Heterogeneous cellular network task unloading method oriented to location privacy protection

Abstract

The invention discloses a heterogeneous cellular network task unloading method oriented to location privacy protection, and belongs to the technical field of communication. The invention provides a heterogeneous cellular network task unloading method oriented to location privacy protection, aiming at solving the problem that location privacy leakage is possibly caused when a mobile user unloads a task to a micro base station edge server to which the mobile user belongs in heterogeneous cellular network edge calculation. According to the method, the privacy leakage risk of task unloading is evaluated according to user position distribution, the number of unit tasks to be unloaded and the resource state of each edge server, the position correlation between users and target servers is weakened by using a relay forwarding mechanism of a macro base station, and a deep reinforcement learning method is adopted, so that the task unloading efficiency is improved. And the position privacy risk and the task unloading strategy of the user are dynamically learned and optimized, so that the position privacy protection of the user is effectively realized and the average unit task energy consumption is minimized under the constraint of average unit task tolerance time delay.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, specifically relating to a method for offloading tasks in heterogeneous cellular networks with location privacy protection. Background Technology

[0002] Mobile edge computing (MEC) reduces terminal load and shortens task completion latency by offloading computing power to the edge, allowing mobile users to offload computationally intensive or latency-sensitive tasks to nearby edge servers. However, traditional "nearby offload" task scheduling strategies can create a strong correlation between a mobile user's location and the edge server. This characteristic could be exploited by cyber attackers to infer the user's geographic location, leading to serious location privacy risks.

[0003] To address the aforementioned issues, this invention proposes a task offloading method for heterogeneous cellular networks oriented towards location privacy protection. This method utilizes the relay forwarding mechanism of macro base stations to weaken the location correlation between users and target servers. It employs a deep reinforcement learning model to dynamically learn and optimize user location privacy risks and task offloading strategies, thereby effectively protecting user location privacy and minimizing average unit task energy consumption under the constraint of average unit task tolerance latency. Summary of the Invention

[0004] This invention aims to solve the problems of the prior art. It proposes a method for offloading tasks in heterogeneous cellular networks with location privacy protection. The technical solution of this invention is as follows:

[0005] A method for offloading tasks in heterogeneous cellular networks with location privacy protection is disclosed. The system includes a macro base station, multiple micro base stations configured with edge servers, and several randomly distributed mobile users. o represents the macro base station, n (∈N) represents the micro base station and its corresponding edge server number, and m (∈M) represents the mobile user. The method includes the following steps:

[0006] 101. A deep reinforcement learning model is established using a dual-delay deep deterministic policy gradient algorithm, including a target server and a computational resource allocation policy generator, a state-policy evaluation module, an experience sample storage area, and a random sample set. The system time is defined as K equal-length time slots of length Δt. A four-element vector group [s] is established. k ,a k ,r k ,s k+1 ], where s k a k r k These represent the system state of slot k, the system target server and computing resource allocation strategy, and the system reward, s k+1For the system state in time slot k+1, initialize k=0;

[0007] 102. If k < K, let k = k + 1, from the system state s k In the process, obtain the real-time location of each user m in set M. Number of unit tasks to be uninstalled Q m And the available computing resources f of each edge server numbered n n The target server and computing resource allocation strategy generator generates the system target server and computing resource allocation strategy. in, If the target server and computing resource allocation strategy for user m in time slot k is determined, proceed to step 103; otherwise, proceed to step 106.

[0008] 103. Based on the system target server and computing resource allocation strategy a k and the real-time location of each user m Calculate the number of unloadable unit tasks for each user m in time slot k. Constructing k-slot task offloading decisions

[0009] 104. Based on the task unloading decision D k Perform task unloading in time slot k and calculate the system reward r for time slot k. k According to s k With a k The system state s of slot k+1 is calculated. k+1 and the quad vector group [s k ,a k ,r k ,s k+1 Add to the experience sample storage area;

[0010] 105. Randomly select Z samples from the experience sample storage area to form a random sample set, and import the random sample set into the target server, the computing resource allocation strategy generator and the state-policy evaluation module for training, and then jump to step 102.

[0011] 106. Algorithm ends.

[0012] Furthermore, step 101 specifically includes:

[0013] Establish the system quad vector group [s k ,a k ,r k ,s k+1 ], where the system state s of time slot k. k System target server and computing resource allocation strategy a k System rewards r kand the system state s of time slot k+1. k+1 As shown in formulas (1), (2), (3), and (4) respectively:

[0014]

[0015] In formulas (1) and (4), and Q represents the real-time location of user m in time slot k or k+1, respectively. m f represents the number of unit tasks to be uninstalled for user m. n This represents the available computing resources of the edge server numbered n; in formula (2), Let represent the target server and computing resource allocation strategy for user m in time slot k, and its calculation method is shown in formula (5); in formula (3), ω1, ω2, and ω3 represent weighting factors. This represents the average unit task latency for user m in slot k. Does it satisfy the condition of being less than or equal to the average unit task tolerance delay τ? m Constraints, if satisfied, otherwise, Wherein, the average unit task latency for user m in time slot k. The calculation method is shown in formula (6). Let m represent the normalized average system energy consumption per unit task for user m in time slot k. Its calculation method is shown in formula (7). The location privacy leakage risk of user m in time slot k is represented by the formula (8); This indicates the risk of privacy breaches related to the location of user m in time slot k. Does it meet the location privacy risk threshold P that is less than or equal to? th Constraints, if satisfied otherwise, w represents the penalty value for failing to meet constraints regarding the risk of user location privacy leakage;

[0016]

[0017]

[0018] In formula (5), n * Indicates the target server number selected by user m for unloading in time slot k, and n * ∈N, This represents the computing resources allocated by the target server in time slot k to user m; in formula (6), This represents the number of unloadable unit tasks for user m in time slot k; in formula (7), Let ε represent the average unit task energy consumption of user m in time slot k, and its calculation method is shown in formula (9); in formula (8), ε is a constant. This indicates that user m in time slot k and user n are... * Distance between target servers, R o This refers to the coverage radius of the macro base station;

[0019]

[0020] In formula (9), This represents the transmission energy consumption of user m task offloading in time slot k. This represents the computational energy consumption of user m task offloading in time slot k.

[0021] Furthermore, in step 103, the number of unloadable unit tasks for each user m in time slot k is calculated. The method is as follows:

[0022] 1) Initialize a temporary set M′ = M, which can unload a unit of tasks.

[0023] 2) If Extract the first element m, and then allocate resources based on the target server and computing resources of user m. Obtain the target server number n * and the computing resources allocated to them Skip to step 3); otherwise, skip to step 8.

[0024] 3) Based on the real-time location of user m in time slot k Obtain the base station number of the network currently accessed by user m.

[0025] 4) If Uninstall mode This indicates that user m in time slot k is not within the coverage area of ​​any micro base station. In this case, user m will forward the task to the macro base station o, which is numbered n. * The micro base station, and delivered with the number n * If the target server has been completely uninstalled, proceed to step 7); otherwise, proceed to step 5.

[0026] 5) If n∈N, and n=n * Uninstall mode This indicates that user m in time slot k is located in slot n. * The coverage area of ​​the micro base station, at this time user m passes through the network numbered n * The micro base station will deliver the task numbered n * If the target server has completed uninstallation, proceed to step 7); otherwise, proceed to step 6.

[0027] 6) If n∈N, and n≠n * Uninstall mode This indicates that user m in time slot k is within the coverage area of ​​micro base station n. In this case, user m forwards the task to macro base station o via micro base station n, and then macro base station o forwards it to micro base station n. * The micro base station, and delivered with the number n * If the target server has been completely uninstalled, proceed to step 7); otherwise, proceed to step 7.

[0028] 7) Based on the task offloading mode of user m in time slot k Target server number n * and the computing resources allocated to them Calculate the number of unloadable unit tasks for user m in time slot k. Jump to step 2);

[0029] 8) Return the number of unloadable unit tasks for each user m in time slot k. The process is complete.

[0030] Furthermore, in step 7), the number of unloadable unit tasks for user m in time slot k. The calculation method is shown in formula (10):

[0031]

[0032] In formula (10), This represents the floor operation, C m Calculate the complexity of user m's task, where U is the unit task size. Let m represent the transmission rate between user m in time slot k and macro base station o. Its calculation method is shown in formula (11). This indicates that macro base station o and number n are... * The transmission rate between micro base stations is calculated using formula (12). This indicates that user m in time slot k and user n are... * The transmission rate between micro base stations is calculated as shown in formula (13), b n,o Let n represent the transmission rate between the micro base station numbered n and the macro base station o. The calculation method is shown in formula (14). Let m represent the transmission rate between user m in time slot k and micro base station n, and its calculation method is shown in formula (15):

[0033]

[0034] In formula (11), W represents the communication bandwidth, pm This represents the transmit power of user m. σ represents the channel gain between user m in time slot k and macro base station o. 2 This represents the power of additive white Gaussian noise; in formula (12), p o This indicates the transmit power of macro base station o. This indicates that macro base station o and number n are... * Channel gain between micro base stations; in formula (13), This indicates that user m in time slot k and user n are... * The channel gain between micro base stations; in formula (14), p n h represents the transmit power of the micro base station numbered n. n,o Let n represent the channel gain between the micro base station numbered n and the macro base station o; in formula (15), This represents the channel gain between user m in time slot k and micro base station numbered n.

[0035] Furthermore, in step 104, the system reward r for the k-slot is obtained. k and the k+1 time slot system state s k+1 The method is as follows:

[0036] 9) Decision D based on task offloading in time slot k k Calculate the task completion status of each user m in the k-timeslot set M. Transmission power consumption Calculate energy consumption Location privacy leakage risk And calculate the system reward r according to formula (3). k ;

[0037] 10) Update the real-time position of each user m in time slot k+1 based on the real-time motion status of each user m in time slot set M. Based on the number of unloadable unit tasks for each user m in the k-slot set M. Update the number of unloaded unit tasks for each user m in time slot k+1.

[0038] 11) Return the system reward r for slot k. k and the system state s of time slot k+1 k+1 The steps are now complete.

[0039] Furthermore, in step 9), the transmission energy consumption of user m in time slot k And calculate energy consumption The calculation methods are shown in formulas (16) and (17), respectively:

[0040]

[0041] In formula (17), κ represents the effective switching capacitor coefficient.

[0042] Furthermore, in step 10), the real-time position of each user m in time slot k+1 is updated according to the real-time motion state of each user m in the k-time slot set M. The method is shown in formula (18):

[0043]

[0044] In formula (18), and These represent the moving speed and direction of user m in time slot k.

[0045] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the location-privacy-oriented heterogeneous cellular network task offloading method as described in any one of claims 1 to 7.

[0046] A non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a location-privacy-oriented heterogeneous cellular network task offloading method as described in any one of the claims.

[0047] The advantages and beneficial effects of this invention are as follows:

[0048] This invention discloses a task offloading method for heterogeneous cellular networks with location privacy protection. Existing research on task offloading in heterogeneous cellular networks mainly focuses on optimizing the latency and energy consumption of task offloading, thus typically employing a "nearest offloading" strategy, where user tasks are preferentially offloaded to the physically nearest edge server, without fully considering the potential location privacy leakage risks during task offloading. Due to the small coverage area of ​​micro base stations in heterogeneous cellular networks, there is a strong correlation between user location and micro base station edge servers. The traditional "nearest offloading" method allows network attackers to quickly infer the user's geographical location based on the task offloading destination, leading to user location privacy leakage. To address this problem, this invention assesses the privacy leakage risk during the offloading process based on user location distribution, the number of unloaded tasks, and the resource status of each edge server. It utilizes a macro base station relay forwarding mechanism to weaken the location correlation between the user and the target server. A deep reinforcement learning method is used to dynamically learn and optimize the user's location privacy risk and task offloading strategy, thereby effectively protecting user location privacy and minimizing average unit task energy consumption under the constraint of average unit task tolerance latency. Attached Figure Description

[0049] Figure 1This is a flowchart of a heterogeneous cellular network task offloading method for location privacy protection, provided by a preferred embodiment of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0051] The technical solution of the present invention to solve the above-mentioned technical problems is:

[0052] The concepts and models involved in this invention are as follows:

[0053] 1. System Model:

[0054] The heterogeneous cellular network system model considered in this invention consists of a macro base station, multiple micro base stations configured with edge servers, and several randomly distributed mobile users. o represents the macro base station, n (∈N) represents the micro base station and its corresponding edge server number, and m (∈M) represents the mobile user. The macro base station, micro base stations, and users all communicate using OFDMA technology. Communication interference is not considered, and the macro base station and each micro base station operate in different frequency bands. The system's operating time is divided into K discrete time slots of equal length with a length of Δt. It is assumed that the system state remains unchanged within any time slot k. The macro base station, acting as a central controller, collects mobile device information (such as location, information related to offloaded tasks, channel status, etc.) and then makes task offloading decisions based on this information to improve overall system performance and resource utilization efficiency.

[0055] 2. Other symbols used in this invention are explained as follows:

[0056] k: Current time slot number

[0057] Δt: Time slot length

[0058] s k System status

[0059] a k System target server and computing resource allocation strategy

[0060] r k Reward function

[0061] Q m Number of remaining unloaded unit tasks

[0062] f n The available computing resources of the edge server numbered n k-slot user m's task offloading mode

[0063] Number of unloadable unit tasks for user m in time slot k Task completion status of user m in time slot k Normalized system average energy consumption per unit task for user m in k time slots Risk of location privacy breach for user m in k-slot. The base station number of the access network for user m in time slot k. The computing resources τ allocated to user m in time slot k m Average unit task tolerance latency for user m Average unit task latency up to k time slots for user m The average system energy consumption per unit task for user m in time slot k

[0064] R o Coverage radius of macro base stations

[0065] Transmission energy consumption of k-slot user m-task offloading Computational energy consumption for k-slot user m-task offloading

[0066] U: Unit task size

[0067] C m User task computational complexity

[0068] W: Channel bandwidth

[0069] p m Transmit power of user m

[0070] p o Transmission power of macro base station o

[0071] p n The transmit power σ of the micro base station numbered n 2 Additive white Gaussian noise power

[0072] Channel gain between user m in time slot k and micro base station numbered n User m in time slot k and user number n * Channel gain h between micro base stations n,o Channel gain between micro base station n and macro base station o Macro base station o and numbered n * Channel gain between micro base stations Channel gain between k-slot user m and macro base station o

[0073] P th Location privacy leakage risk threshold

[0074] The technical solution of the present invention is described as follows:

[0075] 1. System quad vector group [s] k ,a k ,r k ,s k+1 ]

[0076] System state s in time slot k k System target server and computing resource allocation strategy a k System rewards r k and the system state s of time slot k+1. k+1 As shown in formulas (1), (2), (3), and (4) respectively:

[0077]

[0078]

[0079] In formulas (1) and (4), and Q represents the real-time location of user m in time slot k or k+1, respectively. m f represents the number of unit tasks to be uninstalled for user m. n This represents the available computing resources of the edge server numbered n; in formula (2), Let represent the target server and computing resource allocation strategy for user m in time slot k, and its calculation method is shown in formula (5); in formula (3), ω1, ω2, and ω3 represent weighting factors. This represents the average unit task latency for user m in slot k. Does it satisfy the condition of being less than or equal to the average unit task tolerance delay τ? m Constraints, if satisfied, otherwise, Wherein, the average unit task latency for user m in time slot k. The calculation method is shown in formula (6). Let m represent the normalized average system energy consumption per unit task for user m in time slot k. Its calculation method is shown in formula (7). The location privacy leakage risk of user m in time slot k is represented by the formula (8); This indicates the risk of privacy breaches related to the location of user m in time slot k. Does it meet the location privacy risk threshold P that is less than or equal to? th Constraints, if satisfied otherwise, w represents the penalty value for failing to meet constraints regarding the risk of user location privacy leakage;

[0080]

[0081] In formula (5), n* Indicates the target server number selected by user m for unloading in time slot k, and n * ∈N, This represents the computing resources allocated by the target server in time slot k to user m; in formula (6), This represents the number of unloadable unit tasks for user m in time slot k; in formula (7), Let ε represent the average unit task energy consumption of user m in time slot k, and its calculation method is shown in formula (9); in formula (8), ε is a constant. This indicates that user m in time slot k and user n are... * Distance between target servers, R o This refers to the coverage radius of the macro base station;

[0082]

[0083] In formula (9), This represents the transmission energy consumption of user m task offloading in time slot k. This represents the computational energy consumption of user m task offloading in time slot k;

[0084] 2. Number of unloadable unit tasks Calculation method

[0085] Number of unloadable unit tasks for user m in time slot k The calculation method is shown in formula (10):

[0086]

[0087] In formula (10), This represents the floor operation, C m Calculate the complexity of user m's task, where U is the unit task size. Let m represent the transmission rate between user m in time slot k and macro base station o. Its calculation method is shown in formula (11). This indicates that macro base station o and number n are... * The transmission rate between micro base stations is calculated using formula (12). This indicates that user m in time slot k and user n are... * The transmission rate between micro base stations is calculated as shown in formula (13), b n,o Let n represent the transmission rate between the micro base station numbered n and the macro base station o. The calculation method is shown in formula (14). Let m represent the transmission rate between user m in time slot k and micro base station n, and its calculation method is shown in formula (15):

[0088]

[0089]

[0090] In formula (11), W represents the communication bandwidth, p m This represents the transmit power of user m. σ represents the channel gain between user m in time slot k and macro base station o. 2 This represents the power of additive white Gaussian noise; in formula (12), p o This indicates the transmit power of macro base station o. This indicates that macro base station o and number n are... * Channel gain between micro base stations; in formula (13), This indicates that user m in time slot k and user n are... * The channel gain between micro base stations; in formula (14), p n h represents the transmit power of the micro base station numbered n. n,o Let n represent the channel gain between the micro base station numbered n and the macro base station o; in formula (15), This represents the channel gain between user m in time slot k and the micro base station numbered n.

[0091] 3. Calculate the user's transmission energy consumption And calculate energy consumption Method

[0092] Transmission power consumption per user in k-slot And calculate energy consumption The calculation methods are shown in formulas (16) and (17) respectively:

[0093]

[0094] In formula (17), κ represents the effective switching capacitor coefficient;

[0095] 4. Method for updating the real-time location of each user m in time slot k+1

[0096] Real-time location of each user m in time slot k+1 The calculation method is shown in formula (18):

[0097]

[0098] In formula (18), and These represent the moving speed and direction of user m in time slot k, respectively.

[0099] 5. Sub-algorithm 1: Calculate the number of unit tasks that each user m can unload in time slot k.

[0100] Step 1: Initialize the temporary set M′ = M, which can unload a unit of tasks.

[0101] Step 2: If Extract the first element m, and then allocate resources based on the target server and computing resources of user m. Obtain the target server number n * and the computing resources allocated to them Skip to step 3; otherwise, skip to step 8.

[0102] Step 3: Based on the real-time location of user m in time slot k Obtain the base station number of the network currently accessed by user m.

[0103] Step 4: If Uninstall mode This indicates that user m in time slot k is not within the coverage area of ​​any micro base station. In this case, user m will forward the task to the macro base station o, which is numbered n. * The micro base station, and delivered with the number n * If the target server has been completely uninstalled, proceed to step 7; otherwise, proceed to step 5.

[0104] Step 5: If n∈N, and n=n * Uninstall mode This indicates that user m in time slot k is located in slot n. * The coverage area of ​​the micro base station, at this time user m passes through the network numbered n * The micro base station will deliver the task numbered n * If the target server has been completely uninstalled, proceed to step 7; otherwise, proceed to step 6.

[0105] Step 6: If n∈N, and n≠n * Uninstall mode This indicates that user m in time slot k is within the coverage area of ​​micro base station n. In this case, user m forwards the task to macro base station o via micro base station n, and then macro base station o forwards it to micro base station n. * The micro base station, and delivered with the number n * If the target server has been completely uninstalled, proceed to step 7; otherwise, proceed to step 7.

[0106] Step 7: Based on the task offloading mode of user m in time slot k. Target server number n * and the computing resources allocated to them The number of unloadable unit tasks for user m in time slot k is calculated using formula (10). Jump to the next step;

[0107] Step 8: Return the number of unloadable unit tasks for each user m in time slot k. The process is complete.

[0108] 6. Sub-algorithm 2: Obtaining the system reward r in k time slots k and the k+1 time slot system state s k+1

[0109] Step 1: Decision D based on the task offloading in slot k k Calculate the task completion status of each user m in the k-timeslot set M. Transmission power consumption Calculate energy consumption Location privacy leakage risk And calculate the system reward r according to formula (3). k ;

[0110] Step 2: Update the real-time position of each user m in time slot k+1 based on the real-time motion status of each user m in the k-time slot set M. Based on the number of unloadable unit tasks for each user m in the k-slot set M. Update the number of unloaded unit tasks for each user m in time slot k+1.

[0111] Step 3: Return the system reward r for slot k. k and the system state s of time slot k+1 k+1 The steps are now complete.

[0112] A method for offloading tasks in heterogeneous cellular networks with location privacy protection, the specific implementation method of which includes the following steps:

[0113] Step 1: The system includes a macro base station, multiple micro base stations configured with edge servers, and several randomly distributed mobile users. o represents the macro base station, n (∈N) represents the micro base station and its corresponding edge server number, and m (∈M) represents the mobile user. A deep reinforcement learning model is established using a dual-delay deep deterministic policy gradient algorithm, including a target server and a computational resource allocation policy generator, a state-policy evaluation module, an experience sample storage area, and a random sample set. The system time is defined as K equal-length time slots of length Δt. A four-element vector group [s] is established. k ,a k ,r k ,s k+1 ], where s k a k r k The system state, target server and computing resource allocation strategy, and system reward for time slot k (≤K) are respectively. k+1 For the system state in time slot k+1, initialize k=0;

[0114] Step 2: If k < K, let k = k + 1, and start from system state s k In the process, obtain the real-time location of each user m in set M. Number of unit tasks to be uninstalled Q m And the available computing resources f of each edge server numbered n n The target server and computing resource allocation strategy generator generates the system target server and computing resource allocation strategy. in, If the target server and computing resource allocation strategy for user m in time slot k is determined, proceed to step 3; otherwise, proceed to step 6.

[0115] Step 3: Based on the system target server and computing resource allocation strategy a k and the real-time location of each user m Sub-algorithm 1 is invoked to calculate the number of unloadable unit tasks for each user m in time slot k. Constructing k-slot task offloading decisions

[0116] Step 4: Based on the task unloading decision D k Perform task unloading for slot k, call sub-algorithm 2, and calculate the system reward r for slot k. k According to s k With a k The system state s of slot k+1 is calculated. k+1 and the quad vector group [s k ,a k ,r k ,s k+1 Add to the experience sample storage area;

[0117] Step 5: Randomly select Z samples from the experience sample storage area to form a random sample set, and import the random sample set into the target server, the computing resource allocation strategy generator and the state-policy evaluation module for training, and then jump to step 2.

[0118] Step 6: The algorithm ends.

[0119] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions.

[0120] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0121] The above embodiments should be understood as illustrative only and not as limiting the scope of protection of the present invention. After reading the description of the present invention, those skilled in the art can make various alterations or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A method for task offloading in heterogeneous cellular networks with location privacy protection, the system comprising a macro base station, multiple micro base stations configured with edge servers, and several randomly distributed mobile users, where o is the macro base station, n (∈N) are the micro base station numbers and their corresponding edge server numbers, and m (∈M) are mobile users, characterized in that, Includes the following steps:

101. A deep reinforcement learning model is established using a dual-delay deep deterministic policy gradient algorithm, including a target server and a computational resource allocation policy generator, a state-policy evaluation module, an experience sample storage area, and a random sample set. The system time is defined as K equal-length time slots of length Δt. A four-element vector group [s] is established. k ,a k ,r k ,s k+1 ], where s k a k r k These represent the system state of slot k, the system target server and computing resource allocation strategy, and the system reward, s k+1 For the system state in time slot k+1, initialize k=0; 102. If k < K, let k = k + 1, from the system state s k In the process, obtain the real-time location of each user m in set M. Number of unit tasks to be uninstalled Q m And the available computing resources f of each edge server numbered n n The target server and computing resource allocation strategy generator generates the system target server and computing resource allocation strategy. in, If the target server and computing resource allocation strategy for user m in time slot k is determined, proceed to step 103; otherwise, proceed to step 106.

103. Based on the system target server and computing resource allocation strategy a k and the real-time location of each user m Calculate the number of unloadable unit tasks for each user m in time slot k. Constructing k-slot task offloading decisions 104. Based on the task unloading decision D k Perform task unloading in time slot k and calculate the system reward r for time slot k. k According to s k With a k The system state s of slot k+1 is calculated. k+1 and the quad vector group [s k ,a k ,r k ,s k+1 Add to the experience sample storage area; 105. Randomly select Z samples from the experience sample storage area to form a random sample set, and import the random sample set into the target server, the computing resource allocation strategy generator and the state-policy evaluation module for training, and then jump to step 102.

106. Algorithm ends.

2. The method for offloading tasks in heterogeneous cellular networks with location privacy protection according to claim 1, characterized in that, Step 101 specifically includes: Establish the system quad vector group [s k ,a k ,r k ,s k+1 ], where the system state s of time slot k. k System target server and computing resource allocation strategy a k System rewards r k and the system state s of time slot k+1. k+1 As shown in formulas (1), (2), (3), and (4) respectively: In formulas (1) and (4), and Q represents the real-time location of user m in time slot k or k+1, respectively. m f represents the number of unit tasks to be uninstalled for user m. n This represents the available computing resources of the edge server numbered n; in formula (2), Let represent the target server and computing resource allocation strategy for user m in time slot k, and its calculation method is shown in formula (5); in formula (3), ω1, ω2, and ω3 represent weighting factors. This represents the average unit task latency for user m in slot k. Does it satisfy the condition of being less than or equal to the average unit task tolerance delay τ? m Constraints, if satisfied, otherwise, Wherein, the average unit task latency for user m in time slot k. The calculation method is shown in formula (6). Let m represent the normalized average system energy consumption per unit task for user m in time slot k. Its calculation method is shown in formula (7). The location privacy leakage risk of user m in time slot k is represented by the formula (8); This indicates the risk of privacy breaches related to the location of user m in time slot k. Does it meet the location privacy risk threshold P that is less than or equal to? th Constraints, if satisfied otherwise, w represents the penalty value for failing to meet constraints regarding the risk of user location privacy leakage; In formula (5), n * Indicates the target server number selected by user m for unloading in time slot k, and n * ∈N, This represents the computing resources allocated by the target server in time slot k to user m; in formula (6), This represents the number of unloadable unit tasks for user m in time slot k; in formula (7), Let ε represent the average unit task energy consumption of user m in time slot k, and its calculation method is shown in formula (9); in formula (8), ε is a constant. This indicates that user m in time slot k and user n are... * Distance between target servers, R o This refers to the coverage radius of the macro base station; In formula (9), This represents the transmission energy consumption of user m task offloading in time slot k. This represents the computational energy consumption of user m task offloading in time slot k.

3. The method for offloading tasks in heterogeneous cellular networks with location privacy protection according to claim 1, characterized in that, In step 103, the number of unloadable unit tasks for each user m in time slot k is calculated. The method is as follows: 1) Initialize a temporary set M′ = M, which can unload a unit of tasks. 2) If Extract the first element m, and then allocate resources based on the target server and computing resources of user m. Obtain the target server number n * and the computing resources allocated to them Jump to step 3); Otherwise, proceed to step 8); 3) Based on the real-time location of user m in time slot k Obtain the base station number of the network currently accessed by user m. 4) If Uninstall mode This indicates that user m in time slot k is not within the coverage area of ​​any micro base station. In this case, user m will forward the task to the macro base station o, which is numbered n. * The micro base station, and delivered with the number n * If the target server has been completely uninstalled, proceed to step 7); otherwise, proceed to step 5. 5) If n∈N, and n=n * Uninstall mode This indicates that user m in time slot k is located in slot n. * The coverage area of ​​the micro base station, at this time user m passes through the network numbered n * The micro base station will deliver the task numbered n * If the target server has completed uninstallation, proceed to step 7); otherwise, proceed to step 6. 6) If n∈N, and n≠n * Uninstall mode This indicates that user m in time slot k is within the coverage area of ​​micro base station n. In this case, user m forwards the task to macro base station o via micro base station n, and then macro base station o forwards it to micro base station n. * The micro base station, and delivered with the number n * If the target server has been completely uninstalled, proceed to step 7); otherwise, proceed to step 7. 7) Based on the task offloading mode of user m in time slot k Target server number n * and the computing resources allocated to them Calculate the number of unloadable unit tasks for user m in time slot k. Jump to step 2); 8) Return the number of unloadable unit tasks for each user m in time slot k. The process is complete.

4. A method for offloading tasks in heterogeneous cellular networks with location privacy protection according to claim 3, characterized in that, In step 7), the number of unloadable unit tasks for user m in time slot k. The calculation method is shown in formula (10): In formula (10), This represents the floor operation, C m Calculate the complexity of user m's task, where U is the unit task size. Let m represent the transmission rate between user m in time slot k and macro base station o. Its calculation method is shown in formula (11). This indicates that macro base station o and number n are... * The transmission rate between micro base stations is calculated using formula (12). This indicates that user m in time slot k and user n are... * The transmission rate between micro base stations is calculated as shown in formula (13), b n,o Let n represent the transmission rate between the micro base station numbered n and the macro base station o. The calculation method is shown in formula (14). Let m represent the transmission rate between user m in time slot k and micro base station n, and its calculation method is shown in formula (15): In formula (11), W represents the communication bandwidth, p m This represents the transmit power of user m. σ represents the channel gain between user m in time slot k and macro base station o. 2 This represents the power of additive white Gaussian noise; in formula (12), p o This indicates the transmit power of macro base station o. This indicates that macro base station o and number n are... * Channel gain between micro base stations; in formula (13), This indicates that user m in time slot k and user n are... * The channel gain between micro base stations; in formula (14), p n h represents the transmit power of the micro base station numbered n. n,o Let n represent the channel gain between the micro base station numbered n and the macro base station o; in formula (15), This represents the channel gain between user m in time slot k and micro base station numbered n.

5. A method for offloading tasks in heterogeneous cellular networks with location privacy protection according to claim 1, characterized in that, In step 104, the k-slot system reward r is obtained. k and the k+1 time slot system state s k+1 The method is as follows: 9) Decision D based on task offloading in time slot k k Calculate the task completion status of each user m in the k-timeslot set M. Transmission power consumption Calculate energy consumption Location privacy leakage risk And calculate the system reward r according to formula (3). k ; 10) Update the real-time position of each user m in time slot k+1 based on the real-time motion status of each user m in time slot set M. Based on the number of unloadable unit tasks for each user m in the k-slot set M. Update the number of unloaded unit tasks for each user m in time slot k+1. 11) Return the system reward r for slot k. k and the system state s of time slot k+1 k+1 The steps are now complete.

6. A method for offloading tasks in heterogeneous cellular networks with location privacy protection according to claim 5, characterized in that, In step 9), the transmission power consumption of user m in time slot k. And calculate energy consumption The calculation methods are shown in formulas (16) and (17), respectively: In formula (17), κ represents the effective switching capacitor coefficient.

7. A method for offloading tasks in heterogeneous cellular networks with location privacy protection according to claim 4, characterized in that, In step 10), the real-time position of each user m in time slot k+1 is updated according to the real-time motion state of each user m in the k-time slot set M. The method is shown in formula (18): In formula (18), and These represent the moving speed and direction of user m in time slot k.

8. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the location-privacy-oriented heterogeneous cellular network task offloading method as described in any one of claims 1 to 7.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the location-privacy-oriented heterogeneous cellular network task offloading method as described in any one of claims 1 to 7.

Images