A method for maximizing safe computing rate in wireless power supply edge computing network

Abstract

The present invention discloses a method for maximizing the safe computing rate in a wireless power supply edge computing network, wherein the wireless power supply edge computing network includes a base station, a wireless device and an eavesdropper, and the base station transmits wireless radio frequency energy to each wireless device to charge the wireless device, and the base station transmits artificial noise to interfere with the eavesdropper's eavesdropping on each wireless device. The method for maximizing the safe computing rate in the wireless power supply edge computing network first constructs an objective function for maximizing the safe computing rate of the first channel gain, the second channel gain, the third channel gain, the preset unloading decision, the first unloading time ratio, the time ratio of the base station charging the wireless device and the power matrix of the base station transmitting artificial noise, and solves the unloading decision, the first unloading time ratio and the time ratio of the base station charging the wireless device in the objective function, and calculates the objective function to obtain the maximized safe computing rate.

Description

Technical Field

[0001] The present invention belongs to the technical field of secure computing rate, and specifically relates to a method for maximizing secure computing rate in a wireless power supply edge computing network. Background Art

[0002] With the rapid development of the Internet of Things, artificial intelligence, and edge computing technologies, edge computing networks have gradually become core technologies in many application scenarios. Wireless power supply, as an important component of edge computing networks, can effectively support devices to perform computing and data transmission without relying on traditional power supplies. However, although existing technologies have made some progress in wireless power supply and edge computing, there are still some key issues that limit computing performance and network security, especially in the process of maximizing secure computing rates.

[0003] Most of the existing wireless power supply edge computing networks focus on how to provide stable wireless power transmission, but there are relatively few studies on how to maximize the computing rate under the condition of wireless power supply; while improving the computing rate, the current wireless power supply edge computing networks often ignore security issues; data transmission during wireless power supply and computing is vulnerable to network attacks and signal interference, resulting in data leakage and tampering of computing results. In the existing technology, there is insufficient attention to security, lack of effective encryption technology and protection mechanism, resulting in the increase of computing rate often accompanied by increased security risks. Although the wireless power supply network in the existing technology can provide energy support for edge devices, there are still major deficiencies in how to dynamically adjust energy transmission and distribution according to actual computing tasks. Wireless energy management is mostly statically designed, and energy supply cannot be adjusted in real time according to changes in the computing load of the device, thus affecting the computing rate and device stability. Summary of the invention

[0004] The purpose of the present invention is to solve the problems raised in the background technology and to propose a method for maximizing the safe computing rate in a wireless power supply edge computing network.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] The present invention proposes a method for maximizing the safe computing rate in a wireless power supply edge computing network, comprising:

[0007] Step 1: The wireless power supply edge computing network includes a base station, There are wireless devices and an eavesdropper, and the base station transmits wireless radio frequency energy to each wireless device to charge the wireless device, and the base station transmits artificial noise to interfere with the eavesdropper's eavesdropping on each wireless device;

[0008] Step 2: Random Generation The first channel gain between the wireless device and the base station in the time slot is randomly generated The second channel gain between each wireless device and the eavesdropper in time slots is randomly generated The third channel gain between the base station and the eavesdropper for each time slot;

[0009] Step 3: First time slot Inside, randomly initialize the first policy network of reinforcement learning , and a second policy network with randomly initialized reinforcement learning ;

[0010] Step 4: Establish the corresponding time slot an objective function of maximizing a safe computing rate with respect to a first channel gain, a second channel gain, a third channel gain, a preset unloading decision, a first unloading time ratio of all wireless devices, a time ratio of a base station charging each wireless device, and a power matrix of artificial noise emitted by the base station;

[0011] Step 5: Assign each wireless device to a time slot The first channel gain is input to the first strategy network , get the time slot corresponding to each wireless device The unloading probability within the time slot corresponding to each wireless device The unloading probability in the preprocessing is solved to obtain the corresponding time slot Preset uninstallation decision within the system;

[0012] Step 6: Assign each wireless device to a time slot The first channel gain is input to the second strategy network , get the corresponding time slot Second Strategy Network Regarding the output of each wireless device, according to the corresponding time slot Second Strategy Network The corresponding time slot is obtained by calculating the output and offloading decision of each wireless device The second unloading duration ratio between the wireless devices in the wireless network;

[0013] Step 7: Solve to get the corresponding time slot The optimal time ratio for the internal base station to charge the wireless device, and according to the corresponding time slot The second unloading time ratio and the optimal time ratio of the base station charging the wireless device are solved to obtain the first unloading time ratio of all wireless devices;

[0014] Step 8: According to the corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving The internal unloading decision, the first unloading time ratio of all wireless devices and the optimal charging time ratio are calculated to obtain the corresponding time slot The final maximum safe computation rate within ;

[0015] Step 9: First strategy network and the second strategy network Update, use the updated strategy networks as the strategy networks for the next time slot, enter the next time slot, repeat steps 4 to 9, and then get The final maximum security calculation rate for time slots.

[0016] Preferably, the formula of the objective function of maximizing the safe computing rate is as follows:

[0017] ;

[0018] in,

[0019] ;

[0020] ;

[0021] ;

[0022] ;

[0023] ;

[0024] in, To maximize the objective function of safe calculation rate, represents the set of the first channel gain, the second channel gain and the third channel gain, is the set of preset uninstall decisions obtained, and contains Uninstallation decision, For the offloading decisions, and includes the offloading decisions of all wireless devices, For the The first uninstall decision offload decision of each wireless device, is a set of first offloading duration ratios of all wireless devices corresponding to all types of offloading decisions, For the A set of first unloading duration ratios of all wireless devices corresponding to the unloading decision, For the The first uninstall decision The first unloading duration ratio of wireless devices, is the set of the proportion of the time that the base station charges all wireless devices corresponding to all types of offloading decisions, For the The proportion of time that the base station charges all wireless devices corresponding to the offloading decision, and The value range is ( 0,1] , is the power matrix of artificial noise emitted by the base station, is the number of cycles required for each wireless communication device to process one bit of data, is the energy capture efficiency of wireless devices, is the power of the radio frequency energy transmitted by the base station, For the The channel gain from a wireless device to the base station, is the computational efficiency of each wireless device, is the bandwidth from the base station to each wireless device, Offload the communication overhead of tasks for each wireless device, For the The signal-to-noise ratio from a wireless device to the base station, For the The signal-to-noise ratio from a wireless device to an eavesdropper.

[0025] Preferably, the offloading probability includes a probability of staying on the local wireless device and a probability of offloading to the base station.

[0026] Preferably, preprocessing the offloading probability of each wireless device to obtain a preset offloading decision includes:

[0027] The offloading probability of each wireless device is normalized, and then the normalized offloading probabilities are processed in turn. Random sampling The sampling probability of all wireless devices is the normalized probability of remaining on the local wireless device or the normalized probability of being offloaded to the base station;

[0028] When the normalized probability of staying on the local wireless device is obtained by random sampling, the offloading decision of the corresponding wireless device is to stay on the local wireless device, and , when the normalized probability of offloading to the base station is obtained by random sampling, the offloading decision of the corresponding wireless device is to offload to the base station, and 1.

[0029] Preferably, the second policy network The output set of each wireless device is expressed as ,in Represents the second policy network About output of a wireless device.

[0030] Preferably, according to the second strategy network The second offloading duration ratio between the wireless devices is obtained by calculating the output and offloading decision of each wireless device, including:

[0031] Traverse each uninstall decision and uninstall the decision The offloading decision of each wireless device in the second strategy network is respectively The outputs of each wireless device are dot-multiplied one by one, and the result after dot-multiplication is normalized to obtain the second unloading duration ratio between each wireless device, which is expressed as ,in, For the A set of second unloading duration ratios of all wireless devices corresponding to the unloading decision, For the The first uninstall decision The second unloading duration ratio of each wireless device.

[0032] Preferably, for each offloading decision, in the time slot Within, the base station charges all wireless devices. After charging is completed, each wireless device unloads in turn according to the preset unloading decision in the remaining time, that is, the first unloading time ratio of each wireless device and the time ratio of the base station charging all wireless devices meet the following relationship:

[0033] ;

[0034] For each offloading decision, the total time for all wireless devices to offload according to the preset offloading decision is ;

[0035] Multiply the second unloading duration ratio between wireless devices by the total duration The unloading duration of each wireless device is obtained, and the calculation formula is as follows:

[0036] ;

[0037] Right now,

[0038] .

[0039] Preferably, the solution obtains the corresponding time slot The optimal time ratio for the internal base station to charge the wireless device, and according to the corresponding time slot The second unloading time ratio and the optimal time ratio of the base station charging the wireless device are solved to obtain the first unloading time ratio of all wireless devices, including:

[0040] For various uninstall decisions , when all wireless devices’ offloading decisions are to stay in the local wireless device, that is, all When it is 0, the objective function of maximizing the safe computing rate is only proportional to the time the base station charges the wireless device. related to and about is a monotonically increasing function, so the optimal ;

[0041] For various uninstall decisions , when at least one of all wireless devices decides to offload to the base station, that is, at least one When is 1, the objective function of maximizing the safe calculation rate is converted to :

[0042] ;

[0043] At this time, for each uninstall decision , the objective function Only is an unknown number, and the optimal Make is the maximum value, and then various unloading decisions are obtained The optimal time ratio for base stations to charge wireless devices ;

[0044] According to the formula , get various uninstall decisions The first unloading duration ratio of each wireless device .

[0045] Preferably, the time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving The internal unloading decision, the first unloading time ratio of all wireless devices and the optimal charging time ratio are calculated to obtain the corresponding time slot The final maximum safe computation rate within ,include:

[0046] According to the corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving Each uninstall decision within , the first uninstallation time ratio of all wireless devices And the optimal charging time ratio , calculate the objective function to get the corresponding time slot Internal and external uninstall decisions The corresponding maximum security calculation rate is selected from all the maximum security calculation rates as the corresponding time slot The final maximum safe calculation rate .

[0047] Preferably, the method for maximizing the safe computing rate in the wireless power supply edge computing network further includes: The parameters of the benchmark network are copied , and the benchmark network Updated every preset time slot, where the reference network is updated When the first strategy network corresponding to the time slot The parameters are copied to obtain the corresponding benchmark network ;

[0048] First Strategy Network Replace with the baseline network , and follow steps 5 to 8 to get the corresponding maximum safe calculation rate , and in the benchmark network When preprocessing the output offloading probabilities, the normalized baseline network The output offloading probabilities are randomly sampled to obtain the benchmark network A corresponding sampling probability;

[0049] pass and First Strategy Network and the second strategy network Update, and the first strategy network The updated expression is: ,in, is the updated first strategy network, First Strategy Network The learning rate, First Strategy Network The policy network function is First Strategy Network Output action, second policy network The updated expression is: ,in, is the updated second strategy network, Second Strategy Network The learning rate, Second Strategy Network The policy network function is Second Strategy Network Output action.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] The method for maximizing the safe computing rate in the wireless power supply edge computing network first constructs an objective function of maximizing the safe computing rate regarding the first channel gain, the second channel gain, the third channel gain, the preset unloading decision, the first unloading time ratio, the time ratio of the base station charging the wireless device, and the power matrix of the base station emitting artificial noise, and solves the unloading decision, the first unloading time ratio, and the time ratio of the base station charging the wireless device in the objective function, and calculates the objective function to obtain the maximized safe computing rate; at the same time, the method comprehensively considers multiple aspects such as wireless device charging, computing power, and safety, and maximizes the computing rate while ensuring safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a flow chart of a method for maximizing the safe computing rate in a wireless power supply edge computing network of the present invention;

[0053] Figure 2 This is a module block diagram of the wireless power supply edge computing network of the present invention;

[0054] Figure 3 A module block diagram of the method for maximizing the safe computing rate in the wireless power-supply edge computing network of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0056] like Figure 1-Figure 3 As shown, a method for maximizing the safe computing rate in a wireless power supply edge computing network is provided, including:

[0057] Step 1: The wireless power supply edge computing network includes a base station, There are wireless devices and an eavesdropper, and the base station transmits wireless radio frequency energy to each wireless device to charge the wireless device, and the base station transmits artificial noise to interfere with the eavesdropper's eavesdropping on each wireless device;

[0058] It should be noted that the base station is a multi-antenna base station, and each wireless device is a single-antenna wireless device. For example, if the number of wireless devices is 10, and an eavesdropper eavesdrops on each wireless device, the base station integrates an RF energy transmitter to transmit wireless RF energy to each wireless device to charge each wireless device. After each wireless device obtains electricity, it uses it for local calculations. The base station also transmits artificial noise to the eavesdropper to interfere with the eavesdropper's eavesdropping on each wireless device, and each wireless device communicates with the base station in a time division multiplexing manner. In this embodiment, the area where the wireless power supply edge computing network is located is a 20m×20m area, the base station is in the center of the area, and the positions of each wireless device and the eavesdropper are randomly generated.

[0059] Step 2: Random Generation The first channel gain between the wireless device and the base station in the time slot is randomly generated The second channel gain between each wireless device and the eavesdropper in time slots is randomly generated The third channel gain between the base station and the eavesdropper in time slots; .

[0060] Step 3: First time slot Inside, randomly initialize the first policy network of reinforcement learning , and a second policy network with randomly initialized reinforcement learning (The second strategy network Using a deterministic policy network).

[0061] Step 4: Establish the corresponding time slot an objective function of maximizing a safe computing rate with respect to a first channel gain, a second channel gain, a third channel gain, a preset unloading decision, a first unloading time ratio of all wireless devices, a time ratio of a base station charging each wireless device, and a power matrix of artificial noise emitted by the base station;

[0062] Among them, the formula of the objective function of maximizing the safe calculation rate is as follows:

[0063] ;

[0064] in,

[0065] ;

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] in, To maximize the objective function of safe calculation rate, represents the set of the first channel gain, the second channel gain and the third channel gain, is the set of preset uninstall decisions obtained, and contains Uninstallation decision, For the offloading decisions, and includes the offloading decisions of all wireless devices, For the The first uninstall decision offload decision of each wireless device, is a set of first offloading duration ratios of all wireless devices corresponding to all types of offloading decisions, For the The set of first unloading duration ratios of all wireless devices corresponding to the unloading decision ( includes the first unloading duration ratio of all wireless devices), For the The first uninstall decision The first unloading duration ratio of wireless devices, is the set of the proportion of the time that the base station charges all wireless devices corresponding to all types of offloading decisions, For the The proportion of time that the base station charges all wireless devices corresponding to the offloading decision, and The value range is ( 0,1] , is the power matrix of the artificial noise emitted by the base station, where the power matrix The power matrix can be obtained by transforming the unit vector and introducing auxiliary variables and auxiliary functions. The core idea is to convert the vector optimization into a scalar optimization method to obtain the value. The specific solution process belongs to the existing technology and is not described in detail. For example, the existing literature A. Wiesel, Y. Eldar, and A. Beck, "Maximum likelihood estimation inlinear models with a gaussian model matrix," IEEE Signal Processing Letters, vol. 13, no. 5, pp. 292–295, 2006. and the existing literature J. Qin and J. Liu, "Multi-Access Edge Offloading Based on PhysicalLayer Security in C-V2X System," in IEEE Transactions on Vehicular Technology, vol. 71, no. 7, pp. 6912-6923, 2022. describes the specific solution of the power matrix. The process of value, is the number of cycles required for each wireless communication device to process one bit of data, is the energy capture efficiency of wireless devices, is the power of the radio frequency energy transmitted by the base station, For the The channel gain from a wireless device to the base station, is the computational efficiency of each wireless device, is the bandwidth from the base station to each wireless device, Offload the communication overhead of tasks for each wireless device, For the The signal-to-noise ratio from a wireless device to the base station, For the The signal-to-noise ratio from a wireless device to an eavesdropper.

[0071] Step 5: Assign each wireless device to a time slot The first channel gain is input to the first strategy network , get the time slot corresponding to each wireless device The unloading probability within the time slot corresponding to each wireless device The unloading probability in the preprocessing is solved to obtain the corresponding time slot Preset uninstallation decision within the system;

[0072] It should be noted that the First Strategy Network According to the policy network function (i.e. ) Output the offloading probability of each wireless device; the offloading probability includes the probability of staying on the local wireless device and the probability of offloading to the base station;

[0073] The offloading probability of each wireless device is preprocessed to obtain a preset offloading decision, and the process is as follows:

[0074] Normalize the offloading probability of each wireless device (due to the first strategy network Two offloading probabilities are output for each wireless device, namely, the probability of remaining on the local wireless device and the probability of offloading to the base station. The two offloading probabilities of each wireless device are normalized so that the two offloading probabilities are mapped to the same spatial range. Then, the normalized offloading probabilities are processed in sequence. Random sampling The sampling probability of all wireless devices is a random sampling of the offloading probability of all wireless devices, that is, one of the two normalized offloading probabilities of each wireless device is selected, that is, after one sampling, a normalized offloading probability (i.e., sampling probability) of all wireless devices is obtained, and the sampling probability is the normalized probability of remaining on the local wireless device or the normalized probability of offloading to the base station;

[0075] When the normalized probability of staying on the local wireless device is obtained by random sampling, the offloading decision of the corresponding wireless device is to stay on the local wireless device, and (That is, when a wireless device is randomly sampled to obtain the normalized probability of staying on the local wireless device, the unloading decision of the wireless device is to stay on the local wireless device), when the normalized probability of unloading to the base station is obtained by random sampling, the unloading decision of the corresponding wireless device is to unload to the base station, and 1 (that is, when a wireless device is randomly sampled to obtain the normalized probability of offloading to the base station, the offloading decision of the wireless device is to offload to the base station).

[0076] Step 6: Assign each wireless device to a time slot The first channel gain is input to the second strategy network , get the corresponding time slot Second Strategy Network Regarding the output of each wireless device, according to the corresponding time slot Second Strategy Network The corresponding time slot is obtained by calculating the output and offloading decision of each wireless device The second unloading duration ratio between the wireless devices in the wireless network;

[0077] Among them, the second strategy network According to the policy network function (i.e. ) to output, the second strategy network The output set of each wireless device is expressed as ,in Represents the second policy network About Output of wireless devices;

[0078] Among them, according to the second strategy network The second offloading duration ratio between the wireless devices is obtained by calculating the output and offloading decision of each wireless device. The specific process is as follows:

[0079] Traverse each uninstall decision and uninstall the decision The offloading decision of each wireless device in the second strategy network is respectively The outputs of each wireless device are dot-multiplied one by one, and the result after dot-multiplication is normalized to obtain the second unloading duration ratio between each wireless device (and then obtain the second unloading duration ratio between each wireless device corresponding to each unloading decision), and is expressed as ,in, For the A set of second unloading duration ratios of all wireless devices corresponding to the unloading decision, For the The first uninstall decision The second unloading duration ratio of each wireless device.

[0080] Step 7: Solve to get the corresponding time slot The optimal time ratio for the internal base station to charge the wireless device, and according to the corresponding time slot The second unloading time ratio and the optimal time ratio of the base station charging the wireless device are calculated to obtain the first unloading time ratio of all wireless devices. The specific process is as follows:

[0081] First, it should be noted that for each offloading decision, since each wireless device communicates with the base station in a time division multiplexing manner, The base station charges all wireless devices within a certain period of time. After charging is completed, each wireless device will unload in turn according to the preset unloading decision in the remaining time, that is, the first unloading time ratio of each wireless device (where the first unloading time ratio is the time slot occupied by the wireless device The ratio of the time the base station charges all wireless devices satisfies the following relationship:

[0082] ;

[0083] For each offloading decision, the total time for all wireless devices to offload according to the preset offloading decision is ;

[0084] Multiply the second unloading duration ratio between the wireless devices (where the second unloading duration ratio is the ratio of the wireless device to the total unloading duration) by the total duration The unloading duration of each wireless device is obtained, and the calculation formula is as follows:

[0085] ;

[0086] Right now,

[0087] .

[0088] For various uninstall decisions , when all wireless devices’ offloading decisions are to stay in the local wireless device, that is, all When it is 0, the objective function of maximizing the safe computing rate is only proportional to the time the base station charges the wireless device. related to and about is a monotonically increasing function, so the optimal ;

[0089] For various uninstall decisions , when at least one of all wireless devices decides to offload to the base station, that is, at least one When is 1, the objective function of maximizing the safe calculation rate is converted to :

[0090] ;

[0091] At this time, for each uninstall decision , the objective function Only is an unknown number, and the optimal Make is the maximum value (the proportion of time the base station charges all wireless devices Therefore, when at least one wireless device among all wireless devices decides to offload to the base station, in order to make To reach the optimal value, the optimal , so that the value of the objective function is maximized), and then various unloading decisions are obtained The optimal time ratio for base stations to charge wireless devices ;

[0092] It should be noted that the objective function cannot be directly obtained middle The value of is proved as follows:

[0093] For the objective function , let the known quantity , =C, , , , ,but ;

[0094] The objective function Translates to:

[0095] ;

[0096] Because the maximization objective function is to solve the problem when the derivative of the objective function is zero So find the derivative of the objective function and find the value when the derivative is zero. The value of the objective function Find the first derivative and set it to 0:

[0097] ;

[0098] According to the formula of the first-order derivative of the above formula, we can know that: , , , so it is impossible to directly find The value of , so the optimal Make is the maximum value;

[0099] According to the formula , get various uninstall decisions The first unloading duration ratio of each wireless device .

[0100] Step 8: According to the corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving The internal unloading decision, the first unloading time ratio of all wireless devices and the optimal charging time ratio are calculated to obtain the corresponding time slot The final maximum safe computation rate within , the process is as follows:

[0101] According to the corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving Each uninstall decision within , the first uninstallation time ratio of all wireless devices And the optimal charging time ratio , calculate the objective function to get the corresponding time slot Internal and external uninstall decisions The corresponding maximum security calculation rate is selected from all the maximum security calculation rates as the corresponding time slot The final maximum safe calculation rate .

[0102] Step 9: First strategy network and the second strategy network Update and use the updated strategy networks as the strategy networks for the next time slot;

[0103] It should be noted that in step 5, the first strategy network The parameters of the benchmark network are copied , and the benchmark network Updated every preset time slot (reference network Update once every preset time slot (e.g. every 1000 time slots) Once), and update the baseline network When the first strategy network corresponding to the time slot The parameters are copied to obtain the corresponding benchmark network ); The first strategy network Replace with the baseline network , and follow steps 5 to 8 to get the corresponding maximum safe calculation rate , and in the benchmark network When preprocessing the output offloading probabilities, the normalized baseline network The output offloading probabilities are randomly sampled to obtain the benchmark network A corresponding sampling probability, and the unloading decision corresponding to the sampling probability is Figure 3 Medium express;

[0104] pass and First Strategy Network and the second strategy network Update, and the first strategy network The updated expression is: ,in, is the updated first strategy network, First Strategy Network The learning rate, First Strategy Network The policy network function is First Strategy Network Output action (i.e., uninstall probability); Second policy network The updated expression is: ,in, is the updated second strategy network, Second Strategy Network The learning rate, Second Strategy Network The policy network function is Second Strategy Network Output action (i.e. );

[0105] The updated first policy network and the second strategy network As the first policy network for the next time slot and the second strategy network ;

[0106] Next time slot Repeat steps 4 to 9 to obtain the corresponding maximum security calculation rate, and repeat continuously to obtain each time slot (i.e. time slots) corresponding to the final maximized security calculation rate.

[0107] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0108] The method for maximizing the safe computing rate in the wireless power supply edge computing network first constructs an objective function of maximizing the safe computing rate regarding the first channel gain, the second channel gain, the third channel gain, the preset unloading decision, the first unloading time ratio, the time ratio of the base station charging the wireless device, and the power matrix of the base station emitting artificial noise, and solves the unloading decision, the first unloading time ratio, and the time ratio of the base station charging the wireless device in the objective function, and calculates the objective function to obtain the maximized safe computing rate; at the same time, the method comprehensively considers multiple aspects such as wireless device charging, computing power, and safety, and maximizes the computing rate while ensuring safety.

[0109] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for maximizing safe computing rate in a wireless power supply edge computing network, characterized in that: The method for maximizing the safe computing rate in the wireless power supply edge computing network includes: Step 1: The wireless power supply edge computing network includes a base station, There are wireless devices and an eavesdropper, and the base station transmits wireless radio frequency energy to each wireless device to charge the wireless device, and the base station transmits artificial noise to interfere with the eavesdropper's eavesdropping on each wireless device; Step 2: Random Generation The first channel gain between the wireless device and the base station in the time slot is randomly generated The second channel gain between each wireless device and the eavesdropper in time slots is randomly generated The third channel gain between the base station and the eavesdropper for each time slot; Step 3: First time slot Inside, randomly initialize the first policy network of reinforcement learning , and a second policy network with randomly initialized reinforcement learning ; Step 4: Establish the corresponding time slot an objective function of maximizing a safe computing rate with respect to a first channel gain, a second channel gain, a third channel gain, a preset unloading decision, a first unloading time ratio of all wireless devices, a time ratio of a base station charging each wireless device, and a power matrix of artificial noise emitted by the base station; The formula of the objective function of maximizing the safe computing rate is as follows: ； in, ； ； ； ； ； in, To maximize the objective function of safe calculation rate, represents the set of the first channel gain, the second channel gain and the third channel gain, is the set of preset uninstall decisions obtained, and contains Uninstallation decision, For the offloading decisions, and includes the offloading decisions of all wireless devices, For the The first uninstall decision offload decision of each wireless device, is a set of first offloading duration ratios of all wireless devices corresponding to all types of offloading decisions, For the A set of first unloading duration ratios of all wireless devices corresponding to the unloading decision, For the The first uninstall decision The first unloading duration ratio of wireless devices, is the set of the proportion of the time that the base station charges all wireless devices corresponding to all types of offloading decisions, For the The proportion of time that the base station charges all wireless devices corresponding to the offloading decision, and The value range is , is the power matrix of artificial noise emitted by the base station, is the number of cycles required for each wireless communication device to process one bit of data, is the energy capture efficiency of wireless devices, is the power of the radio frequency energy transmitted by the base station, For the The channel gain from a wireless device to the base station, is the computational efficiency of each wireless device, is the bandwidth from the base station to each wireless device, Offload the communication overhead of tasks for each wireless device, For the The signal-to-noise ratio from a wireless device to the base station, For the The signal-to-noise ratio from a wireless device to an eavesdropper; Step 5: Assign each wireless device to a time slot The first channel gain is input to the first strategy network , get the time slot corresponding to each wireless device The unloading probability within the time slot corresponding to each wireless device The unloading probability in the preprocessing is solved to obtain the corresponding time slot Preset uninstallation decision within the system; Step 6: Assign each wireless device to a time slot The first channel gain is input to the second strategy network , get the corresponding time slot Second Strategy Network Regarding the output of each wireless device, according to the corresponding time slot Second Strategy Network The corresponding time slot is obtained by calculating the output and offloading decision of each wireless device The second unloading duration ratio between the wireless devices in the wireless network; Step 7: Solve to get the corresponding time slot The optimal time ratio for the internal base station to charge the wireless device, and according to the corresponding time slot The second unloading time ratio and the optimal time ratio of the base station charging the wireless device are solved to obtain the first unloading time ratio of all wireless devices; Step 8: According to the corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving The internal unloading decision, the first unloading time ratio of all wireless devices and the optimal charging time ratio are calculated to obtain the corresponding time slot The final maximum safe computation rate within ; Step 9: First strategy network and the second strategy network Update, use the updated strategy networks as the strategy networks for the next time slot, enter the next time slot, repeat steps 4 to 9, and then get The final maximum security calculation rate for time slots.

2. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 1, characterized in that: The offloading probability includes the probability of staying on the local wireless device and the probability of offloading to the base station.

3. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 2, characterized in that: The offloading probability of each wireless device is preprocessed to obtain a preset offloading decision, including: The offloading probability of each wireless device is normalized, and then the normalized offloading probabilities are processed in turn. Random sampling The sampling probability of all wireless devices is the normalized probability of remaining on the local wireless device or the normalized probability of being offloaded to the base station; When the normalized probability of staying on the local wireless device is obtained by random sampling, the offloading decision of the corresponding wireless device is to stay on the local wireless device, and , when the normalized probability of offloading to the base station is obtained by random sampling, the offloading decision of the corresponding wireless device is to offload to the base station, and 1.

4. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 1, characterized in that: The second policy network The output set of each wireless device is expressed as ,in Represents the second policy network About output of a wireless device.

5. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 3, characterized in that: According to the second strategy network The second offloading duration ratio between the wireless devices is obtained by calculating the output and offloading decision of each wireless device, including: Traverse each uninstall decision and uninstall the decision The offloading decision of each wireless device in the second strategy network is respectively The outputs of each wireless device are dot-multiplied one by one, and the result after dot-multiplication is normalized to obtain the second unloading duration ratio between each wireless device, which is expressed as ,in, For the A set of second unloading duration ratios of all wireless devices corresponding to the unloading decision, For the The first uninstall decision The second unloading duration ratio of each wireless device.

6. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 5, characterized in that: For each offloading decision, in the time slot Within, the base station charges all wireless devices. After charging is completed, each wireless device unloads in turn according to the preset unloading decision in the remaining time, that is, the first unloading time ratio of each wireless device and the time ratio of the base station charging all wireless devices meet the following relationship: ； For each offloading decision, the total time for all wireless devices to offload according to the preset offloading decision is ; Multiply the second unloading duration ratio between wireless devices by the total duration The unloading duration of each wireless device is obtained, and the calculation formula is as follows: ； Right now, 。 7. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 6, characterized in that: The solution obtains the corresponding time slot The optimal time ratio for the internal base station to charge the wireless device, and according to the corresponding time slot The second unloading time ratio and the optimal time ratio of the base station charging the wireless device are solved to obtain the first unloading time ratio of all wireless devices, including: For various uninstall decisions , when all wireless devices’ offloading decisions are to stay in the local wireless device, that is, all When it is 0, the objective function of maximizing the safe computing rate is only proportional to the time the base station charges the wireless device. related to and about is a monotonically increasing function, so the optimal ; For various uninstall decisions , when at least one of all wireless devices decides to offload to the base station, that is, at least one When is 1, the objective function of maximizing the safe calculation rate is converted to : ； At this time, for each uninstall decision , the objective function Only is an unknown number, and the optimal Make is the maximum value, and then various unloading decisions are obtained The optimal time ratio for base stations to charge wireless devices ; According to the formula , get various uninstall decisions The first unloading duration ratio of each wireless device .

8. The method for maximizing the safe computing rate in a wireless power supply edge computing network according to claim 7, characterized in that: The corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving The internal unloading decision, the first unloading time ratio of all wireless devices and the optimal charging time ratio are calculated to obtain the corresponding time slot The final maximum safe computation rate within ,include: According to the corresponding time slot The first channel gain, the second channel gain, the third channel gain and the power matrix of the artificial noise transmitted by the base station, and the corresponding time slot obtained by solving Each uninstall decision within , the first uninstallation time ratio of all wireless devices And the optimal charging time ratio , calculate the objective function to get the corresponding time slot Internal and external uninstall decisions The corresponding maximum security calculation rate is selected from all the maximum security calculation rates as the corresponding time slot The final maximum safe calculation rate .

9. The method for maximizing safe computing rate in a wireless power supply edge computing network according to claim 1, characterized in that: The method for maximizing the secure computing rate in a wireless power supply edge computing network also includes a first strategy network The parameters of the benchmark network are copied , and the benchmark network Updated every preset time slot, where the reference network is updated When the first strategy network corresponding to the time slot The parameters are copied to obtain the corresponding benchmark network ; First Strategy Network Replace with the baseline network , and follow steps 5 to 8 to obtain the corresponding maximum safe calculation rate , and in the benchmark network When preprocessing the output offloading probabilities, the normalized baseline network The output offloading probabilities are randomly sampled to obtain the benchmark network A corresponding sampling probability; pass and First Strategy Network and the second strategy network Update, and the first strategy network The updated expression is: ,in, is the updated first strategy network, First Strategy Network The learning rate, First Strategy Network The policy network function is First Strategy Network Output action, second policy network The updated expression is: ,in, is the updated second strategy network, Second Strategy Network The learning rate, Second Strategy Network The policy network function is Second Strategy Network Output action.

Images