An edge cloud computing load balancing method

Abstract

The present invention relates to the field of edge cloud computing technology, and discloses an edge cloud computing load balancing method. This method aims to solve the load balancing problem caused by the wide distribution, limited resources and dynamic changes of nodes in the edge computing environment. Traditional load balancing algorithms allocate requests based on static data and cannot accurately predict future load changes, easily leading to node overload or resource idleness. The present invention uses a neural network algorithm, especially the long short-term memory (LSTM) model, to construct and train a load prediction model. According to the historical usage of edge computing nodes, it predicts their future load levels. By deploying a monitoring and data collection system, it collects data on factors affecting node load in real time and inputs them into the prediction model to dynamically formulate resource scheduling strategies. The present invention realizes the dynamic and efficient utilization of edge computing nodes, improves the response speed and operation efficiency of the system, and optimizes the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of edge cloud computing, and specifically to an edge cloud computing load balancing method. Background Art

[0002] With the rapid development of Internet of Things, big data and artificial intelligence technologies, edge computing, as an emerging computing mode, has been gradually widely applied in various industries because it can reduce data transmission latency, improve data processing efficiency and protect user privacy. By deploying data processing capabilities at the network edge, that is, near the device end or terminal, edge computing realizes instant processing and feedback of data, greatly enhancing the response speed and operation efficiency of the system.

[0003] In the edge computing environment, due to the wide distribution of nodes, limited and dynamically changing resources, how to achieve efficient load balancing has become an urgent problem to be solved. Most traditional load balancing algorithms allocate requests based on current or historical static data and cannot accurately predict the future load change trend. Therefore, in the face of sudden traffic or periodic load peaks, some nodes are prone to overload while other node resources are idle, and the dynamic optimal allocation of resources cannot be effectively achieved. The edge computing environment is complex and changeable, and factors such as node performance, network conditions, and user behavior may all affect the load situation. Traditional load balancing methods are difficult to adapt to these dynamic changes in real time and cannot achieve accurate load prediction and resource scheduling, thus affecting the overall performance of the system and user experience. For this reason, we propose an edge cloud computing load balancing method. Summary of the Invention

[0004] The purpose of the present invention is to provide an edge cloud computing load balancing method to solve the problems raised in the above background art.

[0005] The working principle of this method is based on a neural network algorithm, especially the long short-term memory network (LSTM) model, to achieve load balancing in the edge cloud computing environment. It first collects historical load data of edge computing nodes, and uses a feature selection algorithm to screen out features useful for the load prediction model to construct a training data set. Then, the LSTM model is used to iteratively train the training data set to obtain a load prediction model that can accurately predict the future load level of nodes. In practical applications, this method deploys a monitoring and data acquisition system to collect data on factors affecting node load in real time and inputs this data into the trained load prediction model, thereby dynamically predicting the future load level of nodes. Finally, according to the prediction results, a resource scheduling strategy is dynamically formulated and implemented in the load balancer to achieve dynamic and efficient utilization of edge computing nodes.

[0006] To achieve the above object, the present invention provides the following technical solution: An edge cloud computing load balancing method, the steps of the method include:

[0007] Step 1: Use a neural network algorithm to construct and train a load prediction model, which is used to predict the load level of an edge computing node at a future time point according to the usage of the edge computing node. The specific steps of constructing and training the load prediction model include:

[0008] Step 1.1: Collect historical load data of the edge computing node, where the historical load data is all factor data that affects the node load at different load levels;

[0009] Step 1.2: Clean the collected data to handle missing values and outliers;

[0010] Step 1.3: Standardize the cleaned feature data;

[0011] Step 1.4: Use a feature selection algorithm to screen features useful for the load prediction model to form a feature data set;

[0012] Step 1.5: Mark the load level of each piece of data, and together with the corresponding feature data, form a training data set. Further divide the training data set into a training set and a validation set;

[0013] Step 1.6: Select a long short-term memory network (LSTM) model to construct the load prediction model and initialize the parameters of the model;

[0014] Step 1.7: Use the training set data to iteratively train the LSTM model. During the training process, regularly use the validation set to evaluate the model performance until the model performance meets the requirements;

[0015] Step 1.8: After the training is completed, save the weights and biases of the LSTM model and deploy the model for subsequent sampling location prediction;

[0016] Step 2: Deploy a monitoring and data collection system to collect all factor data that affects the load of the edge computing node in real time;

[0017] Step 3: Clean the data collected in Step 2 to handle missing values and outliers;

[0018] Step 4: Input the data cleaned in Step 3 into the load prediction model trained in Step 1 to obtain the load level of the edge computing node at a future time point;

[0019] Step 5: Dynamically formulate a resource scheduling policy according to the load level obtained in Step 4. The scheduling policy includes the number of edge computing nodes required to bear the load;

[0020] Step 6: Implement a scheduling strategy in the load balancer to achieve dynamic and efficient utilization of edge computing nodes.

[0021] Preferably, the historical load data collected for training the load prediction model includes resource usage metrics, network status metrics, and user behavior data of nodes at different load levels; the resource usage metrics include CPU usage rate, memory usage rate, disk I / O, storage space usage rate, number of processes, and number of threads; the network status metrics include network bandwidth usage rate, latency, packet loss rate, number of connections, and network traffic; the user behavior data includes user request volume, user access pattern, user session duration, user geographical location distribution, and user device type.

[0022] Preferably, in Step 1.2, for missing values, a linear interpolation algorithm is used for filling; for outliers, a threshold judgment method is used, and data points outside the preset threshold range are regarded as outliers and replaced with the mean of their neighboring points.

[0023] Preferably, in Step 1.3, the Z-score normalization method is used to subtract the mean of each feature value from it and divide by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for normalization processing is: where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the normalized data.

[0024] Preferably, in Step 1.4, the filter method is used for feature selection. This method evaluates the importance of features by calculating the correlation or statistical metrics between each feature and the target variable; the Pearson correlation coefficient is used to measure the linear correlation between each feature and the target variable. The formula for the Pearson correlation coefficient is:

[0025]

[0026] where r represents the correlation coefficient between the feature and the target variable, n represents the number of samples, x represents the feature value, and y represents the target variable value;

[0027] According to the calculated correlation coefficient, features with an absolute value greater than the set threshold are selected as useful features to form a feature dataset for subsequent training of the load prediction model.

[0028] Preferably, the specific steps for initializing the LSTM model include:

[0029] i) Initialize the structural parameters of the LSTM model, including the feature dimension of the input layer, the number of hidden layers and the number of hidden units in each layer, and the number of nodes in the output layer. The feature dimension of the input layer matches the number of features extracted that affect the load of the edge computing node, and the number of nodes in the output layer corresponds to the dimension of the predicted load level;

[0030] ii) Initialize the model weights and biases, and initialize them with small random numbers;

[0031] iii) Set the activation functions of the LSTM model, including setting the Sigmoid function for the gating mechanism and setting the Tanh function for the calculation of the candidate memory unit and the hidden state;

[0032] iv) Configure the optimization algorithm and set the learning rate parameter;

[0033] Iteratively train the LSTM model using the training set data. The specific method is as follows:

[0034] In each iteration, send a batch of input data into the LSTM model. By calculating the values of the input gate, forget gate, and output gate, update the memory unit and the hidden state, and finally obtain the predicted output result; calculate the loss function value according to the predicted output and the true label; calculate the gradient of the loss function with respect to the model parameters through the backpropagation algorithm; use the optimization algorithm to update the model weights and biases according to the gradient; during the training process, regularly evaluate the model performance using the validation set, and adjust the learning rate or stop training early as needed to prevent overfitting.

[0035] Preferably, in step 5, a threshold-based scheduling algorithm is adopted, and multiple load thresholds are preset, and each threshold corresponds to a different resource scheduling strategy; compare the predicted load level in step 5 with the preset load thresholds to determine the threshold range to which the current load level belongs; select the corresponding resource scheduling strategy according to the belonging threshold range;

[0036] The resource scheduling strategy includes the number of edge computing nodes required to undertake the load. Use a linear mapping function to calculate the required number of nodes. The calculation formula is:

[0037]

[0038] where round() represents the rounding function, the predicted load level represents the load level obtained in step 5, the lowest threshold and the highest threshold respectively represent the preset load threshold range, and the maximum number of nodes and the minimum number of nodes respectively represent the maximum and minimum available edge computing nodes.

[0039] Preferably, in step 2, for each key factor affecting the load of the edge computing node, a corresponding data acquisition module is configured to capture and record the data of the key factor in real time at a preset sampling frequency and time interval; and a data transmission interface between the data acquisition module and the central data storage system is formulated to enable the acquired data to be transmitted to the central data storage system in real time and accurately for subsequent processing and analysis;

[0040] Preferably, the data acquisition module adopts a lightweight data acquisition protocol.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] Through the neural network algorithm, the present invention can accurately predict the future load level of the edge computing node, thus realizing dynamic resource scheduling based on prediction. This method ensures that resources are allocated when needed, avoids the idle and waste of resources, and significantly improves the utilization efficiency of resources. Since the resources are utilized more effectively and the system response speed is accelerated, users will experience less latency and higher service quality when using edge computing services.

[0043] The edge computing environment is complex and changeable. Through real-time monitoring and data acquisition, the present invention can dynamically adapt to changes in factors such as node performance, network conditions, and user behavior, enabling the load balancing strategy to more precisely respond to the actual load situation and improving the stability and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a processing step diagram of the method of the present invention;

[0045] Figure 2 is a flowchart for initializing, training, and validating and optimizing the load prediction model;

[0046] Figure 3 is a step diagram for dynamically formulating a resource scheduling strategy. DETAILED DESCRIPTION OF THE INVENTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0048] Please refer to Figures 1-3 , the present invention provides a technical solution: an edge cloud computing load balancing method, and the steps of the method include:

[0049] Step 1: Build and train a load prediction model. The load prediction model is used to predict the load level of an edge computing node at a future time point based on the usage of the edge computing node. The specific steps for building and training the load prediction model are as follows:

[0050] Step 1.1: Collect historical load data. For an edge computing node, collect its historical load data at different load levels. This data includes all factors affecting the node load, including resource usage metrics, network status metrics, and user behavior data at different load levels. The resource usage metrics include CPU usage rate, memory usage rate, disk I / O, storage space usage rate, number of processes, and number of threads. The network status metrics include network bandwidth usage rate, latency, packet loss rate, number of connections, and network traffic. The user behavior data includes the number of user requests, user access patterns, user session duration, user geographical location distribution, and user device types.

[0051] Step 1.2: Data cleaning. Clean the collected historical load data to handle missing values and outliers, including filling missing values, smoothing outliers, etc., to ensure the integrity and consistency of the data.

[0052] For outliers, use the threshold judgment method. Consider data points outside the preset threshold range as outliers and replace them with the mean of their neighboring points.

[0053] Step 1.3: Data standardization. Standardize the cleaned feature data to eliminate the dimensionality differences between different features and improve the training effect of the model. Use the Z-score standardization method to subtract the mean of each feature value and divide it by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for standardization is: where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

[0054] Step 1.4: Feature selection. Use a feature selection algorithm to screen the features useful for the load prediction model and form a feature dataset, aiming to reduce the data dimension, improve the training efficiency and prediction accuracy of the model.

[0055] Step 1.5: Build a training dataset. Mark the load level of each piece of data and form a training dataset together with the corresponding feature data. Further divide the training dataset into a training set and a validation set to evaluate the performance of the model during the training process.

[0056] Step 1.6. Select and initialize the LSTM model: Select the Long Short-Term Memory (LSTM) network model as the load prediction model and initialize the parameters of the model. The LSTM model is suitable for load prediction because it is good at processing time series data.

[0057] Step 1.7. Model training: Use the training set data to iteratively train the LSTM model. During the training process, regularly evaluate the model performance using the validation set until the model performance meets the requirements, aiming to ensure that the model has good generalization ability and prediction accuracy.

[0058] Step 1.8. Model deployment: After training, save the weights and biases of the LSTM model and deploy the model. The deployed model can be used to predict the load level of edge computing nodes in the future.

[0059] Step 2. Deploy a monitoring and data collection system

[0060] Collect real-time data on all factors affecting the load of edge computing nodes to provide real-time data support for subsequent load prediction and resource scheduling.

[0061] Step 3. Data cleaning

[0062] Clean the real-time data collected in Step 2, and process the missing values and outliers in it to ensure the accuracy and reliability of the data, which helps to improve the accuracy of subsequent load prediction.

[0063] Step 4. Load prediction

[0064] Input the real-time data cleaned in Step 3 into the load prediction model trained in Step 1 to obtain the load level of the edge computing node at a future time point. Through accurate load prediction, it provides support for subsequent resource scheduling.

[0065] Step 5. Develop a resource scheduling strategy

[0066] According to the load level obtained in Step 4, dynamically develop a resource scheduling strategy. The scheduling strategy includes the number of edge computing nodes required to bear the load and the specific allocation plan of the nodes, aiming to ensure the maximization of resource utilization and the optimization of system operation while meeting the load requirements.

[0067] Step 6. Implement the scheduling strategy

[0068] Implement the scheduling strategy developed in Step 5 in the load balancer to achieve dynamic and efficient utilization of edge computing nodes. The last step is the final link of resource scheduling. By implementing the scheduling strategy, real-time, dynamic, and efficient resource allocation and management of edge computing nodes can be achieved.

[0069] The present invention will be further described below in conjunction with Embodiments 1 to 3:

[0070] Embodiment 1:

[0071] In step 1.4, the filtering method is adopted for feature selection. The filtering method (Filter Method) is a feature selection technique that is independent of any subsequent machine learning algorithm. In the implementation of the present invention, the filtering method is used to preliminarily screen those features that may be correlated with the target variable (i.e., the load level). This method is based on the statistical relationship or correlation between the feature and the target variable, and can quickly remove those features that are irrelevant or weakly correlated with the target variable, thereby reducing the dimensionality of the feature space and improving the efficiency of subsequent model training.

[0072] To determine the strength of the correlation between each feature and the target variable, the present invention uses the Pearson Correlation Coefficient as a measurement index. The Pearson Correlation Coefficient is a widely used statistic that can quantify the linear relationship between two variables. In the context of the present invention, features (such as CPU usage rate, memory usage rate, etc.) are regarded as one variable, while the target variable is the load level of the edge computing node. By calculating the Pearson Correlation Coefficient between each feature and the target variable, it can be evaluated which features play an important role in predicting the load level. The calculation formula of the Pearson Correlation Coefficient is:

[0073]

[0074] where r represents the correlation coefficient between the feature and the target variable, n represents the number of samples, x represents the feature value, and y represents the target variable value.

[0075] This formula calculates the covariance between the feature and the target variable and divides it by the standard deviations of the two variables, thereby obtaining a value between -1 and 1. This value represents the strength and direction of the linear correlation between the feature and the target variable. A value close to 1 or -1 indicates a strong correlation, while a value close to 0 indicates a weak correlation or no correlation. In the implementation of the present invention, this formula is used to calculate the Pearson Correlation Coefficient between each feature and the target variable, and based on these coefficients, features useful for the load prediction model are selected.

[0076] After calculating the Pearson correlation coefficient between each feature and the target variable, the next step is to select useful features based on these correlation coefficients. Specifically, a correlation coefficient threshold is set, and then those features whose absolute value is greater than this threshold are selected. The purpose of this step is to screen out features that have a strong linear correlation with the target variable, and these features are considered to make a significant contribution to the accuracy of the load prediction model. By removing those features with weak correlations, the complexity of the feature space can be reduced, and the training efficiency and generalization ability of the model can be improved. Finally, the selected useful features will form a feature subset for subsequent load prediction model training.

[0077] After determining the useful feature subset, the next step is to use these features to form a feature dataset. This feature dataset will be used as input data for training the load prediction model. When forming the feature dataset, it is necessary to ensure the integrity and consistency of the data to provide a high-quality data basis for model training. The feature dataset should contain the numerical values of all selected features, and these numerical values should correspond to the numerical values of the target variable. By using the selected feature subset to form a feature dataset, an optimized and low-dimensional feature space can be provided for subsequent model training, thereby improving the accuracy and efficiency of the load prediction model.

[0078] Example 2:

[0079] As a variant of the Recurrent Neural Network (RNN), the LSTM model is particularly good at processing time series data. In load prediction, historical load data usually exhibits significant time series characteristics, and LSTM can capture the long-term dependencies and periodic patterns in these time series, thereby improving the accuracy of prediction. The load situation in the edge computing environment is often affected by various dynamic factors, such as node performance, network conditions, user behavior, etc. The LSTM model can update its internal state in real time, adjust the prediction results according to the latest input data, so as to adapt to these dynamic changes and improve the real-time performance and accuracy of prediction.

[0080] In load prediction, in addition to historical load data, it may also involve various external factors such as weather, holidays, economic indicators, etc. The LSTM model can fuse these multi-source heterogeneous data into the model by introducing an additional input layer, and jointly act on the prediction results, thereby improving the robustness and comprehensiveness of the prediction.

[0081] The load prediction model is used to predict the load level of an edge computing node at a future time point according to the usage of the edge computing node. For example, it can collect in real time all the factor data that affect the load of the edge computing node, input these real-time data into the deployed load prediction model, and obtain the prediction result of the load level of the node at a future time point (such as 5 minutes, 10 minutes, or 30 minutes later).

[0082] The specific steps for initializing the LSTM model include:

[0083] i) Initialize the structural parameters of the LSTM model. The structural parameters include the feature dimension of the input layer, which matches the number of features extracted that affect the load of the edge computing node. The number of features is determined based on the collected historical load data, which includes resource usage metrics (such as CPU usage rate, memory usage rate, etc.), network status metrics (such as network bandwidth usage rate, latency, etc.), and user behavior data (such as the amount of user requests, user access patterns, etc.). The number of hidden layers and the number of hidden units in each layer determine the complexity and learning ability of the model; the number of nodes in the output layer corresponds to the dimension of the predicted load level.

[0084] ii) Initialize the model weights and biases. To avoid the model falling into the problem of gradient vanishing or explosion in the initial stage of training, small random numbers are used to initialize the weights and biases.

[0085] iii) Set the activation functions of the LSTM model. To achieve the non-linear transformation of the gating mechanism, the Sigmoid function is selected; and to calculate the candidate memory unit and the hidden state, the Tanh function is selected.

[0086] iv) Configure the optimization algorithm and set an appropriate learning rate parameter to control the update step size of the model during training.

[0087] In each iteration, the following steps are executed:

[0088] i) Feed a batch of input data into the LSTM model. The data contains the feature information extracted from the historical load data, such as resource usage metrics, network status metrics, and user behavior data.

[0089] ii) Update the memory unit and the hidden state by calculating the values of the input gate, forget gate, and output gate, and finally obtain the predicted output result.

[0090] iii) Calculate the value of the loss function based on the predicted output and the true labels to evaluate the performance of the model on the current batch of data.

[0091] iv) Calculate the gradients of the loss function with respect to the model parameters through the backpropagation algorithm. These gradients indicate how to adjust the model parameters to reduce the loss.

[0092] v) Use the optimization algorithm to update the model weights and biases according to the gradients, thereby improving the prediction ability of the model.

[0093] During the training process, the model performance is evaluated regularly using a validation set (which also contains comprehensive historical load data of resource usage metrics, network status metrics, and user behavior data). If it is found that the performance of the model on the validation set no longer improves or there are signs of overfitting, the learning rate is adjusted as needed or the training is stopped early.

[0094] By implementing the above steps, the present invention can effectively initialize, train, and optimize the LSTM model for load prediction in edge cloud computing.

[0095] Example 3:

[0096] The method for dynamically formulating a resource scheduling policy is specifically as follows:

[0097] ① Adopt a threshold-based scheduling algorithm:

[0098] To achieve dynamic resource scheduling, the present invention presets multiple load thresholds, which are set based on experience or historical data, and divides the load levels into different ranges, with each range corresponding to a specific resource scheduling policy to ensure effective allocation and use of edge computing nodes at different load levels.

[0099] ② Compare the predicted load level with the preset load thresholds:

[0100] The present invention predicts the future load level through the LSTM model and compares this predicted value with the preset load thresholds. The purpose of the comparison is to determine which preset threshold range the currently predicted load level falls into. This step is crucial for selecting an appropriate resource scheduling policy because it determines how resources should be allocated at a specific load level.

[0101] ③ Select the corresponding resource scheduling policy:

[0102] After determining the threshold range to which the current load level belongs, further select the resource scheduling policy corresponding to this range. These policies are formulated based on historical data and experience and are aimed at achieving optimal resource allocation at different load levels. The policies may include increasing or decreasing the number of edge computing nodes, adjusting the processing capabilities of the nodes, etc., to ensure load balancing and improve the overall service quality.

[0103] ④ Calculate the required number of edge computing nodes:

[0104] The resource scheduling policy includes the number of edge computing nodes required to undertake the load. Use a linear mapping function to calculate the required number of nodes, and the calculation formula is:

[0105]

[0106] Among them, round() represents the rounding function, the predicted load level represents the load level obtained in step 5, the lowest threshold and the highest threshold respectively represent the preset load threshold range, and the maximum number of nodes and the minimum number of nodes respectively represent the maximum and minimum available edge computing node numbers.

[0107] Assume that the preset lowest load threshold is 100, the highest load threshold is 500, the minimum available edge computing node number is 5, and the maximum node number is 20. When the load level predicted by the LSTM model is 300, the number of required nodes is calculated according to the linear mapping function as follows:

[0108]

[0109] Therefore, when the predicted load level is 300, it is necessary to dynamically adjust the resource allocation to ensure that 13 edge computing nodes are available to achieve load balancing.

[0110] ⑤ Dynamically adjust the resource allocation of edge computing nodes:

[0111] Dynamically adjusting the resource allocation of edge computing nodes according to the calculated number of nodes means that when the load is high, more nodes will be added to process requests, and when the load is low, nodes will be reduced to save resources.

[0112] In step 2, in order to achieve accurate prediction of the load of edge computing nodes and dynamic resource scheduling, the present invention configures corresponding data acquisition modules for each key factor affecting the load of edge computing nodes.

[0113] For each key factor affecting the load of edge computing nodes, such as network bandwidth, processor utilization rate, memory utilization rate, storage I / O, etc., the present invention configures a dedicated data acquisition module, and the acquisition module uses a preset sampling frequency and time interval to capture and record the data of key factors in real time. The setting of the sampling frequency and time interval is based on the analysis of historical data and prediction requirements to ensure that the collected data can accurately reflect the real-time state of key factors.

[0114] In order to ensure that the collected data can be transmitted to the central data storage system in real time and accurately for subsequent processing and analysis, the present invention formulates a data transmission interface between the data acquisition module and the central data storage system. This interface is based on an efficient and stable data transmission protocol to ensure reliable data transmission. At the same time, the interface also considers the security and integrity of data, and adopts encryption and verification mechanisms to prevent data from being tampered with or lost during transmission.

[0115] Select or design a lightweight data acquisition protocol, which should have a concise data format and an efficient transmission mechanism to reduce the network load and computational overhead during the data acquisition process. Existing lightweight communication protocols such as MQTT (Message Queuing Telemetry Transport) or CoAP (Constrained Application Protocol) can be considered, or a more streamlined protocol can be customized according to actual requirements. For the selected lightweight protocol, define the data format and transmission rules between the data acquisition module and the central data storage system. The data format should be compact and easy to parse to reduce bandwidth consumption during data transmission. The transmission rules should clarify the data sending frequency, time interval, retransmission mechanism, etc. to ensure the reliability and integrity of the data. Taking the MQTT protocol as an example, the present invention can implement an MQTT client in the data acquisition module, publish the collected key factor data as MQTT messages to a specified topic, and the central data storage system acts as an MQTT broker to subscribe to these topics and receive the messages. The MQTT protocol is very suitable for lightweight data acquisition and transmission in edge computing environments due to its concise data format and efficient transmission mechanism.

[0116] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0117] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An edge cloud computing load balancing method, characterized in that The steps of the method include: Step 1: Construct and train a load prediction model using a neural network algorithm. The load prediction model is used to predict the load level of an edge computing node at a future time point based on the usage of the edge computing node. The specific steps for constructing and training the load prediction model include: Step 1.1: Collect historical load data of the edge computing node. The historical load data is all factor data that affects the node load at different load levels. Step 1.2: Clean the collected data to handle missing values and outliers. Step 1.3: Normalize the cleaned feature data. Step 1.4: Use a feature selection algorithm to screen out features useful for the load prediction model and form a feature data set. Step 1.5: Mark the load level of each piece of data and form a training data set together with the corresponding feature data. Further divide the training data set into a training set and a validation set. Step 1.6: Select a long short-term memory network (LSTM) model to construct the load prediction model and initialize the parameters of the model. Step 1.7: Use the training data set to iteratively train the LSTM model. During the training process, regularly use the validation set to evaluate the model performance until the model performance meets the requirements. Step 1.8: After training is completed, save the weights and biases of the LSTM model and deploy the model for subsequent sampling location prediction. Step 2: Deploy a monitoring and data collection system to collect in real time all factor data that affects the load of the edge computing node. Step 3: Clean the data collected in Step 2 to handle missing values and outliers. Step 4: Input the data cleaned in Step 3 into the load prediction model trained in Step 1 to obtain the load level of the edge computing node at a future time point. Step 5: Dynamically formulate a resource scheduling strategy according to the load level obtained in Step 4. The dynamic formulation of the resource scheduling strategy is to dynamically calculate the number of edge computing nodes required to bear the load. The specific method includes: using a linear mapping function to calculate the required number of nodes, and the calculation formula is: where round() represents the rounding function, the predicted load level represents the load level obtained in Step 4, the minimum threshold and the maximum threshold respectively represent the preset load threshold range, and the maximum number of nodes and the minimum number of nodes respectively represent the maximum and minimum available edge computing node numbers. Step 6: Implement the scheduling strategy in the load balancer to achieve dynamic and efficient utilization of the edge computing nodes.

2. The edge cloud computing load balancing method according to claim 1, wherein: The historical load data collected for training the load prediction model includes the resource usage metrics, network status metrics, and user behavior data of the node at different load levels. The resource usage metrics include CPU usage rate, memory usage rate, disk I / O, storage space usage rate, number of processes, and number of threads. The network status metrics include network bandwidth usage rate, latency, packet loss rate, number of connections, and network traffic. The user behavior data includes the number of user requests, user access patterns, user session duration, user geographical location distribution, and user device types.

3. The edge cloud computing load balancing method according to claim 2, wherein: In step 1.2, for missing values, a linear interpolation algorithm is used for filling; for outliers, a threshold judgment method is adopted, and data points outside the preset threshold range are regarded as outliers and replaced with the mean value of their neighboring points.

4. The edge cloud computing load balancing method according to claim 3, wherein: In step 1.3, using the Z-score normalization method, subtract the mean of each eigenvalue from it and divide by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for the normalization process is: Z = where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the normalized data.

5. The edge cloud computing load balancing method according to claim 4, wherein In step 1.4, a filtering method is used for feature selection. This method evaluates the importance of features by calculating the correlation or statistical metrics between each feature and the target variable; the Pearson correlation coefficient is used to measure the linear correlation between each feature and the target variable, and the calculation formula of the Pearson correlation coefficient is: where r represents the correlation coefficient between the feature and the target variable, n represents the number of samples, x represents the feature value, and y represents the target variable value; According to the calculated correlation coefficient, features with absolute values greater than the set threshold are selected as useful features to form a feature dataset for the subsequent training of the load prediction model.

6. The edge cloud computing load balancing method according to claim 5, characterized in that, The specific steps for initializing the LSTM model include: i) Initialize the structural parameters of the LSTM model, including the feature dimension of the input layer, the number of hidden layers and the number of hidden units in each layer, and the number of nodes in the output layer. Among them, the feature dimension of the input layer matches the number of features affecting the load of the edge computing node, and the number of nodes in the output layer corresponds to the dimension of the predicted load level; ii) Initialize the model weights and biases, and use small random numbers for initialization; iii) Set the activation functions of the LSTM model, including setting the Sigmoid function for the gating mechanism and setting the Tanh function for the calculation of the candidate memory unit and the hidden state; iv) Configure the optimization algorithm and set the learning rate parameter; Use the training set data to perform iterative training on the LSTM model. The specific method is: In each iteration, a batch of input data is fed into the LSTM model. By calculating the values of the input gate, forget gate, and output gate, the memory unit and the hidden state are updated, and finally the predicted output result is obtained; calculate the loss function value according to the predicted output and the true label; calculate the gradient of the loss function with respect to the model parameters through the backpropagation algorithm; use the optimization algorithm to update the model weights and biases according to the gradient; during the training process, regularly evaluate the model performance using the validation set, and adjust the learning rate or stop training early as needed to prevent overfitting.

7. A method for edge cloud computing load balancing according to claim 1, characterized in that: In step 2, for each key factor affecting the load of the edge computing node, configure the corresponding data acquisition module, and use the preset sampling frequency and time interval to capture and record the data of the key factors in real time; And formulate the data transmission interface between the data acquisition module and the central data storage system, so that the collected data can be transmitted to the central data storage system in real time and accurately for subsequent processing and analysis; 8. The edge cloud computing load balancing method according to claim 7, wherein: The data acquisition module adopts a lightweight data acquisition protocol.

Images