data log information), which specifically includes: The task response efficiency label of the daily task of the current edge computing node and the task response efficiency labels of the multiple edge computing nodes are used to respectively establish evaluation lists, thereby obtaining a first evaluation list and a second evaluation list; specifically, the first evaluation list is established by using the task response efficiency label of the daily task of the current edge computing node; the first evaluation list is a collection list of the task response efficiencies of the current edge computing node in different time periods. However, the second evaluation list is formed by using a collection list constructed by using the task response efficiency labels of multiple edge computing nodes within a specific date; The first evaluation list includes a first owning edge node set, a second owning edge node set, a third owning edge node set, and a fourth owning edge node set; The second evaluation list includes a first target edge node set, a second target edge node set, a third target edge node set, and a fourth target edge node set; It should be noted that the two evaluation processes include the clustering analysis and classification results of the task response efficiency labels of the current edge computing node in different time periods, and the other is the clustering analysis and classification results of the task response efficiency labels of multiple edge computing nodes on a specific day; After implementing the two evaluation processes, at least two lists can be established; the first evaluation list is a set list of task response efficiencies of the current edge computing node in different time periods (different dates), and the first evaluation list is divided into four level sets; For example, the first evaluation list divides the current edge computing node from 0:00 on November 2 to 0:00 on November 30 into four sets of task response efficiency label clustering classification results at countless times (note that the interval can be one day), namely, the first edge node set, the second edge node set, the third edge node set, and the fourth edge node set. Figure 5 As shown; Among them, the first original edge node set (this set collects the data of the task response efficiency label clustering classification result of the current edge computing node on a certain day being divided into the high-efficiency task response efficiency label), the second original edge node set (this set collects the data of the task response efficiency label clustering classification result of the current edge computing node on a certain day being divided into the medium-level task response efficiency label), the third original edge node set (this set collects the data of the task response efficiency label clustering classification result of the current edge computing node on a certain day being divided into the low-level task response efficiency label)), the fourth original edge node set (this set collects the data of the task response efficiency label clustering classification result of the current edge computing node on a certain day being divided into the seriously inefficient task response efficiency label)); The second evaluation list is a data set of clustering and categorizing the task response efficiency labels of multiple edge computing nodes according to time periods within a specific date; For example, if the current specific date is a shopping festival on a server website, such as November 11-19, then the task response efficiency label clustering results of multiple edge computing nodes on each date are obtained by dividing them into 1-day intervals from November 11, November 12, November 13, etc., such as Figure 6 As shown; The second evaluation list divides the countless edge nodes from November 11 to 19 into four sets. For example, although the second evaluation list records the clustering and classification results of the task response efficiency labels of countless edge nodes on the current date, the list is only divided into four sets, which are, in order, the first target edge node set (this set is the edge computing nodes corresponding to all high-efficiency task response efficiency labels), the second target edge node set (this set is the edge computing nodes corresponding to all medium task response efficiency labels), the third target edge node set (this set is the edge computing nodes corresponding to all low-level task response efficiency labels), and the fourth target edge node set (this set is all the edge computing nodes corresponding to the severely inefficient task response efficiency labels).
[0042] Specifically, after step S25, it also includes receiving the execution instruction of the distribution task sent by the current server network device, splitting the current distribution task into multiple subtasks, and screening them according to the combination of the number of subtasks, the first evaluation list, and the second evaluation list, outputting the IP addresses of the optimal multiple edge computing nodes to achieve the optimal multiple edge computing nodes to allocate and execute the above multiple subtasks one by one. The specific operation steps are as follows: S251: receiving the execution instruction of the distribution task sent by the current server, detecting the time and date of the current distribution task, and determining the second evaluation list corresponding to the time and date of the current distribution task (there is a predicted second evaluation list every day, so the corresponding second evaluation list should be determined according to the time); Detect the number N of the current subtasks to determine whether the number N of the current subtasks is less than the number of edge computing nodes in the first target edge node set of the second evaluation list (i.e., whether it is less than the number of edge computing nodes corresponding to all high-efficiency task response efficiency labels in the first target edge node set of the second evaluation list); or execute the operation step of S252, so S251 and S252 can be selected and executed in parallel; If so, randomly assigning the same number of edge computing nodes in the first target edge node set to the multiple subtasks in a first priority manner; It should be noted that the number N of current subtasks is detected. If the number N of current subtasks is less than the number of edge computing nodes corresponding to all high-efficiency task response efficiency labels in the first target edge node set in the second evaluation list, then N random edge computing nodes in the first target edge node set are allocated to the above-mentioned multiple subtasks in a first priority manner (that is, if it is less than, then N random edge computing nodes in the first target edge node set that are the same as the number N of current subtasks are screened out and allocated to multiple subtasks (that is, N number of subtasks); the number of current subtasks N is the same as the number of edge computing nodes randomly screened in the first target edge node set), as shown in FIG. Figure 7 As shown; The random allocation of the edge computing nodes to the multiple subtasks also includes another method of randomly allocating N edge computing nodes; or executing S252, matching multiple edge computing nodes in the first target edge node set in the second evaluation list with the first belonging edge node set in the first evaluation list to solve edge computing nodes with the same IP address, and obtaining a plurality of identical edge computing nodes; Clustering the plurality of identical edge computing nodes to form a new data set; allocating a plurality of edge computing nodes in the new data set to the plurality of subtasks; It should be noted that there is another way to randomly assign N edge computing nodes, as follows: screen N edge computing nodes, i.e. S252, match multiple edge computing nodes in the first target edge node set in the second evaluation list from the first belonging edge node set in the first evaluation list to solve edge computing nodes with the same IP address (i.e. the same edge computing nodes), and then screen several identical edge computing nodes to form a new data set, and assign the multiple edge computing nodes in the matched new data set one by one to the above-mentioned multiple subtasks; therefore, the label evaluation in the second evaluation list predicted for the day (i.e. the efficient task response efficiency label) enjoys a higher priority, is first screened from it, and then screened in combination with the first evaluation list, and finally a new data set of the optimal multiple edge computing nodes is obtained, and the IP addresses of the optimal multiple edge computing nodes are pushed to the cloud server, such as Figure 8 As shown; S253: If not, assigning all edge computing nodes in the first target edge node set to multiple subtasks; If all edge computing nodes in the first target edge node set are insufficient, all edge computing nodes in the second target edge node set, the third target edge node set, and the fourth target edge node set are sequentially allocated to multiple subtasks until all N subtasks are allocated; It should be noted that if the number N of current subtasks is greater than or equal to the number of edge computing nodes corresponding to the efficiency labels of all high-efficiency task responses in the first target edge node set in the second evaluation list, all edge computing nodes in the first target edge node set are allocated to the above-mentioned multiple subtasks, and on this basis, the number of all corresponding edge computing nodes in the second, third, and fourth target edge node sets are supplemented successively until all subtasks N are satisfied after supplementation to obtain one-to-one corresponding edge computing nodes; as of S253, a new set of data can also be output separately for push, and of course, the subsequent step S255 can be continued to push the optimal data set (i.e., the optimal edge computing node), such as Fig. 9 As shown; S254: After the above preset logic steps, monitor and calculate in real time the task load score of the edge computing node corresponding to the current subtask that has been screened; S255: Construct a third evaluation list using the task load scores of the multiple edge computing nodes corresponding to the above preset logic steps (i.e., the first target edge node set of S251-S253 and the corresponding edge computing nodes in the first own edge node set assigned to the subtask); the third evaluation list is sorted in descending order according to the task load scores of the multiple edge computing nodes; a new data set is formed according to the corresponding multiple edge computing nodes and pushed to the cloud server; The preset logic step includes any one of step S251, S252 or S253; when sequential screening is performed during the execution of step S255, the task load scores of the edge computing nodes corresponding to the multiple selected edge computing nodes determined in step S251, S252 or S253 are used to form a third evaluation list in which the task load scores are sorted from high to low. After the above preset logic steps (i.e., step S251, S252 or S253), step S254 can be jumped for execution.
[0043] It should be noted that the third evaluation list is actually a sorted data set after filtering again, that is, the task load scores of the edge computing nodes corresponding to the multiple selected edge computing nodes in step S251 or S252 or S253 are filtered again in sequence to form a third evaluation list in which the task load scores are sorted from high to low; the multiple edge computing nodes in the third evaluation list are formed into a new data set and pushed to the cloud server (for example: the second priority filtering behavior of the multiple selected edge computing nodes in step S251 is formed into a third evaluation list, and the multiple edge computing nodes in the third evaluation list are formed into a set and pushed to the cloud server to solve the problem that when the current load task of the selected edge computing node pre-assigned to the subtask is too large, the current edge computing node will be recommended and pushed to realize task collaborative processing among multiple edge computing nodes).
[0044] This embodiment prepares a task load score, for example: the task load scores of the edge computing nodes corresponding to the multiple selected edge computing nodes in step S251 are formed into a third evaluation list in which the task load scores are sorted from high to low by numerical values, and the multiple edge computing nodes in the third evaluation list are formed into a new data set and pushed to the cloud server (or the current server network device); that is, the second priority screening behavior of the multiple selected edge computing nodes in step S251 is formed into a third evaluation list, and the multiple edge computing nodes in the third evaluation list are formed into a set and pushed to the cloud server, so as to solve the problem that when the current load task of the selected edge computing node pre-assigned to the subtask is too large, the current edge computing node is recommended and pushed, so as to realize the collaborative processing of tasks among multiple edge computing nodes.
[0045] The above process ends here.
[0046] Embodiment 2 like Fig.10 As shown, accordingly, the present invention also proposes a data security protection system for an edge reliable elastic network, including: an edge node scanning module 10, an evaluation server network device module 20, and a subsequent processing module 30; The scanning edge node module 10 is used to install an edge device on the server network device, and the edge device scans all edge computing nodes of the server network device; The server network device evaluation module 20 is used for the edge device to obtain original data from each edge computing node of the server network device, obtain data log information of each edge computing node through the original data, and determine the security result of the server network device by evaluating the data log information; The subsequent processing module 30 is used to send an alarm message to the server network device if the safety result of the assessment is unsafe.
[0047] like Fig.11 As shown, on the other hand, this embodiment three is based on a data security protection method for an edge reliable elastic network provided in embodiment one of the invention, and also provides a computer storage medium 1140 (referred to as storage medium).
[0048] A schematic diagram of a computer storage medium structure framework provided in Embodiment 3 of the present invention includes: A memory 1130, used for storing computer programs; The communication interface 1120 is used to connect the memory 1130 to the processor 1110; The processor 1110 is used to execute a computer program to implement a data security protection method for an edge reliable elastic network involved in Example 1 disclosed in combination with any of the above-mentioned implementation modes.
[0049] In summary, the data security protection method and system of an edge reliable elastic network proposed in the example of the present invention scans edge nodes by installing edge devices on server network devices, and analyzes data log information through the original data of each edge node, and the data log information includes task access request information, task response information and performance information; the daily task description feature value Fn of the edge node is calculated according to the task access time of the task access request information, the task response time of the task response information and the bandwidth utilization rate of the performance information, and the task response efficiency status of the edge node on the nth day can be obtained; Furthermore, firstly, by collecting the data of the daily task description eigenvalue Fn of the current edge computing node, and performing eigenvector clustering of the Laplace matrix according to the data of the daily task description eigenvalue Fn within a certain period of time, a task response efficiency label is obtained. The task response efficiency label includes four levels, namely, a high-efficiency task response efficiency label, a medium-efficiency task response efficiency label, a low-efficiency task response efficiency label, and a severely low-efficiency task response efficiency label; the assessment security result of the current edge computing node is judged according to the task response efficiency label; Furthermore, through the task response efficiency label finally obtained by the daily task description characteristic value Fn of the current edge computing node mentioned above, the daily task description characteristic values Fn of multiple edge computing nodes are collected, and the daily task description characteristic values Fn within a specific time are clustered to obtain the task response efficiency label of each edge node, and the evaluation safety result of each edge node is determined.
[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. A person skilled in the art may modify the technical solutions described in the above embodiments, or replace part or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A data security protection method for edge reliable elastic network, It is characterized in that The steps include: Installing an edge device on the server network device, wherein the edge device scans all edge computing nodes of the server network device; The edge device obtains original data from each edge computing node of the server network device, obtains data log information of each edge computing node through the original data, and determines the security result of the server network device by evaluating the data log information; if the evaluated security result is unsafe, an alarm message is sent to the server network device and an emergency response operation of enhanced protection processing is implemented.
2. According to the data security protection method of the edge reliable elastic network described in claim 1, It is characterized in that The edge device obtains original data from each edge computing node of the server network device, obtains data log information of each edge computing node through the original data, and determines the security result of the server network device by evaluating the data log information. The specific operation steps are as follows: Collect data from each edge node through the edge device to obtain raw data; analyze the raw data to obtain data log information of each edge node, the data log information includes: task access request information, task response information and performance information; Obtain the task description feature value Fn of the edge node corresponding to each task according to the task access request information of the current edge computing node and each task response information and performance information; Collect the data of the daily task description eigenvalue Fn of the current edge computing node, perform Laplace matrix eigenvector clustering on the daily task description eigenvalue Fn corresponding to the current edge computing node within the preset time period, realize clustering output of the daily task description eigenvalue of the current edge computing node, and finally output the task response efficiency label of the daily task of the current edge computing node; Collect the data of daily task description eigenvalues Fn of multiple edge computing nodes, perform eigenvector clustering of the Laplace matrix on the task description eigenvalues Fn of the current multiple edge computing nodes at a specific time node day, realize clustering output of the task description eigenvalues of the current different edge nodes, and finally output the task response efficiency labels of multiple edge computing nodes to realize the evaluation output of multiple edge computing nodes.
3. According to the data security protection method of the edge reliable elastic network described in claim 2, It is characterized in that The task description characteristic value Fn of the edge node corresponding to each task is obtained, including: calculating the daily task description characteristic value Fn of the nth day, the daily task volume of the integrated edge node, the response time and access time of each task, the average response time and average access time of the daily task, and the edge node bandwidth utilization at the current moment; the formula is as follows: ; Where, Fn: daily task description feature value of day n; h: total number of tasks processed daily; i: the i-th task running on that day; Zi: task response time of the i-th task; Ki: task access time of the i-th task; D: average task response time of daily tasks; C: Average task access time of daily tasks; b: bandwidth utilization of edge nodes at the current moment; α, β, γ: weight coefficients, used to adjust the impact of various parameters on the eigenvalues of task description.
4. According to the data security protection method of the edge reliable elastic network described in claim 3, It is characterized in that Collect the data of the daily task description eigenvalue Fn of the current edge computing node, perform Laplace matrix eigenvector clustering on the daily task description eigenvalue Fn corresponding to the current edge computing node within the preset time period, realize clustering output of the daily task description eigenvalue of the current edge computing node, and finally output the task response efficiency label of the daily task of the current edge computing node; the specific operation steps are as follows: Collect data on the daily task description feature value Fn; Calculate the similarity matrix W between the data points of the current edge computing node, use the Gaussian kernel function to calculate, and construct the similarity matrix; The Gaussian kernel function calculation formula is as follows: ; Where σ is the parameter of the Gaussian kernel; when calculating the similarity matrix W between two data points, Fi and Fj represent any two data points; where Fi is the daily task description eigenvalue of the current edge computing node on the i-th day; Fj is the daily task description eigenvalue of the current edge computing node on the j-th day; Constructing a Laplace matrix L using the similarity matrix W; First, calculate the degree matrix D: The Laplace matrix L is calculated from the above degree matrix D: L = D − W; Calculate the eigenvectors corresponding to the first k smallest eigenvalues of the Laplace matrix L to form the matrix U; Taking the rows of the eigenvector matrix U as new data points, clustering the new data points using K-means, and outputting the clustering results of the current new data points; The clustering results of the new data points are mapped to the original data, and the clustering results are output to assign a clustering label to the daily task description feature value Fn of each current edge computing node.
5. According to the data security protection method of the edge reliable elastic network described in claim 4, It is characterized in that Collect the data of daily task description eigenvalues Fn of multiple edge computing nodes, perform Laplace matrix eigenvector clustering on the task description eigenvalues Fn of the current multiple edge computing nodes at a specific time node day, realize clustering output of task description eigenvalues of current different edge nodes, and finally output task response efficiency labels of multiple edge computing nodes to realize evaluation output of multiple edge computing nodes. The specific operation steps are as follows: Collect the data of daily task description feature values Fn of multiple edge computing nodes at a specific time node t; at the same time, collect the historical efficiency average FT of the efficiency values of all edge nodes at the same historical time t; Calculate the daily task description feature value Fn' corresponding to the current edge computing node at a specific time node t; Fn'=Fn-historical efficiency average FT; Calculate the similarity matrix W' between the daily task description feature values Fn' corresponding to the current specific time node day of any two edge nodes, use the Gaussian kernel function to calculate, and construct the similarity matrix; The Gaussian kernel function calculation formula is as follows: ; in; Indicates the current The daily task description feature value corresponding to the edge node at the current specific time node day; It is represented as the daily task description feature value corresponding to the current p-th edge node at the current specific time node day; The similarity matrix W' between the daily task description feature values corresponding to the current specific time node day of any two edge nodes is calculated by the above formula; Constructing a Laplace matrix L' using the similarity matrix W'; First, calculate the degree matrix D'; ; The Laplace matrix L' calculated from the above degree matrix D' is: L'=D'−W'; Calculate the eigenvectors corresponding to the first k' smallest eigenvalues of the Laplace matrix L' to form the matrix U'; Taking the rows of the eigenvector matrix U' as new data points, clustering the new data points using K-means, and outputting the clustering results; The clustering results are mapped back to the original data, and the clustering results are output to assign a clustering label to the daily task description feature value Fn' of each current edge computing node; the output of the evaluation security result is confirmed by the clustering label obtained by each current edge computing node.
6. According to the data security protection method of the edge reliable elastic network described in claim 5, It is characterized in that After the clustering label obtained by the edge computing node confirms the output of the evaluation security result, it also includes: establishing evaluation lists respectively through the task response efficiency label of the daily task of the current edge computing node and the task response efficiency labels of the multiple edge computing nodes to obtain a first evaluation list and a second evaluation list; The first evaluation list includes a first owning edge node set, a second owning edge node set, a third owning edge node set, and a fourth owning edge node set; The second evaluation list includes a first target edge node set, a second target edge node set, a third target edge node set, and a fourth target edge node set.
7. According to claim 6, a data security protection method for edge reliable elastic network, It is characterized in that After the clustering label obtained by the edge computing node confirms the output of the assessment security result, it also includes receiving the execution instruction of the distribution task sent by the current server network device, splitting the current distribution task into multiple subtasks, and screening according to the combination of the subtask node and all edge computing nodes, outputting the optimal multiple edge computing nodes to allocate multiple subtasks. The specific operation steps are as follows: S251: receiving an execution instruction of the distribution task sent by the current server, detecting the time and date of the current distribution task, and determining a second evaluation list corresponding to the time and date of the current distribution task; Detecting the number N of current subtasks, and determining whether the number N of current subtasks is less than the number of edge computing nodes in the first target edge node set in the second evaluation list; If so, randomly assigning the same number of edge computing nodes in the first target edge node set to the multiple subtasks in a first priority manner; Or execute S252; S252: match multiple edge computing nodes in the first target edge node set in the second evaluation list with the first belonging edge node set in the first evaluation list to solve the edge computing nodes with the same IP address, and obtain a number of identical edge computing nodes; cluster the several identical edge computing nodes to form a new data set; assign the multiple edge computing nodes in the new data set to the multiple subtasks; S253: If not, assign all edge computing nodes in the first target edge node set to multiple subtasks; if all edge computing nodes in the first target edge node set are still insufficient, assign all edge computing nodes in the second target edge node set, the third target edge node set, and the fourth target edge node set to multiple subtasks in order until all N subtasks are assigned; S254: After the above preset logic steps, monitor and calculate in real time the task load score of the edge computing node corresponding to the current subtask that has been screened; S255: Construct a third evaluation list using the task load scores of multiple edge computing nodes corresponding to the above preset logical steps; sort the third evaluation list from high to low according to the task load scores of the multiple edge computing nodes; form a new data set based on the corresponding multiple edge computing nodes and push it to the cloud server.
8. According to the data security protection method of the edge reliable elastic network described in claim 6, It is characterized in that Wherein, during the execution of S255: a third evaluation list is constructed using the task load scores of the plurality of edge computing nodes corresponding to the above-mentioned preset logic steps; The preset logical steps include any one of steps S251 or S252 or S253; when sequential screening is performed during the execution of step S255, the task load scores of the edge computing nodes corresponding to the multiple selected edge computing nodes determined in steps S251 or S252 or S253 are formed into a third evaluation list in which the task load scores are sorted from high to low by numerical value.
9. A data security protection system for edge reliable elastic network, It is characterized in that include: Scan edge node module, evaluate server network equipment module, and subsequent processing module; The scanning edge node module is used to install an edge device on the server network device, and the edge device scans all edge computing nodes of the server network device; The server network device evaluation module is used for the edge device to obtain original data from each edge computing node of the server network device, obtain data log information of each edge computing node through the original data, and determine the security result of the server network device by evaluating the data log information; The subsequent processing module is used to send an alarm message to the server network device and implement emergency response operations to enhance protection processing if the security result of the assessment is unsafe.
Citation Information
Patent Citations
Cloud network end cooperative defense method and system based on end-side edge computing
CN112769796A
Big data and edge computing-based data integration method and the storage medium
CN113515367A
Cloud edge collaboration method based on intelligent edge Internet of Things
CN114285855A
Power business evaluation method based on data fusion and edge calculation
CN117195292A
5g-based edge node scheduling method and apparatus, and medium and device
WO2022237670A1
