Calculation network energy integrated collaborative task scheduling platform and scheduling method

Abstract

The invention relates to the technical field of computing power distribution, in particular to a computing network energy integrated collaborative task scheduling platform and scheduling method. The scheduling platform comprises an infrastructure layer, a management and control layer and an application service layer. The infrastructure layer is composed of a computing power supplier, a network supplier and an energy supplier; the application service layer at least comprises a user platform, and the user platform is at least used for receiving and analyzing task requests initiated by all users and reporting the task requests to the management and control layer; the management and control layer at least comprises a sensing module, a measurement module and a computing network energy task scheduling module, the sensing module is used for sensing computing power resources, transmitting network resources and energy resources, and the measurement module is used for measuring the computing power resources, the network resources and the energy resources required for processing each task; and the calculation network energy task scheduling module is used for selecting an optimal calculation power supplier and an optimal network supplier of each task based on the total time delay, the total energy consumption and the total carbon emission. According to the platform, the computing resource utilization rate and the task processing speed can be improved, and carbon emission can be effectively reduced.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computing power allocation technology, and in particular to a computing-network-integrated collaborative task scheduling platform and scheduling method. Background Art

[0002] In the context of the current large-scale deployment of computing infrastructure such as edge computing, artificial intelligence data centers, and supercomputing data centers, it has become more convenient and faster for end users to access and use computing resources. How to efficiently use these resources has become an urgent problem to be solved. However, with the continuous advancement of science and technology and the development of computing networks, the total amount of data in society has exploded, and the demand for data resource storage, computing, and application has increased significantly. The energy consumption and carbon emissions of the computing network itself have increased sharply. For this reason, future development needs to adopt more energy-saving and environmentally friendly computing and storage technologies to ensure the intelligent combination of efficient computing and energy conservation of the computing network.

[0003] However, existing computing power scheduling technologies such as crystal network technology and CEFT-LB scheduling algorithm not only have low computing resource utilization and service speed, but also do not take energy conservation into consideration, which easily leads to waste of computing power resources and energy.

[0004] Therefore, there is an urgent need for a computing-network-integrated collaborative task scheduling platform and scheduling method. Summary of the invention

[0005] In order to solve the problem that the existing computing power scheduling technology not only has low computing resource utilization and service speed, but also does not take energy conservation into consideration, which easily causes waste of computing power resources and energy, the embodiments of the present invention provide a computing-network-energy integrated collaborative task scheduling platform and scheduling method.

[0006] In a first aspect, an embodiment of the present invention provides a computing-network-energy integrated collaborative task scheduling platform, the platform comprising: an infrastructure layer, a management and control layer, and an application service layer;

[0007] The infrastructure layer is composed of computing power suppliers, network suppliers and energy suppliers, which are used to provide computing power resources, transmission network resources and energy resources; wherein the energy includes clean energy and thermal energy;

[0008] The application service layer at least includes a user platform, which is used to manage users, receive and analyze task requests initiated by each user, and report tasks to the control layer;

[0009] The management and control layer includes at least a perception module, a measurement module and a computing network energy task scheduling module. The perception module is used to perceive the computing power resources, transmission network resources and energy resources that each computing power supplier, network supplier and energy supplier can provide. The measurement module is used to measure the computing power resources, network resources and energy resources required to process each task. The computing network energy task scheduling module is used to select the optimal computing power supplier and the optimal network supplier for each task based on the total delay, total energy consumption and total carbon emissions; wherein the total delay includes computing delay and communication delay, the total energy consumption includes computing energy consumption and communication energy consumption, and the total carbon emissions include computing carbon emissions and communication carbon emissions.

[0010] In a second aspect, an embodiment of the present invention further provides a scheduling method based on the platform described in any embodiment of the specification, the method comprising:

[0011] The user platform of the application service layer receives and analyzes the task requests initiated by each user, and reports the tasks to the management and control layer;

[0012] The control layer uses the measurement module to measure the computing resources, network resources and energy resources required to process each task;

[0013] The management and control layer uses the perception module to perceive the computing resources, transmission network resources and energy resources that can be provided by each computing power supplier, network supplier and energy supplier in the infrastructure layer;

[0014] The management and control layer uses the computing network energy task scheduling module to select the optimal computing power supplier and the optimal network supplier for each task based on the total delay, total energy consumption and total carbon emissions; wherein the total delay includes computing delay and communication delay, the total energy consumption includes computing energy consumption and communication energy consumption, and the total carbon emissions include computing carbon emissions and communication carbon emissions.

[0015] The embodiments of the present invention provide a computing, network and energy integrated collaborative task scheduling platform and scheduling method. The scheduling platform has the distributed resource management and decision-making capabilities of the network between computing power suppliers and energy suppliers. By considering the total delay, total energy consumption and total carbon emissions of processing and transmission tasks, the optimal computing power supplier and the optimal network supplier are selected in complex scenarios to realize the scheduling and processing of user tasks. This can not only improve the computing resource utilization and task processing speed, but also effectively reduce carbon emissions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, 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 some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a schematic diagram of the composition of a computing-network-energy integrated collaborative task scheduling platform provided by an embodiment of the present invention;

[0018] Figure 2 It is a flow chart of a computing-network-energy integrated collaborative task scheduling method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] Please refer to Figure 1 ,The embodiment of the present invention provides a computing network energy integrated collaborative task scheduling platform, including: an infrastructure layer, a management and control layer, and an application service layer;

[0021] The infrastructure layer is composed of computing power suppliers, network suppliers, and energy suppliers, which are used to provide computing power resources, transmission network resources, and energy resources; among them, energy includes clean energy and thermal energy;

[0022] The application service layer includes at least a user platform, which is used to manage users, receive and analyze task requests initiated by each user, and report tasks to the management and control layer;

[0023] The management and control layer includes at least a perception module, a measurement module and a computing network energy task scheduling module. The perception module is used to perceive the computing power resources, transmission network resources and energy resources that each computing power supplier, network supplier and energy supplier can provide. The measurement module is used to measure the computing power resources, network resources and energy resources required to process each task. The computing network energy task scheduling module is used to select the optimal computing power supplier and the optimal network supplier for each task based on the total delay, total energy consumption and total carbon emissions; among which, the total delay includes computing delay and communication delay, the total energy consumption includes computing energy consumption and communication energy consumption, and the total carbon emissions include computing carbon emissions and communication carbon emissions.

[0024] In some implementations, the application service layer further includes: a green computing energy trading module, a green computing energy operation module, and an intelligent operation and maintenance module;

[0025] The green computing energy trading module is used to enable users to purchase computing power resources and transmission network resources based on the optimal computing power supplier and optimal network supplier output by the computing network energy task scheduling module, and to enable computing power suppliers and network suppliers to purchase energy from energy suppliers;

[0026] The green computing energy operation module is used for computing energy and energy operation management planning;

[0027] The intelligent operation and maintenance module is used to perform network operation, management and maintenance.

[0028] In the embodiment of the present invention, the scheduling platform has the distributed resource management and decision-making capabilities of the network between the computing power supplier and the energy supplier, which can solve the problems of complex data service processes and a large amount of redundant traffic between the end users and the data center in the traditional data center. By considering the total delay, total energy consumption and total carbon emissions of processing and transmission tasks, the optimal computing power supplier and the optimal network supplier are selected in complex scenarios to realize the scheduling and processing of user tasks, which can not only improve the computing resource utilization and task processing speed, but also effectively reduce carbon emissions.

[0029] In this embodiment, if Figure 1 As shown in FIG. 1 , it is mainly composed of six main parts, namely, users, computing power suppliers, network suppliers, computing network energy task scheduling module, green computing energy trading module and energy suppliers. represents a collection of users, represents the set of computing power providers, and Represents the set of network suppliers. Users complete their respective tasks by purchasing computing power services and network services according to the scheduling results given by the computing network energy task scheduling module. These tasks may have different QoS requirements such as latency and network reliability. Computing power suppliers refer to the parties that provide computing power services to the platform. They purchase different proportions of electricity from the green computing energy trading module, including clean renewable energy such as photovoltaics, wind power, and non-renewable energy such as thermal power, to build their own energy proportion. This clean energy proportion is directly related to the sustainability and environmental protection of the platform. Network suppliers refer to the parties that provide network services to the platform. They also purchase electricity through the green computing energy trading module to ensure the efficient operation and reliability of the network. The quality of network services is crucial to the user experience, especially in tasks with strict requirements on latency and network reliability. The green computing energy trading module acts as a bridge between computing network suppliers and energy suppliers, providing computing power suppliers and network suppliers with a variety of electricity options, including clean renewable energy such as photovoltaics, wind power, and thermal power generation energy. Through energy trading, the platform can optimize the use of green energy while reducing dependence on traditional energy and promoting sustainable development. The computing network task scheduling module is the brain of the platform. It analyzes user task requirements and current resource conditions to perform intelligent resource scheduling to efficiently complete tasks. This not only improves the overall performance of the platform, but also helps achieve the goal of green computing and makes the platform more energy-efficient and environmentally friendly.

[0030] Next, the computing network energy task scheduling module is described in detail.

[0031] In some implementations, the network energy task scheduling module contains a carbon emission model, a task model, a delay model, an energy consumption model, and an optimization algorithm.

[0032] (1) Carbon emission model

[0033] The carbon emission model contains the energy proportion of each computing power supplier and the energy proportion of each network supplier; among them, the energy proportion includes the clean energy proportion and the thermal energy proportion, and the sum of the clean energy proportion and the thermal energy proportion is 1.

[0034] In this embodiment, the energy ratio of the computing power supplier and the network supplier includes clean energy and thermal energy. Clean energy produces almost no direct carbon emissions during the power generation process, while thermal power generation produces a large amount of carbon dioxide. The energy ratios of different suppliers are different. Specifically, the computing power supplier and the network supplier can formulate their respective energy ratios by purchasing energy from the green computing energy trading module and the power grid. The choice of energy ratio will directly affect the supplier's impact on the environment during the production and operation process, resulting in different carbon emissions from different suppliers when processing tasks. Assume that the energy consumption carbon emission coefficient (carbon dioxide emissions generated per unit of energy consumption) α represents the carbon emissions generated per unit of thermal energy. For the computing power supplier Use and Indicates the proportion of thermal energy and clean energy, and Likewise, and Network provider The proportion of thermal energy and clean energy,

[0035] (2) Task Model

[0036] Each task in the task model consists of a tuple, which stores the total amount of data and the number of cycles required for the task. A task can only be processed by one computing power provider.

[0037] In this embodiment, considering the video processing tasks in entertainment, monitoring, medical and other scenarios, the user hopes to use distributed computing resources to process his video analysis tasks. The video processing task is T u , T u By a binary Composition, of which Indicates the total amount of data for this task, in MB; is the number of cycles required to process this task. By default, a user's task cannot be split, that is, a task can only be processed by one computing power provider, and a computing power provider can process multiple tasks at the same time. It is assumed that the task arrival interval follows a negative exponential distribution with parameter λ. Since the output data size of task processing is much smaller than the input data, the delay and energy consumption of transmitting the task processing results back to the user can be ignored.

[0038] (3) Delay Model

[0039] The delay model includes a computing delay model and a communication delay model; among them, the communication delay model is used to calculate the wireless transmission delay of the task from the user to the base station and the wired transmission delay of the task from the base station to the computing power supplier via the computing power network edge gateway. The wired transmission link represents the network supplier, and the result data obtained by the computing power supplier after processing the task is small, and the delay of transmitting the result back to the user is ignored.

[0040] In this embodiment, the route from the user initiating the task to the task being returned to the user after it is processed is as follows: the user initiates the task and transmits it wirelessly to the base station; the task is transmitted from the base station to the computing power network edge gateway via a wired link of a network provider; the task is transmitted from the computing power network edge gateway to the computing power provider via a wired link of the same network provider; the computing power provider processes the task, and the task result is returned to the user with negligible delay.

[0041] In the embodiment of the present invention, the delay calculation model is:

[0042]

[0043] In the formula, Processing task T for computing power provider m u The time required, For task T u The total amount of data, Assign computing power provider m to task T u The computing power ratio of this ratio ranges from 0 to 1. If A low value of means that only a small part of the computing power of the computing power supplier m is allocated to task T u , which will increase the processing time. On the contrary, if A higher value of indicates that more computing power is allocated to task T. u , the processing time will be reduced. m is the CPU frequency of computing power provider m.

[0044] The wireless transmission delay of the task from the user to the base station in the communication energy consumption model is calculated as follows:

[0045]

[0046] In the formula, is the wireless transmission delay of the task from the user to the base station, For task T u The total amount of data, R u is the uplink transmission rate, β is the actual available bandwidth ratio of the user, 0≤β≤1, when β=1, it means that there is no other user sharing the bandwidth, B is the uplink channel bandwidth, p is the user's transmission power, h uis the fading power gain between the user and the base station, σ 2 represents Gaussian noise power, γ is the influence coefficient of interference on signal transmission, and when γ=1, it means there is no interference.

[0047] In this embodiment, it is assumed that the wireless channel between the user and the base station is an independent, identically distributed block fading channel, the positions of the user and the base station are fixed when transmitting the result, and the channel is static in each time slot. If the channel sharing and interference between different users are considered, the uplink transmission rate R u It can be expressed by the above formula.

[0048] In the communication energy consumption model, the wired transmission delay of transmitting the task from the base station to the computing power provider via the computing power network edge gateway is calculated as follows:

[0049]

[0050] In the formula, To transfer task T u The wired transmission delay from the base station to the computing power provider via the computing power network edge gateway, where n represents the network provider of the wired transmission link. For task T u The total amount of data, the transmission rate from the base station to the edge gateway of the computing network is equal to the transmission rate from the edge gateway of the computing network to the computing power supplier, both of which are R un ;

[0051] So task T u The communication delay from the user to the designated computing power provider for:

[0052]

[0053] Since the delay of task result transmission is ignored, the calculation and transmission task T u Total delay for:

[0054]

[0055] (4) Energy consumption model

[0056] The energy consumption model includes a computing energy consumption model and a communication energy consumption model. The computing energy consumption model is used to calculate the energy consumption of the computing power supplier in processing tasks. The communication energy consumption model is used to calculate the energy consumption of transmitting tasks from the base station to the computing power supplier via the computing power network edge gateway. The wireless transmission energy consumption of tasks transmitted from users to base stations is small and can be ignored.

[0057] In some embodiments, the computing energy consumption model determines the computing energy consumption by: for:

[0058]

[0059] In the formula, v m is the energy coefficient of the computing power supplier m when processing tasks, which depends on the chip architecture of the server, and f m is the CPU frequency of computing power provider m, The number of cycles required to process the task;

[0060] The communication energy consumption model calculates the communication energy consumption in the following way for:

[0061]

[0062] In the formula, the transmission power of the link from the base station to the computing power network edge gateway is equal to the transmission power from the computing power network edge gateway to the designated computing power supplier, both of which are p n , To transfer task T u The wired transmission delay from the base station to the computing power provider via the computing power network edge gateway.

[0063] (5) Optimization algorithm

[0064] The optimization algorithm is used to determine the computing carbon emissions and communication carbon emissions based on the computing energy consumption, communication energy consumption and energy consumption composition of each computing power supplier when processing tasks in the carbon emission model, as well as the energy consumption composition of each network supplier when transmitting data, so as to determine the constraints by utilizing the total delay, total energy consumption and total carbon emissions, and to weight the total energy consumption and total carbon emissions to determine the optimization function, so as to determine the optimal computing power supplier and the optimal network supplier for processing the task while minimizing energy consumption and carbon emissions.

[0065] In this embodiment, user u has at most one task to be processed. If there is no task, the relevant energy consumption, carbon emissions and delay are all 0. It should be noted that although the wired links of different network providers have the same transmission rate, the power consumed to transmit the same task is different, that is, the transmission power is different. Moreover, due to the different energy ratios of each network provider, even if the power consumed to transmit the same task is the same, the carbon emissions are different. Therefore, task scheduling not only needs to consider the total delay to ensure service speed, but also needs to consider the total energy consumption and total carbon emissions at the same time to save electricity and reduce carbon emissions.

[0066] In this embodiment, the optimization algorithm is:

[0067] Determine the decision result x of the computing power supplier of the task:

[0068]

[0069] Network supplier decision result y:

[0070]

[0071] The total energy consumption E(x,y) is expressed as:

[0072]

[0073] In the formula, To calculate the energy consumption, For task T u The decision result of whether to dispatch to the computing power provider m, is the communication energy consumption, For task T u The decision result of whether to schedule to network provider n, where M is the number of computing power providers, U is the number of users, and N is the number of network providers;

[0074] The total carbon emissions C(x,y) can be expressed as:

[0075]

[0076] In the formula, α is the carbon emission per unit of thermal energy, The proportion of thermal energy consumed by computing power supplier m when processing tasks. The proportion of thermal energy consumed by network supplier n when transmitting data;

[0077] Weighting the total energy consumption and total carbon emissions yields the optimization function:

[0078] Min(E(x,y),C(x,y))

[0079] The constraints are:

[0080]

[0081] In the formula, is the total delay, D umax is the maximum delay that user u can tolerate, E is the maximum electric energy that can be provided, and C is the maximum tolerable carbon emission.

[0082] On the basis of satisfying the constraints, the optimization function is minimized, that is, the energy consumption and carbon emissions are minimized, to select the optimal computing power supplier and the optimal network supplier for each task and achieve green computing.

[0083] It should be noted that when there are many computing power suppliers, network suppliers and tasks, using the optimization algorithm to sequentially calculate the energy consumption and carbon emissions of various combinations of computing power suppliers and network suppliers requires a lot of computing resources and a long calculation time. Therefore, in this embodiment, a genetic algorithm can be used in combination with an optimization algorithm to achieve fast scheduling.

[0084] In some implementations, the computing network energy task scheduling module further includes a genetic algorithm unit for non-dominated sorting based on an elite strategy;

[0085] The genetic algorithm unit is used to use the optimization function and the constraint conditions as the objective function to find a set containing multiple optimal solutions from the population; wherein the population contains a number of individuals, each individual is composed of a computing power supplier and a network supplier, and the optimal solution is composed of the optimal computing power supplier and the optimal network supplier.

[0086] In some embodiments, the genetic algorithm unit is used to perform:

[0087] S1, randomly initialize a first-generation parent population with a population size of N, and perform non-dominated sorting on all individuals to obtain several levels of non-dominated sets;

[0088] S2, using selection, crossover, and mutation operators until the first generation of subgroups with a population size of N is generated, and the number of evolution generations is set to 2;

[0089] For each subsequent evolution, execute:

[0090] S3, after merging the parent population and the sub-population obtained in the last evolution, perform non-dominated sorting to obtain several levels of non-dominated sets, and calculate the crowding degree of each individual;

[0091] S4, starting from the first level non-dominated set, put the non-dominated sets of each level into the new parent population in ascending order. If the size of the new parent population will exceed N after all individuals in a non-dominated set are added to the new parent population, sort the crowding degree of the individuals in the non-dominated set in ascending order, and put the individuals into the new parent population in ascending order of crowding degree until a new parent population with a population size of N is generated;

[0092] S5, use selection, crossover, and mutation operators until a new subgroup of population size N is generated, and the evolutionary generation number is increased by 2;

[0093] S6, determine whether the current evolutionary generation is less than the evolutionary generation threshold. If so, start the next evolution and jump to step S3 until the evolutionary generation is not less than the evolutionary generation threshold, and obtain the optimal solution set.

[0094] In this embodiment, the computational complexity of the genetic algorithm is O(mN 2 ), determined by the non-dominated sorting part of the algorithm, m is the number of objective functions, and N is the population size. The calculation time can be controlled by setting the population size N. Therefore, the genetic algorithm unit based on the non-dominated sorting of the elite strategy in this embodiment can realize fast scheduling optimization based on the integration of computing and network.

[0095] In some implementations, non-dominated sorting is performed in the genetic algorithm unit in the following manner:

[0096] Obtain the first parameter and the second parameter of each individual in the population that currently needs to be non-dominated sorted; wherein the first parameter is the number of solution individuals that dominate the individual in the population, and the second parameter is the set of solution individuals dominated by the individual i;

[0097] Determine the individuals whose first parameter is 0 from the population, that is, all individuals i in the population that are not dominated by other individuals, and store them in the first-level non-dominated set;

[0098] For each subsequent classification, execute:

[0099] For each individual in the previous level non-dominated set, obtain the set of solution individuals dominated by the individual in the second parameter of these individuals, and after subtracting the first parameter of each individual in these solution individual sets by one, determine the individual whose current first parameter is 0 from these solution individual sets, and store it in the current level non-dominated set;

[0100] Until all individuals in the population are classified, several levels of non-dominated sets are obtained.

[0101] This embodiment can achieve multi-objective optimization, adopt advanced energy-saving technology and optimization algorithms, and effectively improve the energy efficiency of data centers and computing equipment. By combining optimization algorithms and genetic algorithms, energy consumption in the computing process is reduced, and the demand for power resources is reduced, which helps to reduce carbon emissions and energy waste, and promote the development of green computing.

[0102] like Figure 2 As shown, the embodiment of the present invention also provides a scheduling method based on the platform of any embodiment of the specification, the method comprising:

[0103] Step 200, using the user platform of the application service layer to receive and parse the task requests initiated by each user, and report the task to the control layer;

[0104] Step 202: The management and control layer uses the measurement module to measure the computing resources, network resources, and energy resources required to process each task;

[0105] Step 204: The management and control layer uses the perception module to perceive the computing resources, transmission network resources, and energy resources that can be provided by each computing power supplier, network supplier, and energy supplier in the infrastructure layer;

[0106] Step 206: The management and control layer uses the computing network energy task scheduling module to select the optimal computing power supplier and the optimal network supplier for each task based on the total delay, total energy consumption and total carbon emissions; wherein the total delay includes computing delay and communication delay, the total energy consumption includes computing energy consumption and communication energy consumption, and the total carbon emissions include computing carbon emissions and communication carbon emissions.

[0107] The content of the above method is based on the same concept as the platform embodiment of the present invention. For the specific content, please refer to the description in the platform embodiment of the present invention, and it will not be repeated here.

[0108] 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 such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0109] A person of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc., various media that can store program codes.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A computing, network and energy integrated collaborative task scheduling platform, characterized in that: include: Infrastructure layer, management and control layer, and application service layer; The infrastructure layer is composed of computing power suppliers, network suppliers and energy suppliers, which are used to provide computing power resources, transmission network resources and energy resources; wherein the energy includes clean energy and thermal energy; The application service layer at least includes a user platform, which is used to manage users, receive and analyze task requests initiated by each user, and report tasks to the control layer; The management and control layer includes at least a perception module, a measurement module and a computing network energy task scheduling module. The perception module is used to perceive the computing power resources, transmission network resources and energy resources that each computing power supplier, network supplier and energy supplier can provide. The measurement module is used to measure the computing power resources, network resources and energy resources required to process each task. The computing network energy task scheduling module is used to select the optimal computing power supplier and the optimal network supplier for each task based on the total delay, total energy consumption and total carbon emissions; wherein the total delay includes computing delay and communication delay, the total energy consumption includes computing energy consumption and communication energy consumption, and the total carbon emissions include computing carbon emissions and communication carbon emissions.

2. The platform according to claim 1, characterized in that The network energy task scheduling module contains a carbon emission model, a task model, a delay model, an energy consumption model and an optimization algorithm; The carbon emission model contains the energy ratio of each computing power supplier and the energy ratio of each network supplier; wherein the energy ratio includes the clean energy ratio and the thermal energy ratio, and the sum of the clean energy ratio and the thermal energy ratio is 1; Each task in the task model consists of a binary group, which stores the total amount of data and the number of cycles required for the task. A task can only be processed by one computing power provider. The delay model includes a computing delay model and a communication delay model; wherein the communication delay model is used to calculate the wireless transmission delay of the task from the user to the base station and the wired transmission delay of the task from the base station to the computing power supplier via the computing power network edge gateway. The wired transmission link represents the network supplier, and the result data obtained by the computing power supplier after processing the task is small, ignoring the delay of the result transmitted back to the user; The energy consumption model includes a computing energy consumption model and a communication energy consumption model. The computing energy consumption model is used to calculate the energy consumption of the computing power supplier for processing tasks. The communication energy consumption model is used to calculate the energy consumption of transmitting tasks from the base station to the computing power supplier via the computing power network edge gateway. The wireless transmission energy consumption of tasks transmitted from users to base stations is small and can be ignored. The optimization algorithm is used to determine the computing carbon emissions and communication carbon emissions based on the computing energy consumption, communication energy consumption, and the energy consumption composition of each computing power supplier when processing the task in the carbon emission model and the energy consumption composition of each network supplier when transmitting data, so as to determine the constraint conditions by using the total delay, total energy consumption and total carbon emissions, and to weight the total energy consumption and total carbon emissions to determine the optimization function, so as to determine the optimal computing power supplier and the optimal network supplier for processing the task while minimizing the energy consumption and carbon emissions.

3. The platform according to claim 2, characterized in that: The calculation delay model is: In the formula, Processing task T for computing power provider m u The time required, For task T u The total amount of data, Assign computing power provider m to task T u The computing power ratio, f m is the CPU frequency of computing power provider m; The wireless transmission delay of the task from the user to the base station in the communication energy consumption model is calculated as follows: In the formula, is the wireless transmission delay of the task from the user to the base station, For task T u The total amount of data, R u is the uplink transmission rate, β is the actual available bandwidth ratio of the user, B is the uplink channel bandwidth, p is the user's transmission power, and h u is the fading power gain between the user and the base station, σ 2 represents the Gaussian noise power, γ is the influence coefficient of interference on signal transmission; In the communication energy consumption model, the wired transmission delay of transmitting the task from the base station to the computing power supplier via the computing power network edge gateway is calculated as follows: In the formula, To transfer task T u The wired transmission delay from the base station to the computing power provider via the computing power network edge gateway, where n represents the network provider of the wired transmission link. For task T u The total amount of data, the transmission rate from the base station to the edge gateway of the computing network is equal to the transmission rate from the edge gateway of the computing network to the computing power supplier, both of which are R un ; So task T u The communication delay from the user to the designated computing power provider for: Computation and transmission tasks T u Total delay for:

4. The platform according to claim 2, characterized in that: The computing energy consumption model determines the computing energy consumption in the following way for: In the formula, v m is the energy coefficient of the computing power supplier m when processing tasks, f m is the CPU frequency of computing power provider m, The number of cycles required to process the task; The communication energy consumption model calculates the communication energy consumption in the following way for: In the formula, the transmission power of the link from the base station to the computing power network edge gateway is equal to the transmission power from the computing power network edge gateway to the designated computing power supplier, both of which are p n , To transfer task T u The wired transmission delay from the base station to the computing power provider via the computing power network edge gateway.

5. The platform according to claim 2, characterized in that: The optimization algorithm is: Determine the decision result x of the computing power supplier of the task: Network supplier decision result y: The total energy consumption E(x,y) is expressed as: In the formula, To calculate the energy consumption, For task T u The decision result of whether to dispatch to the computing power provider m, is the communication energy consumption, For task T u The decision result of whether to schedule to network provider n, where M is the number of computing power providers, U is the number of users, and N is the number of network providers; The total carbon emissions C(x,y) can be expressed as: In the formula, α is the carbon emission per unit of thermal energy, The proportion of thermal energy consumed by computing power supplier m when processing tasks. The proportion of thermal energy consumed by network supplier n when transmitting data; Weighting the total energy consumption and total carbon emissions yields the optimization function: Min(E(x,y),C(x,y)) The constraints are: E(x,y)≤E C(x,y)≤C In the formula, is the total delay, D umax is the maximum delay that user u can tolerate, E is the maximum electric energy that can be provided, and C is the maximum tolerable carbon emission.

6. The platform according to claim 5, characterized in that The computing network energy task scheduling module also includes a genetic algorithm unit for non-dominated sorting based on elite strategy; The genetic algorithm unit is used to use the optimization function and the constraint conditions as objective functions to find a set containing multiple optimal solutions from a population; wherein the population contains a number of individuals, each individual is composed of a computing power supplier and a network supplier, and the optimal solution is composed of an optimal computing power supplier and an optimal network supplier.

7. The platform according to claim 6, characterized in that The genetic algorithm unit is used to perform: S1, randomly initialize a first-generation parent population with a population size of N, and perform non-dominated sorting on all individuals to obtain several levels of non-dominated sets; S2, using selection, crossover, and mutation operators until the first generation of subgroups with a population size of N is generated, and the number of evolution generations is set to 2; For each subsequent evolution, execute: S3, after merging the parent population and the sub-population obtained in the last evolution, perform non-dominated sorting to obtain several levels of non-dominated sets, and calculate the crowding degree of each individual; S4, starting from the first level non-dominated set, put the non-dominated sets of each level into the new parent population in ascending order. If the size of the new parent population will exceed N after all individuals in a non-dominated set are added to the new parent population, sort the crowding degree of the individuals in the non-dominated set in ascending order, and put the individuals into the new parent population in ascending order of crowding degree until a new parent population with a population size of N is generated; S5, use selection, crossover, and mutation operators until a new subgroup of population size N is generated, and the evolutionary generation number is increased by 2; S6, determine whether the current evolutionary generation is less than the evolutionary generation threshold. If so, start the next evolution and jump to step S3 until the evolutionary generation is not less than the evolutionary generation threshold, and obtain the optimal solution set.

8. The platform according to claim 7, characterized in that The non-dominated sorting is performed in the genetic algorithm unit in the following manner: Obtain the first parameter and the second parameter of each individual in the population that currently needs to be non-dominated sorted; wherein the first parameter is the number of solution individuals that dominate the individual in the population, and the second parameter is the set of solution individuals dominated by the individual; Determine individuals whose first parameter is 0 from the population, and store them in a first-level non-dominated set; For each subsequent classification, execute: For each individual in the previous level non-dominated set, obtain the set of solution individuals dominated by the individual in the second parameter of these individuals, and after subtracting the first parameter of each individual in these solution individual sets by one, determine the individual whose current first parameter is 0 from these solution individual sets, and store it in the current level non-dominated set; Until all individuals in the population are classified, several levels of non-dominated sets are obtained.

9. The platform according to any one of claims 1 to 8, characterized in that: The application service layer also includes: a green computing energy trading module, a green computing energy operation module and an intelligent operation and maintenance module; The green computing energy trading module is used to enable users to purchase computing power resources and transmission network resources based on the optimal computing power supplier and the optimal network supplier output by the computing network energy task scheduling module, and to enable computing power suppliers and network suppliers to purchase energy from energy suppliers; The green computing energy operation module is used to carry out operation management planning of computing energy and energy; The intelligent operation and maintenance module is used to perform operation, management and maintenance of the network.

10. A scheduling method based on the platform according to any one of claims 1 to 9, characterized in that: The method comprises: The user platform of the application service layer receives and analyzes the task requests initiated by each user, and reports the tasks to the management and control layer; The control layer uses the measurement module to measure the computing resources, network resources and energy resources required to process each task; The management and control layer uses the perception module to perceive the computing resources, transmission network resources and energy resources that can be provided by each computing power supplier, network supplier and energy supplier in the infrastructure layer; The management and control layer uses the computing network energy task scheduling module to select the optimal computing power supplier and the optimal network supplier for each task based on the total delay, total energy consumption and total carbon emissions; wherein the total delay includes computing delay and communication delay, the total energy consumption includes computing energy consumption and communication energy consumption, and the total carbon emissions include computing carbon emissions and communication carbon emissions.

Images