A baseline-based automated edge computing network resource allocation method

Abstract

The application belongs to the field of mobile edge computing, and discloses a baseline-based automatic edge computing network resource allocation method. The method comprises the following steps: acquiring channel gains between each wireless device and each hybrid access point; inputting the channel gains into discrete network branch prediction offloading decisions; inputting the channel gains and the offloading decisions into parallel first and second continuous network branches to predict energy allocation ratios and offloading time allocation ratios; taking the parallel branches after historical iteration training as a baseline, outputting historical prediction results according to the channel gains of the current time slot; calculating maximum computing rates according to the prediction results of the current time slot and the history respectively, and updating the discrete network branches; calculating the maximum computing rate of the current time slot according to the prediction results of the current time slot, and updating the first continuous network branch and the second continuous network branch. In the case that distributed renewable energy is used for wireless device functions, the maximum computing rate of the overall edge computing network is high.

Description

Technical Field

[0001] This invention belongs to the field of mobile edge computing, specifically relating to a baseline-based automated edge computing network resource allocation method. Background Technology

[0002] In recent years, the Internet of Things (IoT) has experienced tremendous growth, with the scale of IoT devices expanding rapidly and their application in scenarios such as autonomous driving, smart homes, and smart cities showing a continuous increasing trend. Simultaneously, artificial intelligence (AI) technology is also constantly evolving, making IoT devices increasingly intelligent. The large-scale deployment of devices and highly intelligent technologies generates massive amounts of computing data and real-time demands. However, IoT devices are often limited by factors such as battery capacity, preventing them from performing a large number of high-performance computing tasks. Therefore, new technological approaches are urgently needed to simultaneously improve computing power and energy self-sufficiency.

[0003] Mobile edge computing (MEC) technology has emerged as a potential solution to this challenge. Deploying edge servers near terminal devices to assist or even replace them in performing computational tasks can significantly reduce latency and improve real-time computing capabilities. For terminal devices with limited battery life, distributed power sources based on renewable energy can provide a continuous and stable supply of clean energy, ensuring reliable energy for local computation and data offloading. Optimizing limited resources to maximize computing performance is an effective way to overcome the computational limitations of IoT devices. Currently, various algorithms exist to address the problems inherent in typical IoT devices, such as dynamic programming-based algorithms to maximize computation completion rates and heuristic algorithms to minimize service latency. However, these algorithms often perform well with small sample sizes. As the edge computing network expands, the search space grows exponentially, leading to a decline in algorithm performance and a significant increase in computation time. There are also deep reinforcement learning (such as deep Q-networks) to minimize latency and energy consumption. Ordinary deep reinforcement learning (DRL) algorithms are only used to optimize offloading decisions and cannot incorporate other resource optimizations, such as time allocation. They need to be combined with other mathematical optimization algorithms. This approach requires designing mathematical optimization algorithms for specific problems and is only effective for convex optimization problems, making it difficult to apply to non-convex optimization problems. Summary of the Invention

[0004] The purpose of this invention is to provide a baseline-based automated edge computing network resource allocation method that achieves a higher overall maximum computing rate for the edge computing network when distributed renewable energy serves as the function of wireless devices.

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

[0006] A baseline-based automated edge computing network resource allocation method, wherein the edge computing network has A hybrid access point and A wireless device, including the following steps:

[0007] After the start of each time slot, the channel gain between each wireless device and each hybrid access point is obtained to form the channel set of the current time slot;

[0008] The channel set of the current time slot is input into a parallel discrete network branch to predict the offloading decision of each wireless device, thus obtaining the offloading decision set of the current time slot.

[0009] Input the channel set and offload decision set of the current time slot into the first continuous network branch in parallel to predict the energy allocation ratio of each wireless device;

[0010] The channel set and offload decision set of the current time slot are input into the parallel second continuous network branch to predict the offload time allocation ratio of each wireless device;

[0011] Take the first in history After each iteration of training, the parallel discrete network branch, the first continuous network branch, and the second continuous network branch serve as the baseline. Based on the channel set of the current time slot, the historical offloading decision, energy allocation ratio, and offloading time allocation ratio are output.

[0012] Based on the current time slot and historical unloading decisions, energy allocation ratios, and unloading time allocation ratios, the maximum computation rate is calculated to form a discrete loss function to update the discrete network branches.

[0013] Based on the unloading decision, energy allocation ratio, and unloading time allocation ratio of the current time slot, calculate the maximum computation rate of the current time slot, form a continuous loss function to update the first and second continuous network branches, the time slot ends, and training is repeated until all rounds.

[0014] Several alternative methods are provided below, but they are not intended as additional limitations on the overall solution above. They are merely further additions or optimizations. Provided there are no technical or logical contradictions, each alternative method can be combined individually with respect to the overall solution above, or multiple alternative methods can be combined with each other.

[0015] Preferably, the discrete network branch includes a discrete network and a normalization layer, wherein the discrete network includes an input layer, a multi-scale pyramid network, and an output layer;

[0016] The channel set of the current time slot is input into the input layer of the discrete network. After passing through the multi-scale pyramid network, the output layer outputs the first matrix. The first matrix is ​​normalized by the SoftMax normalization operation of the normalization layer to obtain the probability matrix. Random offloading sampling is performed according to the probability distribution in the probability matrix, and the offloading decision of each wireless device is output.

[0017] Preferably, the first continuous network branch includes a first continuous network and a normalization layer, wherein the first continuous network includes an input layer, a multi-scale pyramid network, and an output layer.

[0018] The current time slot's channel set and offload decision set are combined into a two-channel matrix, which is input into the input layer of the first continuous network. After passing through a multi-scale pyramid network, the output layer outputs a second matrix. After the second matrix undergoes SoftMax normalization operation in the normalization layer, the energy allocation ratio obtained by the local calculation of all wireless devices is obtained.

[0019] Preferably, the second continuous network branch includes a second continuous network and a normalization layer, wherein the second continuous network includes an input layer, a multi-scale pyramid network, and an output layer.

[0020] The channel set and offloading decision set of the current time slot are combined into a two-channel matrix and input into the input layer of the second continuous network. After passing through the multi-scale pyramid network, the output layer outputs a third matrix. Based on the offloading decision set of the current time slot, the normalization layer performs SoftMax normalization on the element values ​​of the wireless devices offloading to the same hybrid access point in the third matrix to obtain the offloading time allocation ratio of each wireless device.

[0021] Preferably, the multi-scale pyramid network includes convolutional pyramid layers, pooling pyramid layers, and fully connected layers;

[0022] The convolutional pyramid layer contains The pooling pyramid layer contains dilated convolutions with different kernel sizes. A pooled pyramid, and each pooled pyramid performs... Downsampling operations in different pooling regions, the input data is passed through a convolutional pyramid layer to obtain... Each feature map is processed through a pooling pyramid to obtain the final result. Each feature map will Each feature map is unfolded into a one-dimensional vector and concatenated. The concatenated features are then passed through a fully connected layer to obtain the output data of the multi-scale pyramid network.

[0023] Preferably, the baseline is obtained in the following way:

[0024] After each round of training iterations, the discrete network branches, the first continuous network branch, and the second continuous network branch are stored in a cache queue. If the number of cached branches exceeds the size of the cache queue, the earliest cached element is removed from the cache queue according to the first-in-first-out principle, and the first element in the cache queue is taken as the baseline for training.

[0025] Preferably, the step of forming a discrete loss function to update the discrete network branch includes:

[0026] Calculate the maximum computing rate of the current time slot based on the offloading decision, energy allocation ratio, and offloading time allocation ratio of the current time slot;

[0027] Calculate the historical maximum computing rate based on historical unloading decisions, energy allocation ratios, and unloading time allocation ratios;

[0028] The discrete loss function is calculated based on the maximum computation rate of the previous time slot and the historical maximum computation rate as follows:

[0029]

[0030] In the formula, For discrete loss functions, The probability matrix is ​​the nth List, The absolute value of the difference in maximum computational speed. This represents the maximum computation rate for the current time slot. This represents the highest computing speed in history. For decision parameters, if ,but ;otherwise .

[0031] Preferably, the continuous loss function is the reciprocal of the logarithm of the maximum computation rate of the current time slot.

[0032] The baseline-based automated edge computing network resource allocation method provided by this invention has the following advantages compared with the prior art:

[0033] First, this invention proposes a baseline-based method for evaluating the effectiveness of optimization schemes. Unlike current supervised learning methods that use a base algorithm as a baseline to measure the percentage improvement in algorithm performance, where the baseline in most supervised learning systems is a static value serving only as a comparison point and not participating in algorithm training and updates, the baseline proposed in this invention is based on the network itself and dynamically adjusts to different baselines at different training epochs. This allows for the evaluation of the effectiveness of optimization strategies produced by the network in the current state. The baseline value participates in the calculation of the loss function, providing a new metric for unsupervised problems to guide network parameter updates, enabling the model to converge quickly and improving model performance.

[0034] Secondly, this invention constructs a multi-scale pyramid network structure. By combining convolutional pyramids and pooling pyramids, this network structure can transform input data of different dimensions into feature vectors of a unified dimension. It is applicable to edge computing network scenarios where the number of nodes varies, eliminating the need to redesign network parameters for model training. To reduce the problem of excessively large decision spaces, common multi-agent algorithms divide the entire edge computing network into multiple sub-networks, with each agent responsible for one sub-network. However, this approach only allows the agent to see a portion of the overall network, passing its observations to the master node for processing and executing the master node's decisions. In this mode, the information obtained by the master network heavily depends on the sub-networks; any deviation in any sub-network will affect the master network's judgment. The master network struggles to obtain the original global information, easily leading to decision errors. This invention employs a parallel multi-network architecture design where each sub-network receives global information and can independently perform loss calculations and parameter updates. This makes network training more flexible, facilitates individual network debugging, improves model training efficiency, and enhances the final prediction performance of the model.

[0035] Finally, this invention adopts a parallel multi-network architecture design. The discrete network can predict the unloading decision in the optimization problem, and the parallel continuous network can predict the energy and time allocation. This method can easily adjust the number of networks and be extended to other optimization scenarios. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of an edge computing network structure;

[0037] Figure 2 This is a flowchart of the baseline-based automated edge computing network resource allocation method of the present invention;

[0038] Figure 3 This is a schematic diagram of the structure of the three parallel neural networks of the present invention;

[0039] Figure 4 This is a schematic diagram of the multi-scale pyramid network structure of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention.

[0042] like Figure 1 As shown, the edge computing network in this embodiment has A hybrid access point and The system comprises several wireless devices, each with a hybrid access point equipped with an edge server and power lines, and powered by renewable energy sources. This embodiment employs Time Division Multiple Access (TDMA) technology for data communication between the wireless devices (terminal devices) and the edge server. When multiple terminal devices simultaneously offload data to the same edge server, different time slots are allocated to different wireless devices. During data offloading, a partial offloading mode is used, meaning data on the wireless device can be arbitrarily divided. During data offloading, any proportion of data can be selected to be offloaded to the edge server, and each wireless device can only select one edge server for data offloading.

[0043] In one embodiment, the entire edge computing network scenario comprises the following components: 3 hybrid access points and 5 wireless devices; [Let the hybrid access point set...] and wireless device collection The entire system time is divided into multiple very short time slots. It is the length of a time slot. Within a time slot, the energy obtained by each wireless device from the renewable energy source is denoted as... , Indicates the first The renewable energy acquisition rate of a wireless device is typically measured in energy per unit time (e.g., joules per second). The wireless device and the first The channel gain between the hybrid access points is denoted as . .

[0044] During local task computation, each wireless device generates data that needs to be computed in each time slot. Of the wireless devices, the proportion of energy harvested used for local computing is [percentage missing]. Therefore, the remaining energy used for offloading to the hybrid access point is... ; Indicates the first The CPU (Central Processing Unit) computing speed of a wireless device, measured in cycles per second; This represents the number of loops required for the CPU to process 1 bit of data; therefore, in The total computational load of the data within a given time period can be expressed as: The CPU requires the following energy: The CPU consumes the following energy: Then you can get ; So, for the first The computing speed of a wireless device's local computation can be expressed as:

[0045]

[0046] in, For the first The computation rate of local computation by each wireless device, a constant. , This is the CPU's dynamic power consumption factor.

[0047] For task offloading computation, the wireless device needs to first offload the data to the hybrid access point, and then the hybrid access point performs the computation; for the first... For an individual wireless device, assuming it offloads data to a hybrid access point The percentage of time required for uninstallation is marked as follows. Therefore, the unloading time of all devices and the time for energy acquisition should not exceed the duration of the time slot. , can be represented as:

[0048]

[0049] If the The wireless device offloads the data to the first... On a hybrid access point If the bandwidth is represented by the offload rate, then the offload rate can be expressed as:

[0050]

[0051] in, Indicates the first The wireless device offloads the data to the first... Offload rate on a hybrid access point, constant , This indicates additional data transferred during the unloading process. This represents Gaussian noise.

[0052] like Figure 2 As shown, this embodiment presents a baseline-based automated edge computing network resource allocation method to address resource allocation issues in edge computing networks. This method aims to address data communication scenarios between wireless device terminals and edge servers by proposing a baseline-based resource allocation strategy evaluation approach. It employs a multi-scale pyramid structure to design a network architecture for offloading decision-making and resource allocation prediction. This network architecture provides a unified output across different input dimensions without requiring modification of network parameters, adapting to dynamic changes in nodes within the edge computing network. Furthermore, it utilizes multiple networks in parallel for prediction, avoiding complex optimization algorithms and achieving flexible, simple, and efficient optimization results. Specifically, it includes the following steps:

[0053] Step 1: After the start of each time slot, obtain the channel gain between each wireless device and each hybrid access point to form the channel set for the current time slot. Channel Set , representing the channel gain between the five wireless devices and all hybrid access points, where, Indicates the first Channel gain of a wireless device.

[0054] Step 2: Input the channel set of the current time slot into the parallel discrete network branch to predict the offloading decision of each wireless device and obtain the offloading decision set of the current time slot.

[0055] like Figure 3 As shown, the discrete network branch in this embodiment includes a discrete network and a normalization layer. The discrete network includes an input layer, a multi-scale pyramid network, and an output layer. The channel set of the current time slot is input to the input layer of the discrete network, and after passing through the multi-scale pyramid network, the first matrix is ​​output by the output layer. The first matrix The probability matrix is ​​obtained after the SoftMax normalization operation of the normalization layer. Random unloading sampling is performed according to the probability distribution in the probability matrix, and the unloading decision for each wireless device is output.

[0056] Taking this embodiment as an example, the input layer of the discrete network is a set of channels. , representing the channel gain between the 5 wireless devices and all hybrid access points; the input layer is followed by a multi-scale pyramid network, and the output is The first matrix For the first matrix Perform SoftMax normalization to obtain the probability matrix. According to the probability matrix Random offloading sampling is performed using the probability distribution in the data (e.g., the probability of a device offloading to three hybrid access points is respectively...). Then there is a 0.1 probability of sampling to hybrid access point 1, a 0.2 probability of sampling to hybrid access point 2, and a 0.7 probability of sampling to hybrid access point 3, resulting in the offloading decision set. ,in, Indicates the first The decision to offload a wireless device, among which... Indicates the first The wireless device offloads the data to the first... On a hybrid access point, conversely... This indicates that the software has not been uninstalled.

[0057] Step 3: Input the channel set and offload decision set of the current time slot into the first continuous network branch in parallel to predict the energy allocation ratio of each wireless device.

[0058] In this embodiment, the first continuous network branch includes a first continuous network and a normalization layer. The first continuous network includes an input layer, a multi-scale pyramid network, and an output layer. The channel set and offload decision set of the current time slot are combined into a two-channel matrix and input into the input layer of the first continuous network. After passing through the multi-scale pyramid network, the output layer outputs a second matrix. After the second matrix is ​​normalized by the SoftMax normalization operation of the normalization layer, the energy allocation ratio obtained by the local calculation of all wireless devices is obtained.

[0059] Step 4: Input the channel set and offload decision set of the current time slot into the parallel second continuous network branch to predict the offload time allocation ratio of each wireless device.

[0060] In this embodiment, the second continuous network branch includes a second continuous network and a normalization layer. The second continuous network includes an input layer, a multi-scale pyramid network, and an output layer. The channel set and offloading decision set of the current time slot are combined to form a two-channel matrix, which is input to the input layer of the second continuous network. After passing through the multi-scale pyramid network, the output layer outputs a third matrix. According to the offloading decision set of the current time slot, the normalization layer performs SoftMax normalization on the element values ​​of the wireless devices offloaded to the same hybrid access point in the third matrix to obtain the offloading time allocation ratio of each wireless device.

[0061] This embodiment utilizes a parallel continuous network design to obtain energy and time allocation; the input layer of the continuous network is a set of channels. and unloading decision set The two-channel matrix formed The input layer is followed by a multi-scale pyramid network; the first continuous network outputs the second matrix. For the second matrix Perform a sigmoid normalization operation to obtain the set of energy allocation ratios calculated locally by all wireless devices. The output of the second continuous network is the third matrix. According to the unloading decision set Perform SoftMax normalization on the values ​​of wireless devices offloaded to the same hybrid access point to obtain the set of offload time allocation ratios. ,in Indicates the first A hybrid access point, Used with To associate, specifically the first A hybrid access point corresponding to each wireless device.

[0062] For example, there are 5 wireless devices, and their numbers are as follows: This means that the 1st and 5th wireless devices are offloaded to the same hybrid access point, and the 2nd and 4th wireless devices are offloaded to the same hybrid access point. When performing the SoftMax normalization operation, the data corresponding to the 1st and 5th wireless devices in the third matrix jointly perform the SoftMax normalization operation once, the data corresponding to the 2nd and 4th wireless devices jointly perform the SoftMax normalization operation once, and the data corresponding to the 3rd wireless device performs the SoftMax normalization operation separately, thereby obtaining the offload time allocation ratio.

[0063] like Figure 4 As shown, the multi-scale pyramid network used in this embodiment across three parallel branches includes convolutional pyramid layers, pooling pyramid layers, and fully connected layers. The convolutional pyramid layer contains... Dilated convolutions with different kernel sizes, pooling pyramid layers contain A pooled pyramid, and each pooled pyramid performs... Downsampling operations in different pooling regions, the input data is passed through a convolutional pyramid layer to obtain... Each feature map is processed through a pooling pyramid to obtain the final result. Each feature map will Each feature map is unfolded into a one-dimensional vector and concatenated. The concatenated features are then passed through a fully connected layer to obtain the output data of the multi-scale pyramid network.

[0064] For example, this embodiment provides a multi-scale pyramid network structure design, constructing three dilated convolutions with different kernel sizes, the kernel size being... The input data is convolved using three convolution kernels, with kernel sizes of [sizes to be filled in]. , , Obtain 3 feature maps Then, for each feature map, downsampling is performed on two different pooling regions (downsampling of different pooling regions can be achieved by setting different pooling convolution kernel sizes and / or strides, or by directly sampling pooling layers or performing pooling convolutions), finally resulting in 6 feature maps. All feature maps are then unfolded into one-dimensional vectors and concatenated, followed by fully connected layers and output layers to output data.

[0065] Step 5: Retrieve the historical number After each iteration of training, the parallel discrete network branch, the first continuous network branch, and the second continuous network branch serve as the baseline. Based on the channel set of the current time slot, the historical offloading decision, energy allocation ratio, and offloading time allocation ratio are output.

[0066] To overcome the shortcomings of static baselines in existing technologies, this embodiment stores the parallel discrete network branches, the first continuous network branch, and the second continuous network branch after each round of training iterations in a cache queue. If the number of cached branches exceeds the size of the cache queue, the earliest cached element is removed from the cache queue according to the first-in-first-out principle, and the first element in the cache queue (denoted as the historical first element) is retrieved. The parallel network model trained in the previous iteration is used as a baseline for training.

[0067] For example, in this embodiment, the queue size is set to 100. If the number of cached networks exceeds 100, the earliest model is removed according to the first-in, first-out principle of the queue. The number of training rounds of the model is denoted as... ,if If the baseline is set to 0, then the historical maximum calculation rate will be used in subsequent loss calculations. ;if Then, the first set of network parameters in the buffer queue is used to predict the channel under the current state, and the baseline network output result is obtained. , For the set of historical unloading decisions, A collection of historical energy distribution ratios. A set of historical unloading time allocation ratios.

[0068] Step 6: Calculate the maximum computation rate based on the current time slot and historical unloading decisions, energy allocation ratios, and unloading time allocation ratios, and form a discrete loss function to update the discrete network branches.

[0069] This embodiment provides the maximum computing speed. The formula is as follows:

[0070]

[0071] Therefore, when calculating the discrete loss function, the maximum computation rate of the current time slot is calculated based on the unloading decision, energy allocation ratio, and unloading time allocation ratio of the current time slot, denoted as . Based on historical unloading decisions, energy allocation ratios, and unloading time allocation ratios, calculate the historical maximum computing rate, denoted as . The discrete loss function is then calculated based on the maximum computation rate of the previous time slot and the historical maximum computation rate as follows:

[0072]

[0073] In the formula, For discrete loss functions, The probability matrix is ​​the nth List, The absolute value of the difference in maximum computational speed. For decision parameters, if Greater than ,but ;otherwise .

[0074] Step 7: Calculate the maximum computation rate of the current time slot based on the unloading decision, energy allocation ratio, and unloading time allocation ratio of the current time slot. Form a continuous loss function to update the first and second continuous network branches. The time slot ends, and training is repeated until all rounds.

[0075] In this embodiment, when calculating the continuous loss function, the reciprocal of the logarithm of the maximum computation rate of the current time slot is used. , as a continuous loss function.

[0076] During network training, the entire architecture comprises three parallel neural networks (one discrete network and two continuous networks). The batch size for neural network training is 1, meaning that each iteration involves retrieving channel data from the environment for the current time slot and updating the neural network model. A total of 40,000 iterations are trained. All three neural networks use the AdamW optimizer (Adam with Weight Decay). The learning rate for the discrete network is 3e-5, and the learning rate for the two continuous networks is 3e-6. Training is repeated until all iterations are completed. After training, the three parallel neural networks can output the offloading decision, energy allocation ratio, and offloading time allocation ratio for each wireless network based on the channel data of the current time slot.

[0077] This invention proposes a baseline-based multi-scale pyramid network architecture, utilizing a multi-network parallel deep reinforcement learning method to solve the resource allocation problem in edge computing networks powered by renewable energy. It also proposes a baseline-based optimization strategy evaluation method to assess the merits of the current network output strategy and guide the update of neural network parameters. This invention constructs a multi-scale pyramid network architecture that can adapt to data inputs of different dimensions, transforming them into a unified feature vector. In edge computing networks, if the number of nodes changes, the network model can continue to be used directly without redesigning the network structure. This invention employs multi-network parallel computation, utilizing discrete networks to obtain data offloading decisions and multiple continuous networks to obtain energy and time allocation ratios, achieving simplicity and efficiency. This avoids the problems of high optimization algorithm difficulty and time-consuming iterations required in traditional optimization methods. The resources to be optimized in this invention include the data offloading ratio and the time allocated to each terminal device in time-division multiple access.

[0078] For example, in an edge computing network containing 5 wireless devices and 3 hybrid access points, the corresponding channel matrix h = [[1.8e-3, 5.2e-5, 2.1e-6], [5.1e-5, 2.7e-5, 9.6e-5], [1.3e-6, 1.9e-5, 2.9e-5], [2.1e-5, 9.2e-4, 6.1e-6], [2.1e-5, 3.4e-5, 6.1e-6]], bandwidth... Hertz, the number of cycles required for the CPU of each wireless device to process 1 bit. Dynamic power consumption coefficient Additional data transfer during data offloading Gaussian noise Each time slot Seconds, the energy each device receives in each time slot kilowatts, then the optimized equipment unloading decision is: Energy distribution ratio Time allocation ratio .

[0079] It should be understood that, although Figure 2 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed 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 some of the sub-steps or stages of other steps.

[0080] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0081] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A baseline-based automated edge computing network resource allocation method, wherein the edge computing network has A hybrid access point and A wireless device, characterized in that... Includes the following steps: After the start of each time slot, the channel gain between each wireless device and each hybrid access point is obtained to form the channel set of the current time slot; The channel set of the current time slot is input into a parallel discrete network branch to predict the offloading decision of each wireless device, thus obtaining the offloading decision set of the current time slot. Input the channel set and offload decision set of the current time slot into the first continuous network branch in parallel to predict the energy allocation ratio of each wireless device; The channel set and offload decision set of the current time slot are input into the parallel second continuous network branch to predict the offload time allocation ratio of each wireless device; Take the first in history After each iteration of training, the parallel discrete network branch, the first continuous network branch, and the second continuous network branch serve as the baseline. Based on the channel set of the current time slot, the historical offloading decision, energy allocation ratio, and offloading time allocation ratio are output. Based on the current time slot and historical unloading decisions, energy allocation ratios, and unloading time allocation ratios, the maximum computation rate is calculated to form a discrete loss function to update the discrete network branches. Based on the unloading decision, energy allocation ratio, and unloading time allocation ratio of the current time slot, calculate the maximum computation rate of the current time slot, form a continuous loss function to update the first and second continuous network branches, the time slot ends, and training is repeated until all rounds.

2. The baseline-based automated edge computing network resource allocation method according to claim 1, characterized in that, The discrete network branch includes a discrete network and a normalization layer. The discrete network includes an input layer, a multi-scale pyramid network, and an output layer. The channel set of the current time slot is input into the input layer of the discrete network. After passing through the multi-scale pyramid network, the output layer outputs the first matrix. The first matrix is ​​normalized by the SoftMax normalization operation of the normalization layer to obtain the probability matrix. Random offloading sampling is performed according to the probability distribution in the probability matrix, and the offloading decision of each wireless device is output. The elements in the probability matrix represent the probability of the wireless device offloading to the hybrid access point.

3. The baseline-based automated edge computing network resource allocation method according to claim 1, characterized in that, The first continuous network branch includes a first continuous network and a normalization layer. The first continuous network includes an input layer, a multi-scale pyramid network, and an output layer. The current time slot's channel set and offload decision set are combined into a two-channel matrix, which is input into the input layer of the first continuous network. After passing through a multi-scale pyramid network, the output layer outputs a second matrix. After the second matrix undergoes SoftMax normalization operation in the normalization layer, the energy allocation ratio obtained by the local calculation of all wireless devices is obtained.

4. The baseline-based automated edge computing network resource allocation method according to claim 1, characterized in that, The second continuous network branch includes a second continuous network and a normalization layer. The second continuous network includes an input layer, a multi-scale pyramid network, and an output layer. The channel set and offloading decision set of the current time slot are combined into a two-channel matrix and input into the input layer of the second continuous network. After passing through the multi-scale pyramid network, the output layer outputs a third matrix. Based on the offloading decision set of the current time slot, the normalization layer performs SoftMax normalization on the element values ​​of the wireless devices offloading to the same hybrid access point in the third matrix to obtain the offloading time allocation ratio of each wireless device.

5. The baseline-based automated edge computing network resource allocation method according to any one of claims 2-4, characterized in that, The multi-scale pyramid network includes convolutional pyramid layers, pooling pyramid layers, and fully connected layers; The convolutional pyramid layer contains The pooling pyramid layer contains dilated convolutions with different kernel sizes. A pooled pyramid, and each pooled pyramid performs... Downsampling operations in different pooling regions, the input data is passed through a convolutional pyramid layer to obtain... Each feature map is processed through a pooling pyramid to obtain the final result. Each feature map will Each feature map is unfolded into a one-dimensional vector and concatenated. The concatenated features are then passed through a fully connected layer to obtain the output data of the multi-scale pyramid network.

6. The baseline-based automated edge computing network resource allocation method according to claim 1, characterized in that, The baseline is obtained as follows: After each round of training iterations, the discrete network branches, the first continuous network branch, and the second continuous network branch are stored in a cache queue. If the number of cached branches exceeds the size of the cache queue, the earliest cached element is removed from the cache queue according to the first-in-first-out principle, and the first element in the cache queue is taken as the baseline for training.

7. The baseline-based automated edge computing network resource allocation method according to claim 2, characterized in that, The process of forming a discrete loss function to update the discrete network branches includes: Calculate the maximum computing rate of the current time slot based on the offloading decision, energy allocation ratio, and offloading time allocation ratio of the current time slot; Calculate the historical maximum computing rate based on historical unloading decisions, energy allocation ratios, and unloading time allocation ratios; The discrete loss function is calculated based on the current maximum computation rate and the historical maximum computation rate as follows: ； In the formula, For discrete loss functions, The probability matrix is ​​the nth List, The absolute value of the difference in maximum computational speed. This represents the maximum computation rate for the current time slot. This represents the highest computing speed in history. For decision parameters, if > ,but ;otherwise .

8. The baseline-based automated edge computing network resource allocation method according to claim 1, characterized in that, The continuous loss function is the reciprocal of the logarithm of the maximum computation rate of the current time slot.

Images