Resource collaborative scheduling method in edge computing environment

Abstract

The invention belongs to the technical field of edge computing environment resource scheduling, and relates to a resource collaborative scheduling method in an edge computing environment. According to the method, the energy consumption model is constructed on the basis of the energy consumption data, nodes with low energy consumption under a specific load can be preferentially selected by predicting energy consumption and scheduling a system during task allocation, the energy consumption of the whole system is effectively reduced, and the energy consumption of the whole system is effectively reduced on the basis of historical power supply data and environmental data. The power supply capacity of renewable energy sources in several hours or even several days in the future is predicted, and the system is helped to carry out more accurate energy consumption management and task scheduling in a short time; based on the predicted energy consumption condition and the power supply capacity of the renewable energy, the system preferentially distributes tasks to nodes which are within the current power supply capacity range and have the lowest energy consumption and stable power supply. Therefore, the method achieves the efficient cooperative scheduling of resources in an edge computing environment, makes full use of renewable energy sources, and reduces the difficulty of energy consumption management caused by unstable power supply.

Description

Technical Field

[0001] The present application belongs to the technical field of resource scheduling in edge computing environments, and more specifically, to a resource collaborative scheduling method in an edge computing environment. Background Art

[0002] With the rapid development of the Internet of Things (IoT) and 5G communication technology, edge computing, as an emerging computing paradigm, has gradually become a key technology to solve the needs of large-scale data processing and low latency. In an edge computing environment, a large number of computing tasks are distributed to edge nodes close to the data source for processing, thereby reducing the delay of data transmission and improving the response speed of the system. However, edge computing node resources are limited, especially in terms of energy consumption, computing power and power supply, which puts higher requirements on resource coordinated scheduling.

[0003] In the traditional centralized computing model, computing tasks are mainly completed in cloud data centers. These data centers usually have strong computing capabilities and stable energy supply, and can improve energy efficiency by optimizing the scheduling of computing tasks. However, edge computing nodes are distributed at the edge of the network and may be located in geographically dispersed areas, and the energy consumption characteristics and power supply conditions of each node vary greatly. In this environment, relying solely on existing energy consumption management methods cannot effectively address the following challenges:

[0004] Resource heterogeneity: The computing power, storage resources, and network bandwidth of edge computing nodes are heterogeneous, which increases the complexity of task allocation and resource scheduling.

[0005] Energy supply fluctuations: Compared with data centers, the energy supply of edge computing nodes is more dependent on the local power grid. Especially in the scenario of using renewable energy, the instability of power supply further increases the difficulty of energy consumption management.

[0006] Dynamic changes in task load: The task load on edge computing nodes changes frequently, and traditional static scheduling strategies are difficult to adapt to such dynamic changes. Summary of the invention

[0007] The present invention provides a resource collaborative scheduling method in an edge computing environment, which is intended to solve the technical problem that the current instability of power supply exacerbates the difficulty of energy consumption management.

[0008] The resource collaborative scheduling method in the edge computing environment includes the following steps:

[0009] Step 1: Collect energy consumption data, workload data, and historical load data on edge computing nodes, as well as power supply data and environmental data;

[0010] Step 2: Build an energy consumption model based on the collected energy consumption data to predict the energy consumption of each edge computing node under different loads;

[0011] Step 3: Forecast the short-term and medium-term power supply capacity of renewable energy based on historical power supply data and environmental data;

[0012] Step 4: Allocate tasks on the edge computing nodes based on the predicted energy consumption and the short-term and medium-term power supply capabilities of renewable energy.

[0013] The present invention constructs an energy consumption model based on energy consumption data. By predicting energy consumption, the scheduling system can give priority to nodes with lower energy consumption under specific loads when allocating tasks, thereby effectively reducing the energy consumption of the overall system and alleviating the pressure of energy consumption management caused by unstable power supply. Based on historical power supply data and environmental data, the power supply capacity of renewable energy in the next few hours or even days is predicted to help the system perform more accurate energy consumption management and task scheduling in the short term. Based on the predicted energy consumption and the power supply capacity of renewable energy, the system will give priority to assigning tasks to nodes with the lowest energy consumption and stable power supply within the current power supply capacity. This can not only effectively reduce the overall energy consumption, but also give priority to the execution of key tasks when the energy supply is insufficient. Therefore, the present invention realizes efficient collaborative scheduling of resources in an edge computing environment, makes full use of renewable energy, and reduces the difficulty of energy consumption management caused by unstable power supply.

[0014] Preferably, the energy consumption model constructed in step 2 is a linear regression model, which is as follows:

[0015]

[0016] Where: E represents the predicted energy consumption; β0 represents the intercept term; β1 to β8 represent the regression coefficients of the corresponding independent variables; CPU usage Indicates CPU usage; Memory usage Indicates memory usage; Historical load Indicates historical load; Temperature indicates the temperature of the environment where the edge computing node is located; Network traffic Indicates network traffic; Concurrency indicates task concurrency; Priority indicates task priority; Data transfer Indicates the amount of data transmission, describing the amount of data transmission between nodes; Indicates the task type, a dummy variable for different task types; Represents the interaction effect between task types; Node state represents the current state of the node;∈ represents the error term; represents the sum of regression coefficients of task type; The sum of regression coefficients representing the interaction effects between task types.

[0017] Preferably, the loss function of the linear regression model is as follows:

[0018]

[0019] Where: w i represents the weight of the i-th sample; E i represents the actual energy consumption of the i-th sample; represents the predicted energy consumption of the i-th sample; n represents the total number of samples;

[0020] Add energy consumption constraint to the loss function:

[0021]

[0022] Where: λ represents the hyperparameter; represents the average energy consumption of all nodes; To ensure that the loss is only penalized when the node's energy consumption is higher than the average.

[0023] Preferably, the weight w of the sample i The calculation method is as follows:

[0024]

[0025] Where: α1, α2, α3 represent weight coefficients, satisfying α1+α2+α3=1; Priority i Indicates the priority of the i-th task; Represents the historical energy consumption of the i-th task; Concurrency i Indicates the concurrency of the i-th task.

[0026] Preferably, in step 3, an LSTM model is used to predict the short-term and medium-term power supply capacity of renewable energy.

[0027] Preferably, step 4 comprises the following steps:

[0028] Task priority determination: Determine the priority of each task based on its importance and urgency;

[0029] Renewable energy supply capacity: Based on the predicted short-term and medium-term renewable energy supply capacity of each node and combined with the current node power demand, the total power supply capacity of the node is evaluated;

[0030] Task and node matching: The matching conditions are as follows:

[0031] The current load of the node matches the task load; whether the available power of the node can support the task execution, the specific matching formula is as follows:

[0032]

[0033] Where: E j represents the current energy consumption of node j; E task,i represents the predicted energy consumption of task i; L j represents the total power supply capacity of node j; L task,i represents the load forecast value of task i;

[0034] Global optimal task allocation is achieved based on genetic algorithm.

[0035] Preferably, the optimization goal of achieving global optimal task allocation based on genetic algorithm is as follows:

[0036]

[0037] Among them C ij represents a binary variable, indicating whether the task can be assigned to node j. When the value is 1, it means that it can be assigned;

[0038] The constraints are as follows:

[0039]

[0040] Where: represents the maximum power supply capability of node j.

[0041] The beneficial effects of the present invention include:

[0042] The present invention constructs an energy consumption model based on energy consumption data. By predicting energy consumption, the scheduling system can give priority to nodes with lower energy consumption under specific loads when allocating tasks, thereby effectively reducing the energy consumption of the overall system and alleviating the pressure of energy consumption management caused by unstable power supply. Based on historical power supply data and environmental data, the power supply capacity of renewable energy in the next few hours or even days is predicted to help the system perform more accurate energy consumption management and task scheduling in the short term. Based on the predicted energy consumption and the power supply capacity of renewable energy, the system will give priority to assigning tasks to nodes with the lowest energy consumption and stable power supply within the current power supply capacity. This can not only effectively reduce the overall energy consumption, but also give priority to the execution of key tasks when the energy supply is insufficient. Therefore, the present invention realizes efficient collaborative scheduling of resources in an edge computing environment, makes full use of renewable energy, and reduces the difficulty of energy consumption management caused by unstable power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 The overall system architecture diagram provided for the embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the technical problems, technical solutions and beneficial effects to be solved by 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.

[0046] See also Figure 1 As shown, the best embodiment of the present invention is further described;

[0047] The resource collaborative scheduling method in the edge computing environment includes the following steps:

[0048] Step 1: Collect energy consumption data, workload data, and historical load data on edge computing nodes, as well as power supply data and environmental data;

[0049] Energy consumption data: power consumption of each edge computing node, power consumption of hardware resources such as CPU, GPU, and memory.

[0050] Workload data: CPU usage, memory usage, network traffic, task concurrency (i.e. the number of tasks running simultaneously), type and priority of current tasks.

[0051] Historical load data: The load status of the node over the past period of time (such as historical records of CPU usage, memory usage, network traffic, etc.).

[0052] Electricity supply data: actual electricity generation from renewable energy sources, electricity fluctuations.

[0053] Environmental data: temperature, humidity, light intensity, and wind speed of the physical environment where the node is located.

[0054] Step 2: Build an energy consumption model based on the collected energy consumption data to predict the energy consumption of each edge computing node under different loads;

[0055] The energy consumption model constructed in step 2 is a linear regression model, which is as follows:

[0056]

[0057] Where: E represents the predicted energy consumption; β0 represents the intercept term; β1 to β8 represent the regression coefficients of the corresponding independent variables; CPU usage Indicates CPU usage; Memory usage Indicates memory usage; Historical load Indicates historical load; Temperature indicates the temperature of the environment where the edge computing node is located; Network traffic Indicates network traffic; Concurrency indicates task concurrency; Priority indicates task priority; Data transfer Indicates the amount of data transmission, describing the amount of data transmission between nodes; Indicates the task type, a dummy variable for different task types; Represents the interaction effect between task types; Node state represents the current state of the node;∈ represents the error term; represents the sum of regression coefficients of task type; The sum of regression coefficients representing the interaction effects between task types.

[0058] Preferably, the loss function of the linear regression model is as follows:

[0059]

[0060] Where: w i represents the weight of the i-th sample; E i represents the actual energy consumption of the i-th sample; represents the predicted energy consumption of the i-th sample; n represents the total number of samples;

[0061] Add energy consumption constraint to the loss function:

[0062]

[0063] Where: λ represents the hyperparameter; represents the average energy consumption of all nodes; To ensure that the loss is only penalized when the node's energy consumption is higher than the average.

[0064] Preferably, the weight w of the sample i The calculation method is as follows:

[0065]

[0066] Where: α1, α2, α3 represent weight coefficients, satisfying α1+α2+α3=1; Priority i Indicates the priority of the i-th task; Represents the historical energy consumption of the i-th task; Concurrency i Indicates the concurrency of the i-th task.

[0067] In this embodiment, multiple key factors (such as CPU usage, memory usage, historical load, temperature, etc.) are included in the regression analysis, which can fully reflect the energy consumption of edge computing nodes under different load and environmental conditions. The regression coefficient of each variable represents the contribution of the factor to energy consumption, which makes the model have high explanatory power and flexibility, and can adapt to a variety of task scenarios and computing nodes.

[0068] The model introduces dummy variables of task type (Task_type) and interaction effect between tasks (Task_interaction), so that the model can capture the different effects of different task types on energy consumption and the interactions between tasks. Through the design of these variables, the model can not only predict the energy consumption of conventional tasks, but also cope with complex task combinations and interdependent task scenarios.

[0069] The model includes node state variables, which can reflect the impact of nodes in different states (such as high load, low load or idle) on energy consumption. This makes the model more dynamically adaptable and can adjust energy consumption predictions according to the actual operating environment, thereby optimizing resource scheduling.

[0070] By introducing weight coefficients, the model can perform weighted processing on data samples of different importance. The design of weights takes into account task priority, historical energy consumption, and task concurrency, so that important tasks or high-energy-consuming tasks have a greater weight in the model, thereby improving the prediction accuracy of the model. This is especially important for scheduling scenarios that need to ensure the efficient execution of key tasks.

[0071] Adding energy consumption constraints to the loss function can effectively prevent the problem of excessive node energy consumption. This constraint ensures that the model's loss is only penalized when the node energy consumption is higher than the average level, thereby encouraging the model to keep the node energy consumption within a reasonable range when optimizing energy consumption prediction. This design helps to reduce the energy consumption of the overall system and reduce the load of nodes with excessively high energy consumption, achieving the goal of optimizing resource allocation.

[0072] The hyperparameter λ in the loss function can be adjusted according to the actual application scenario to control the strength of the energy consumption constraint. This allows the model to flexibly adapt to different scheduling strategy requirements. For example, in energy-sensitive scenarios, the λ value can be increased to strengthen the energy consumption constraint, while in performance-first scenarios, the λ value can be reduced.

[0073] In summary, through multivariate analysis, the introduction of task types and interaction effects, the consideration of node status, the optimization of weighted least squares method, and the setting of energy consumption constraints, this linear regression model can provide accurate energy consumption prediction, support resource collaborative scheduling in edge computing environments, and achieve an optimal balance between system performance and energy consumption.

[0074] Step 3: Based on historical power supply data and environmental data, predict the short-term and medium-term power supply capacity of renewable energy; in step 3, the LSTM model is used to predict the short-term and medium-term power supply capacity of renewable energy.

[0075] Since it is relatively common in the art to predict the short-term and medium-term power supply capacity of renewable energy based on the LSTM model, the existing LSTM model is used in the embodiment to predict the short-term and medium-term power supply capacity;

[0076] LSTM (Long Short-Term Memory) is a deep learning model suitable for processing time series data. It consists of a series of memory cells, each of which contains three gating mechanisms: input gate, forget gate, and output gate. These gating mechanisms enable LSTM to effectively capture long-term and short-term dependencies in data.

[0077] LSTM cell structure

[0078] Input Gate: Controls the relevant information input to the current time step.

[0079] Forget Gate: determines which information is discarded from memory.

[0080] Output Gate: Determines the information to be output at the current time step.

[0081] To predict the power supply capacity of renewable energy, the LSTM model requires the following input data:

[0082] Historical electricity supply data

[0083] Content: Records the actual power supply data of renewable energy (such as solar energy, wind energy, etc.) over a period of time in the past.

[0084] Purpose: Helps the model learn historical power supply patterns and seasonal variations.

[0085] Environmental data

[0086] Content: Includes meteorological data, such as sunshine intensity, wind speed, temperature, humidity, etc.

[0087] Purpose: Provides information on environmental conditions directly related to renewable energy generation.

[0088] Data preprocessing

[0089] Standardization: Standardize the input data and convert it into a distribution with a mean of 0 and a variance of 1 to improve the training efficiency of the model.

[0090] Serialization: Convert the data into a time series form and construct input samples of multiple time steps for processing by the LSTM model.

[0091] Model training process:

[0092] Multiple LSTM layers are stacked to form a deep network structure to capture complex time series features.

[0093] Add a Dense layer as the output layer to output the predicted power supply capacity.

[0094] The mean squared error (MSE) is used as the loss function to minimize the difference between the predicted value and the true value.

[0095] Training process:

[0096] The processed input data is fed into the LSTM model and the weights of the model are updated using the back-propagation algorithm.

[0097] The model parameters are adjusted through multiple iterations (epochs) to optimize the loss function.

[0098] According to the performance of the model, the prediction accuracy of the model can be improved by adjusting hyperparameters such as learning rate, number of hidden layer units, batch size, etc.

[0099] Short-term forecast:

[0100] Input current environmental conditions and recent historical power supply data to predict power supply capabilities in the short term.

[0101] Medium-term forecast:

[0102] Based on historical trends and predicted short-term results, extend the forecast to medium-term power supply capacity to help develop longer-term dispatch plans.

[0103] Output:

[0104] Predicted power supply capacity: The output of the LSTM model is the predicted value of the power supply capacity of renewable energy in the short and medium term.

[0105] Uncertainty assessment: By analyzing the uncertainty of forecast results, it helps decision makers consider potential risks when making resource scheduling plans.

[0106] Through the application of the LSTM model, the power supply capacity of renewable energy can be accurately predicted under complex environmental conditions, the scheduling and allocation of resources can be optimized, and the overall efficiency and reliability of the system can be improved.

[0107] Step 4: Allocate tasks on the edge computing nodes based on the predicted energy consumption and the short-term and medium-term power supply capabilities of renewable energy.

[0108] The step 4 comprises the following steps:

[0109] Task priority determination: Determine the priority of each task based on its importance and urgency;

[0110] Renewable energy supply capacity: Based on the predicted short-term and medium-term renewable energy supply capacity of each node and combined with the current node power demand, the total power supply capacity of the node is evaluated;

[0111] Task and node matching: The matching conditions are as follows:

[0112] The current load of the node matches the task load; whether the available power of the node can support the task execution, the specific matching formula is as follows:

[0113]

[0114] Where: E j represents the current energy consumption of node j; E task,i represents the predicted energy consumption of task i; L j represents the total power supply capacity of node j; L task,i represents the load forecast value of task i;

[0115] The load prediction value L of task i is task,i Make predictions based on the LSTM model;

[0116] L task,i =LSTM(Historical load ,Current load );

[0117] Among them: Historical load Represents historical load data; Current load Indicates the current load data;

[0118] In this embodiment, the future load data of the task is predicted based on the historical load data and current load data of the task.

[0119] The LSTM model can capture the dynamic changes of load and provide accurate load forecasts for the system, thereby optimizing resource utilization and task allocation.

[0120] Global optimal task allocation is achieved based on genetic algorithm.

[0121] The optimization objectives of achieving global optimal task allocation based on genetic algorithm are as follows:

[0122]

[0123] Among them C ij represents a binary variable, indicating whether the task can be assigned to node j. When the value is 1, it means that it can be assigned;

[0124] The constraints are as follows:

[0125]

[0126] Where: represents the maximum power supply capability of node j.

[0127] The specific steps of target optimization based on genetic algorithm are as follows:

[0128] Initialize the population:

[0129] Chromosome representation: The task allocation scheme is represented as a chromosome. Each chromosome is an array of length N, and each gene in the array represents the node index to which the task is assigned.

[0130] For example, chromosome [2,1,3,4] means that task 1 is assigned to node 2, task 2 is assigned to node 1, task 3 is assigned to node 3, and task 4 is assigned to node 4.

[0131] Population generation: Randomly generate an initial population consisting of P chromosomes. This population covers the diversity of possible task allocation schemes.

[0132] Fitness function design:

[0133] Fitness function: The fitness function is used to evaluate the quality of each chromosome. The fitness function is defined as the inverse of the objective function;

[0134] Select an action:

[0135] Selection method: Use Roulette Wheel Selection to select chromosomes with higher fitness from the population as the parents of the next generation.

[0136] Selection steps:

[0137] Calculate the fitness value of each chromosome.

[0138] Chromosomes are selected according to the ratio of fitness values. Chromosomes with higher fitness values ​​have a greater probability of being selected.

[0139] Crossover operation:

[0140] Crossover method: Use the two-point crossover method to generate new individuals.

[0141] Crossover steps:

[0142] Randomly select two chromosomes as parents.

[0143] One or two crossover points are randomly selected, and the gene segments of the parents are exchanged at the crossover points to generate two new offspring.

[0144] After the crossover, the offspring inherits some of the characteristics of the parent generation and forms a new task allocation scheme.

[0145] Mutation operation:

[0146] Mutation method: Use random mutation method to make small random adjustments to the genes in the chromosome.

[0147] Mutation steps:

[0148] A chromosome in the population is selected with a small probability (usually between 0.01 and 0.1).

[0149] Randomly select one or more genes in the chromosome and randomly change the gene value so that the corresponding tasks are assigned to different nodes.

[0150] The mutated chromosomes may provide new and unexplored task allocation schemes, making it possible to find better solutions.

[0151] New generation of population:

[0152] The offspring generated by crossover and mutation operations together with the parents form a new generation of population.

[0153] Elite retention strategy: In order to avoid the optimal solution being lost during the evolution process, the chromosome with the highest fitness in the current population can be directly copied to the next generation.

[0154] Termination conditions:

[0155] Maximum number of iterations: Set the maximum number of iterations. When the set number of iterations is reached, the algorithm terminates.

[0156] Convergence: If the optimal solution of several consecutive generations does not change significantly, it means that the algorithm may have converged and can be terminated early.

[0157] Final output:

[0158] When the algorithm terminates, the chromosome with the highest fitness in the current population is output. The task allocation plan represented by this chromosome is the optimized global optimal task allocation plan.

[0159] The present invention constructs an energy consumption model based on energy consumption data. By predicting energy consumption, the scheduling system can give priority to nodes with lower energy consumption under specific loads when allocating tasks, thereby effectively reducing the energy consumption of the overall system and alleviating the pressure of energy consumption management caused by unstable power supply. Based on historical power supply data and environmental data, the power supply capacity of renewable energy in the next few hours or even days is predicted to help the system perform more accurate energy consumption management and task scheduling in the short term. Based on the predicted energy consumption and the power supply capacity of renewable energy, the system will give priority to assigning tasks to nodes with the lowest energy consumption and stable power supply within the current power supply capacity. This can not only effectively reduce the overall energy consumption, but also give priority to the execution of key tasks when the energy supply is insufficient. Therefore, the present invention realizes efficient collaborative scheduling of resources in an edge computing environment, makes full use of renewable energy, and reduces the difficulty of energy consumption management caused by unstable power supply.

[0160] The above are only preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A resource collaborative scheduling method in an edge computing environment, characterized in that: The following steps are involved: Step 1: Collect energy consumption data, workload data, and historical load data on edge computing nodes, as well as power supply data and environmental data; Step 2: Build an energy consumption model based on the collected energy consumption data to predict the energy consumption of each edge computing node under different loads; Step 3: Forecast the short-term and medium-term power supply capacity of renewable energy based on historical power supply data and environmental data; Step 4: Allocate tasks on the edge computing nodes based on the predicted energy consumption and the short-term and medium-term power supply capabilities of renewable energy.

2. The resource collaborative scheduling method in the edge computing environment according to claim 1 is characterized in that: The energy consumption model constructed in step 2 is a linear regression model, which is as follows: Where: E represents the predicted energy consumption; β0 represents the intercept term; β1 to β8 represent the regression coefficients of the corresponding independent variables; CPU usage Indicates CPU usage; Memory usage Indicates memory usage; Historical load Indicates historical load; Temperature indicates the temperature of the environment where the edge computing node is located; Network traffic Indicates network traffic; Concurrency indicates task concurrency; Priority indicates task priority; Data transfer Indicates the amount of data transmission, describing the amount of data transmission between nodes; Indicates the task type, a dummy variable for different task types; Represents the interaction effect between task types; Node state represents the current state of the node;∈ represents the error term; represents the sum of regression coefficients of task type; The sum of regression coefficients representing the interaction effects between task types.

3. The resource collaborative scheduling method in the edge computing environment according to claim 2 is characterized in that: The loss function of the linear regression model is as follows: Where: w i represents the weight of the i-th sample; E i represents the actual energy consumption of the i-th sample; represents the predicted energy consumption of the i-th sample; n represents the total number of samples; Add energy consumption constraint to the loss function: Where: λ represents the hyperparameter; represents the average energy consumption of all nodes; To ensure that the loss is only penalized when the node's energy consumption is higher than the average.

4. The resource collaborative scheduling method in the edge computing environment according to claim 3 is characterized in that: The weight w of the sample i The calculation method is as follows: Where: α1, α2, α3 represent weight coefficients, satisfying α1+α2+α3=1; Priority i Indicates the priority of the i-th task; Represents the historical energy consumption of the i-th task; Concurrency i Indicates the concurrency of the i-th task.

5. The resource collaborative scheduling method in the edge computing environment according to claim 1, characterized in that: In step 3, the LSTM model is used to predict the short-term and medium-term power supply capacity of renewable energy.

6. The resource collaborative scheduling method in the edge computing environment according to claim 1, characterized in that: The step 4 comprises the following steps: Task priority determination: Determine the priority of each task based on its importance and urgency; Renewable energy supply capacity: Based on the predicted short-term and medium-term renewable energy supply capacity of each node and combined with the current node power demand, the total power supply capacity of the node is evaluated; Task and node matching: The matching conditions are as follows: The current load of the node matches the task load; whether the available power of the node can support the task execution. The specific matching formula is as follows: Where: E j represents the current energy consumption of node j; E task,i represents the predicted energy consumption of task i; L j represents the total power supply capacity of node j; L task,i represents the load forecast value of task i; Global optimal task allocation is achieved based on genetic algorithm.

7. The resource collaborative scheduling method in the edge computing environment according to claim 6 is characterized in that: The optimization objectives of achieving global optimal task allocation based on genetic algorithm are as follows: Among them C ij represents a binary variable, indicating whether the task can be assigned to node j. When the value is 1, it means that it can be assigned; The constraints are as follows: Where: represents the maximum power supply capability of node j.

Images