A method, apparatus and device for protecting edge computing node addresses

Abstract

A method, device and equipment for protecting edge computing node address, the method comprising: step one: after the optimal edge computing node processes the demand information, generating feedback information; step two: the optimal edge computing node randomly selects a pseudo address, takes the node corresponding to the pseudo address as a relay node, and sends the feedback information to the user through the relay node; step three: the optimal edge computing node judges whether the interaction frequency with the user exceeds the critical interaction frequency, if not, continues to use the pseudo address, if yes, changes the pseudo address and returns to execute the step two; using the pseudo address to directly interact with the user, when the pseudo address has a leakage risk, the user cannot immediately find the specific location of the optimal edge computing node which contacts with it, and the optimal edge computing node can update the pseudo address in time, so that the address of the optimal edge computing node can be well protected.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network information security control, in particular to a method, device and equipment for protecting edge computing node address. BACKGROUND

[0002] In recent years, with the rapid development of mobile Internet technology, the number of mobile terminals in the world is increasing day by day. According to statistics, the number of devices connected to the Internet in 2023 will be more than three times the global population. Correspondingly, people will have new demands for network performance. The traditional cloud computing mode needs to upload massive data to the cloud server. In the face of the increase of various data and the increase of computing pressure, although the cloud computing has strong computing power, it is also difficult to meet the requirements of users on data transmission rate, time delay and service quality. In order to reduce the delay of computing system, reduce data transmission bandwidth, relieve the pressure of cloud server, improve availability, and strengthen the protection of data security and privacy, it is proposed to transfer the cloud function to the edge of the network, and give the task to the computing resource close to the data source. Mobile edge computing (MEC) emerges as the times require.

[0003] MEC is not a substitute for cloud computing, but an extension of cloud computing. It is a distributed service architecture that migrates computing and storage resources from cloud platforms to network edges. It is completed by multiple edge computing nodes located between cloud servers and local devices. MEC as a new computing paradigm can collect a large number of idle computing capabilities and storage spaces of nodes distributed at the network edge. Because it is closer to local devices, it can perform computing-intensive and delay-critical tasks such as Internet of Vehicles and smart city. Cloud computing and edge computing are closely related. Compared with cloud computing, edge computing emphasizes more on the coordination of resources between edge devices, and is a technology for processing upstream data on the cloud or downstream data on the Internet of Things. Therefore, edge computing nodes carry a large amount of private information. However, in actual application, edge computing nodes are mostly located in untrusted environments and have poor mobility, and are vulnerable to various attacks. Therefore, how to protect edge computing nodes has become a problem to be solved in the Internet industry.

[0004] In order to protect the security of private information in the communication process, some static and passive security protection measures are currently taken. For example, a firewall, intrusion detection and defense, distributed denial of service (DDos) attack detection and protection, etc. These protection measures mostly detect in a priori manner, and need to identify the specific attack type through features and then make a response, so the attacker has enough time to analyze, and when he finds the loophole through long-term observation and launches a targeted attack, the current defense measures are extremely passive. The prior art combines IP randomization technology with deception technology, changes the IP address of a large number of deception nodes with the actual server, thereby prolonging the scanning time of the attacker, and reducing the probability of the actual server being identified and attacked by unauthorized users, but increasing the defense overhead of the communication network. SUMMARY

[0005] The purpose of the present application is to provide a method, device and equipment for protecting the address of an edge computing node.

[0006] The technical solution of the present application is as follows:

[0007] The present application provides a method for protecting the address of an edge computing node, comprising:

[0008] Step 1: After the optimal edge computing node processes the demand information, feedback information is generated;

[0009] Step 2: The optimal edge computing node randomly selects a pseudo address, takes the node corresponding to the pseudo address as a relay node, and sends the feedback information to the user through the relay node;

[0010] Step 3: The optimal edge computing node judges whether its interaction frequency with the user exceeds a critical interaction frequency, if not, continues to use the pseudo address, if so, changes the pseudo address and returns to execute Step 2.

[0011] The method for protecting the address of an edge computing node as described above, the method for obtaining the optimal edge computing node is:

[0012] The total loss degree of each edge computing node is calculated, and after sorting, the minimum total loss degree is obtained, and the edge computing node corresponding to the minimum total loss degree is the optimal edge computing node.

[0013] The formula of the total loss degree Q of the edge computing node is: loss

[0014] Q loss = u1Q x + u2Q y + u3Q mean ,​

[0015] Q x Q represents the lateral geographical loss from the edge computing node to the cloud server. y Q represents the vertical geographic loss from the edge computing node to the cloud server. mean Let u1 be the semantic loss degree and u1 be the Q value. x The weight value, u2 is Q y The weight value, u3 is Q mean The weight value.

[0016] As described above, in a method for protecting the address of an edge computing node, the location range of the pseudo-address is within a circle centered on the optimal edge computing node with radius R, where R = Q. max -Q minloss Q max For the maximum loss, Q minloss The total loss of the node at the optimal edge is calculated.

[0017] The selection range of the pseudo address is within the set of pseudo addresses with a privacy loss of (0, a], where the value of a ranges from (0, 1). The method for obtaining the pseudo address set is as follows: calculate the privacy loss of all addresses within the location circle, count the addresses whose privacy loss is within the range of (0, a], and form the pseudo address set.

[0018] The formula for the privacy loss degree W is as follows:

[0019]

[0020] Q ux Q is the lateral geographic loss from the address to the optimal edge node. uy σ is the vertical geographic loss from the address to the optimal edge computing node. ux For Q ux The bandwidth parameter, σ uy For Q uy The bandwidth parameters.

[0021] As described above, in a method for protecting the address of an edge computing node, the critical interaction frequency f d The formula for (t) is:

[0022]

[0023] μ1 is the discount factor for attack revenue, μ2 is the discount factor for update revenue, ε1 is the discount factor for attack cost, ε2 is the discount factor for update cost, ρ1 is the cost of a single attack, and ρ2 is the cost of a single update.

[0024] The present invention also provides a device for protecting the address of an edge computing node, comprising:

[0025] The feedback information module is used to generate feedback information.

[0026] The pseudo-address operation module is used to select a pseudo-address to transmit the feedback information;

[0027] The judgment module is used to determine whether the pseudo address needs to be changed.

[0028] The present invention also provides a device for protecting the address of an edge computing node, comprising a processor and a memory, wherein the processor implements the above-described method for protecting the address of an edge computing node when executing a computer program stored in the memory.

[0029] The present invention also provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the above-described method for protecting the address of an edge computing node.

[0030] The beneficial effects of this invention are as follows:

[0031] This invention provides a method for protecting the address of an edge computing node. When a cloud server sends a request, it can select a suitable edge computing node, i.e., the optimal edge computing node, based on the request's preference for semantic privacy and location privacy. Then, it randomly selects a pseudo address from the pseudo address set of the optimal edge computing node and uses the node corresponding to the pseudo address as a relay node. Through the relay node, it enables brief communication between the optimal edge computing node and the user. This effectively prevents the user from finding the location of the optimal edge computing node during information transmission, thus achieving the function of protecting the address of the optimal edge computing node.

[0032] This invention provides a method for protecting the address of an edge computing node. When the communication between the optimal edge computing node and the user exceeds a critical interaction frequency, in order to prevent the user from attacking the optimal edge computing node after recognizing the fake address, the optimal edge computing node will immediately discard the original fake address and replace it with a new fake address to continue indirect communication with the user, which can further enhance the function of protecting the address of the optimal edge computing node. Attached Figure Description

[0033] The solutions and advantages of this application will become clear to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0034] In the attached diagram:

[0035] Figure 1 This is a flowchart illustrating the protection method in the embodiment;

[0036] Figure 2 This is a schematic diagram of the protective device in the embodiment;

[0037] Figure 3 This is a schematic diagram of the protective device in the embodiment. Detailed Implementation

[0038] Exemplary embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings.

[0039] Example

[0040] This embodiment provides a method for protecting the addresses of edge computing nodes. See [link to relevant documentation] Figure 1 ,include:

[0041] Step 1: After processing the demand information, the optimal edge computing node generates feedback information;

[0042] Step 2: The optimal edge computing node randomly selects a pseudo address, uses the node corresponding to the pseudo address as a relay node, and sends the feedback information to the user through the relay node;

[0043] Step 3: The optimal edge computing node determines whether its interaction frequency with the user exceeds the critical interaction frequency. If it does not exceed the critical interaction frequency, it continues to use the pseudo address. If it does exceed the critical interaction frequency, it changes the pseudo address and returns to execute Step 2.

[0044] Specifically:

[0045] 1. The user sends the request information to the cloud server.

[0046] Users send their needs to the cloud server according to their own wishes. In order to reduce their own computing workload and prevent system crashes due to over-processing, the cloud server will pass the needs information down to the edge computing nodes for processing.

[0047] 2. Select the optimal edge computing node

[0048] The information uploaded by users to the cloud server includes geographic and semantic attributes, which correspond to the geographic and semantic privacy of user information, respectively. Different edge computing nodes have different sensitivities and preferences for the requested information, resulting in different processing outcomes. Therefore, it is necessary to find the most suitable edge computing node (optimal edge computing node) to process the requested information, so that the feedback obtained can meet the user's needs.

[0049] To obtain the optimal edge computing nodes, the applicant of this invention sends the demand information to all edge computing nodes through a cloud server and calculates the total loss Q for each edge computing node. loss , all Qloss The minimum total loss Q is obtained by sorting the values ​​from smallest to largest. minloss Q minloss The corresponding edge computing node is the optimal edge computing node.

[0050] When a cloud server sends demand information to different edge computing nodes, to prevent the demand information from being stolen during transmission, the cloud server adds a certain amount of redundant data, which is the semantic loss Q. mean Q mean This can be expressed as the ratio of twice the redundant data to the required information after adding the redundant data, with a value range of [0,1]. When the redundant data accounts for more than half, it is considered invalid data. The geographical distance from the cloud server to each edge computing node is regarded as the geographical location loss of the edge computing node, including the lateral loss Q. x and longitudinal loss Q y Lateral loss Q x and longitudinal loss Q y All calculations are performed using Euclidean distance, and Q x and Q y All are not less than 0, thus obtaining the Q of the edge computing node. loss =u1Q x +u2Q y +u3Q mean u1 is Q x The weight value, u2 is Q y The weight value, u3 is Q mean The weight values ​​of u1, u2 and u3 are all in the range of [0,1].

[0051] To obtain the optimal edge computing node, the applicant of this invention uses the Analytic Hierarchy Process (AHP) for analysis and computation. AHP is a decision-making method applicable to multiple-option choices in complex systems. It obtains the best alternative by considering the relative priorities of standard or alternative solutions. AHP decomposes a problem into individual parts in a hierarchical form, and then, through pairwise comparisons, the decision-maker can understand the role of the considered factors in the hierarchical structure. This provides the decision-maker with a simple and highly flexible model and an easy-to-apply auxiliary decision-making method.

[0052] The specific process of the Analytic Hierarchy Process (AHP) is as follows:

[0053] Step 1: Construct a structural model with a target layer of minimum loss, criterion layers of geographic and semantic loss, and a solution layer. Because Q x Q y and Q mean Since the magnitudes are different, these three attributes must first be normalized.

[0054] Step 2: Construct a pairwise factor comparison judgment matrix based on Table 1. The corresponding relevant factor indicators Q... x Q y and Q mean There is a dominant relationship. By comparing the advantages and disadvantages of each factor indicator in pairs, and selecting 1 to 5 and their reciprocals as the scale for judging the relative advantages and disadvantages of the factors according to the user's privacy protection preferences, a judgment matrix C is established.

[0055] Table 1 Scale Determination Table

[0056]

[0057]

[0058] Step 3: Normalize the judgment matrix column-wise, sum the rows, and divide by n, where n is the index number (the number of rows in the judgment matrix), to obtain Q. x Q y and Q mean The weight value for the minimum loss.

[0059] Step 4: Perform a consistency check. Where λ max To determine the largest eigenvalue of matrix C, RI is the average random consistency index, with values ​​shown in Table 2. When CR ≤ 0.1, the consistency of the comparison matrix is ​​acceptable; when CR > 0.1, the comparison matrix needs appropriate adjustment.

[0060] Table 2 RI Value Table

[0061]

[0062] To facilitate understanding by those skilled in the art on how to obtain the optimal edge computing node, the applicant of this invention provides an example as follows:

[0063] Taking five edge computing nodes as an example, a simulation analysis is performed. A coordinate system is established with the cloud server as the origin and an index. Geographical location is expressed in km, and semantic loss is expressed as a decimal between 0 and 1 (i.e., the percentage of loss). The geographic location of each edge computing node and the semantic loss of this information transmitted to each edge computing node are shown in Table 3.

[0064] Table 3 Edge Computing Node Information

[0065]

[0066]

[0067] Since the three indicators in the table have different orders of magnitude, they need to be normalized. This means transforming each element into its proportion (since the geographical loss considered represents distance, their absolute values ​​need to be taken), for example, transforming element (2, 2) into... The normalized results are shown in Table 4:

[0068] Table 4 Normalization Results

[0069]

[0070] If we directly assign weight values ​​to each attribute and simply add them together, we can obtain Q. loss The approach is too subjective, and the weighting values ​​are difficult to determine. Therefore, based on Table 5, a judgment matrix is ​​constructed by comparing the pairwise importance of the indicators, thus scientifically setting the weights. Users can dynamically construct the judgment matrix based on their preference for information privacy, as shown in Table 5.

[0071] Table 5 shows the judgment matrix based on preferences.

[0072]

[0073] Here, a32 = 1 / 2 means the x-axis is more important than the y-axis; a42 = 2 means semantic loss is more important than y-axis, i.e., semantic loss > x-axis loss > y-axis loss in importance. Since assignment is subjective, a43 = 1 / 3 represents y-axis loss being more important than semantic loss, which is unreasonable. Therefore, we need to... and Perform a consistency check (RI values ​​refer to Table 3). If the consistency ratio CR = 0, the judgment matrix is ​​consistent and there will be no contradictions; if CR < 0.1, the error is acceptable and the consistency check can be passed; if CR ≥ 0.1, the judgment matrix needs to be adjusted until CR < 0.1. The adjusted judgment matrix is ​​shown in Table 6, where CR = 0.0092 < 0.1, which meets the requirements.

[0074] Table 6 Adjusted Judgment Matrix

[0075]

[0076] The judgment matrix that passed the consistency test is normalized column-wise. Each row is summed, and the sum is divided by n. The resulting column vector is the weight vector, as shown below:

[0077]

[0078] Substitute the obtained weights into the total loss function Q. loss =0.478Q x+0.164Q y +0.539Q mean The Q values ​​of each edge computing node are obtained. loss The sorting results are shown in Table 7, where the Q of edge computing node number 2 is... loss The minimum value is found, therefore the coordinates corresponding to edge computing node number 2 are the optimal edge computing node.

[0079] Table 7 Q of each edge computing node loss Ascending sort results

[0080]

[0081] 3. Selecting a pseudo address

[0082] Because edge computing nodes carry a large amount of information and have poor mobility, direct interaction with users could lead to vulnerabilities that users could exploit for deception attacks. To prevent users from stealing the addresses of edge computing nodes, their relative independence should be maintained as much as possible.

[0083] To address this issue, the applicant of this invention utilizes the optimal edge computing node to randomly select a pseudo address, uses the node corresponding to the pseudo address as a relay node, and sends feedback information to the user through the relay node. In this way, even if the user finds a vulnerability and attempts to steal the address, the address obtained is the pseudo address (the address of the relay node), not the address of the optimal edge computing node, thus protecting the edge computing node's address information from being stolen.

[0084] The location range of the pseudo-address is within a circle centered at the optimal edge computation node, with radius R, where R = Q. max -Q minloss Q max To determine the maximum loss, a value of 2.5 is chosen for Q. minloss The total loss of the node is calculated for the optimal edge.

[0085] In addition, the selection range of pseudo addresses is within the pseudo address set with a privacy loss of (0, a], where the value of a ranges from (0, 1). The pseudo address set is obtained by calculating the privacy loss of all addresses within the location circle, and counting the addresses with privacy loss within the range of (0, a] to form the pseudo address set.

[0086] In order to ensure both service efficiency and privacy protection, the applicant of this invention prefers a = 0.5, that is, the value range of privacy loss is (0, 0.5).

[0087] Since edge computing nodes do not include redundant data when sending feedback information to relay nodes, relay nodes do not consider semantic loss; and the size of redundant data is much larger than the data loss during data transmission, so the loss during data transmission can be ignored; furthermore, pseudo-addresses are relay nodes derived from edge computing nodes, so the loss between edge computing nodes and relay nodes is not considered. Therefore, the privacy loss W of addresses in the pseudo-address set can be expressed as W = (u x (u) y ), u x and u y These represent the horizontal and vertical leakage rates, respectively, and can be represented using a Gaussian kernel function, i.e. Q ux Q is the lateral geographic loss value calculated from the address to the optimal edge node. uy σ is the vertical geographic loss degree calculated from the address to the optimal edge node. ux For Q ux The bandwidth parameter, σ uy For Q uy The bandwidth parameters.

[0088] 4. Determine the critical interaction frequency

[0089] To prevent users from discovering the vulnerability of the pseudo address during multiple indirect interactions with the optimal edge computing node, thereby launching a follow-up attack and finding the address of the optimal edge computing node, it is necessary to change the pseudo address.

[0090] In this embodiment, whenever feedback information is transmitted to the user, the optimal edge computing node begins to determine whether its interaction frequency with the user exceeds the critical interaction frequency. If it does not exceed the critical interaction frequency, it continues to use the original pseudo address. If it does exceed the critical interaction frequency, the optimal edge computing node discards the original pseudo address, randomly selects a new pseudo address from the pseudo address set, uses the node corresponding to the new pseudo address as the new relay node, and transmits the feedback information to the user through the new relay node.

[0091] In this embodiment, the applicant of the present invention obtains the critical interaction frequency through game theory. μ1 is the discount factor for attack gains, μ2 is the discount factor for update gains, ε1 is the discount factor for attack costs, ρ1 is the cost of a single attack, and ρ2 is the cost of a single update. The optimal edge computing node uses f... d The (t) method is used to decide whether to change the pseudo address, which avoids wasting network resources due to excessive frequency and poor defense effect due to excessively slow frequency.

[0092] f dThe solution process for (t) involves a master-slave game where the user's access frequency and the pseudo-address update frequency form a master-slave dynamic, with the user as the leader (within period t). a (t) represents the user interaction frequency (since the attack frequency is difficult to observe, the interaction frequency can be used as an approximation of the attack frequency); f b (t) represents the pseudo-address update frequency; the attack cost can be expressed as ε1ρ1f a (t), the update cost is ε2ρ2f b (t), the return can be obtained by multiplying the frequency difference by their respective discount factors μ1 and μ2:

[0093] Net attack gain R a (t)=μ1(f a (t)-f b (t))f a (t)-ε1ρ1f a (t)+ε2ρ2f b (t);

[0094] Net income of defenders R b (t)=μ2(f b (t)-f a (t))f b (t)-ε2ρ2f b (t)+ε1ρ1f a (t);

[0095] The attacker acts as the leader, and the defender as the follower. d (t) about f b (t) When the partial derivative is equal to 0, the solution for f is... b (t) is the critical interaction frequency f. d (t), that is:

[0096]

[0097] f d (t) Substitute into R a (t) yields:

[0098]

[0099] The attacker needs to find f at this moment. a The optimal solution for (t), i.e., for R a (t) about f a Taking the partial derivative of (t) and setting it to 0, we get:

[0100]

[0101] Substituting formula (2) back into formula (1) yields:

[0102]

[0103] To better assist those skilled in the art in understanding this invention, this embodiment also provides the operating mechanism of users, cloud servers, optimal edge computing nodes, and relay nodes:

[0104] User A sends a request for an "apple vector graphic" to the cloud server for the first time. The cloud server selects the optimal edge computing node A to process the "apple vector graphic" and generate an "apple vector". Figure 1 Based on the feedback information, the optimal edge computing node A selects the node corresponding to the pseudo-address A as relay node A, and transmits the "apple vector" through relay node A. Figure 1 Send to user A.

[0105] User A sends a request for a "red apple vector image" to the cloud server for the second time. The cloud server directly sends the request to the optimal edge computing node A, which then processes and generates the "red apple vector image". Figure 1 The feedback information is sent to the "Red Apple Vector" relay node A. Figure 1 Send to user A.

[0106] User A sends a request for a "green apple vector image" to the cloud server for the third time. The cloud server directly sends the request to the optimal edge computing node A, which then processes and generates the "green apple vector image". Figure 1 The feedback information is sent to "Green Apple Vector" via relay node A. Figure 1 Send to user A.

[0107] ...

[0108] User A sends a request for a "Big Apple vector image" to the cloud server for the 31st time. The cloud server directly forwards the request to the optimal edge computing node A. Optimal edge computing node A determines that the number of interactions between it and user A within 1 minute exceeds 30. Optimal edge computing node A replaces the pseudo address A and selects the node corresponding to the pseudo address A' as relay node A'. The processed "Big Apple vector image" is then transmitted through relay node A'. Figure 1 The feedback message "" was sent to user A.

[0109] User A sends a request for a "banana vector image" to the cloud server for the 32nd time. The cloud server directly sends the request to the optimal edge computing node A, which processes and generates the "banana vector image". Figure 1 The feedback information is sent to the relay node A' and the "banana vector" is transmitted through the relay node A'. Figure 1 Send to user A.

[0110] User A sends a request for an "orange vector image" to the cloud server for the 33rd time. The cloud server directly sends the request to the optimal edge computing node A, which processes and generates the "orange vector image". Figure 1 The feedback information is sent to the relay node A' and the "orange vector" is transmitted. Figure 1 Send to user A.

[0111] User B sends a request for an "apple vector graphic" to the cloud server for the first time. The cloud server selects the optimal edge computing node B to process the "apple vector graphic" and generate an "apple vector". Figure 2 Based on the feedback information, the optimal edge computing node B selects the node corresponding to the pseudo-address B as relay node B, and transmits the "apple vector" through relay node B. Figure 2 Send to user B.

[0112] The present invention also provides a device for protecting the address of edge computing nodes, see [link to device]. Figure 2 ,include:

[0113] The feedback information module is used to generate feedback information.

[0114] The pseudo-address operation module is used to select a pseudo-address to transmit the feedback information;

[0115] The judgment module is used to determine whether the pseudo address needs to be changed.

[0116] This invention also provides a device for protecting the addresses of edge computing nodes, see [link to related document]. Figure 3 It includes a processor and a memory, wherein the processor implements the above-described method for protecting the address of an edge computing node when executing a computer program stored in the memory.

[0117] The present invention also provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the above-described method for protecting the address of an edge computing node.

[0118] In summary, the present invention provides a method for protecting the address of an edge computing node. When a cloud server sends a request, it can select a suitable edge computing node, i.e., the optimal edge computing node, based on the preference for semantic privacy and location privacy in the request. Then, it randomly selects a pseudo address from the pseudo address set of the optimal edge computing node and uses the node corresponding to the pseudo address as a relay node. The relay node enables brief communication between the optimal edge computing node and the user. This effectively prevents the user from finding the location of the optimal edge computing node during information transmission, thus achieving the function of protecting the address of the optimal edge computing node.

[0119] This invention provides a method for protecting the address of an edge computing node. When the communication between the optimal edge computing node and the user exceeds a critical interaction frequency, in order to prevent the user from attacking the optimal edge computing node after recognizing the fake address, the optimal edge computing node will immediately discard the original fake address and replace it with a new fake address to continue indirect communication with the user, which can further enhance the function of protecting the address of the optimal edge computing node.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for protecting the address of an edge computing node, characterized in that, include: Step 1: After processing the demand information, the optimal edge computing node generates feedback information; The optimal edge computing node is obtained by calculating the total loss of each edge computing node, sorting them, and then finding the edge computing node with the minimum total loss. The edge computing node with the minimum total loss is the optimal edge computing node. Q loss The formula is: Q loss =u 1 Q x + u 2 Q y +u 3 Q mean , Q x The horizontal geographical loss from edge computing nodes to cloud servers. Q y The vertical geographical loss from edge computing nodes to cloud servers. Q mean For semantic loss, u 1 for Q x The weight value, u 2 for Q y The weight value, u 3 for Q mean The weight value; Step 2: The optimal edge computing node randomly selects a pseudo-address, uses the node corresponding to the pseudo-address as a relay node, and sends the feedback information to the user through the relay node; the location range of the pseudo-address is centered on the optimal edge computing node. R The position is inside the circle with radius. R=Q max -Q minloss , Q max To maximize the loss, Q minloss The total loss of the node is calculated for the optimal edge; the selection range of the pseudo address is within the pseudo address set with a privacy loss of (0, a], where the value of a is in the range of (0, 1); the method for obtaining the pseudo address set is: calculate the privacy loss of all addresses within the location circle, and the addresses with privacy loss in the range of (0, a] form the pseudo address set; Privacy loss The formula is: ， Q ux The lateral geographic loss is calculated from the address to the optimal edge node. Q uy Calculate the vertical geographic loss from the address to the optimal edge node. σ ux for Q ux bandwidth parameters, σ uy for Q uy Bandwidth parameters; Step 3: The optimal edge computing node determines whether its interaction frequency with the user exceeds the critical interaction frequency. If it does not exceed the critical interaction frequency, it continues to use the pseudo address. If it does exceed the critical interaction frequency, it changes the pseudo address and returns to Step 2.

2. The method for protecting edge computing node addresses according to claim 1, characterized in that, Critical interaction frequency The formula is: ， As a discount factor for attack gains, To update the discount factor for returns, As a discount factor for the cost of attack, To update the discount factor for costs, Cost per attack Cost per update.

3. An apparatus for protecting the address of an edge computing node, used to implement the method for protecting the address of an edge computing node as described in claim 1, characterized in that, include: The feedback information module is used to generate feedback information; The pseudo-address operation module is used to select a pseudo-address to transmit the feedback information; The judgment module is used to determine whether the pseudo address needs to be changed.

4. A device for protecting the address of an edge computing node, characterized in that, The device includes a processor and a memory, wherein the processor executes a computer program stored in the memory to implement a method for protecting the address of an edge computing node as described in any one of claims 1-2.

5. A computer-readable storage medium, characterized in that, Used to store a computer program, wherein the computer program, when executed by a processor, implements a method for protecting the address of an edge computing node according to any one of claims 1-2.

Images