Computing power network task scheduling method and device, storage medium and program product

Abstract

The application provides a computing power network task scheduling method and device, a storage medium and a program product, and relates to the technical field of computers. The method comprises the following steps: obtaining a task scheduling optimization problem by defining a target function between network energy consumption of a computing power network and network carbon emission of the computing power network and a constraint condition corresponding to the target function; the optimization target of the target function is to reduce the network energy consumption and the network carbon emission by optimizing computing power nodes, forwarding paths and forwarding time corresponding to a computing power task; the task scheduling optimization problem is solved by using a preset optimization algorithm to obtain a task scheduling method of the computing power network; the task scheduling method is used for indicating a forwarding path of the computing power task in the computing power network, a computing power node for executing the computing power task and forwarding time corresponding to the computing power node; the application can solve the problems of low scalability and management efficiency of task allocation; and the flexibility and management efficiency of task allocation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer technology, and in particular to a computing power network task scheduling method and device, a storage medium and a program product. BACKGROUND

[0002] With the rapid development of information technology, the number of networking devices in intelligent application systems has increased dramatically, generating unprecedented big data. These data are characterized by large volume, multiple types, and high speed, which puts higher demands on computing and network capabilities. Although cloud edge architecture improves communication and computing efficiency, resource allocation is mainly limited to a single domain, resulting in decreased resource utilization and limiting the development of data-intensive applications. The emergence of computing power networks (CPNs) has solved these problems. By connecting and coordinating heterogeneous computing resources in the cloud, edge, and endpoint, CPNs can effectively integrate computing and network resources for computing power task scheduling.

[0003] Currently, traditional computing power task scheduling methods include: a task allocation and management method based on deep reinforcement learning and software-defined networks; through deep reinforcement learning algorithms, the best task allocation strategy is learned autonomously through interaction with the environment, and optimization of energy consumption and delay is achieved; dynamic task scheduling and resource management are achieved through the centralized control characteristics of software-defined networks.

[0004] However, due to the increasing dynamic and unpredictable traffic in the Internet of Things, computing power task scheduling becomes complex, especially in real-time scheduling, which requires a highly adaptive mechanism. In addition, the resources of fog nodes are usually limited and are powered by batteries, and computing and energy resources are limited, which poses challenges to task scheduling and energy management. Therefore, it is very difficult to balance the relationship between network delay, scheduling accuracy, energy consumption, and response time in a heterogeneous Internet of Things environment, and there are problems of scalability and low management efficiency of task allocation. SUMMARY

[0005] In view of this, the embodiments of the present application provide a computing power network task scheduling method, device, storage medium and program product to eliminate or improve one or more defects in the prior art.

[0006] One aspect of the present application provides a computing power network task scheduling method, which comprises the following steps:

[0007] Defining a target function between network energy consumption of a computing power network and network carbon emissions of the computing power network and a constraint condition corresponding to the target function, obtaining a task scheduling optimization problem; the optimization goal of the target function is to reduce network energy consumption and network carbon emissions by optimizing computing power nodes, forwarding paths and forwarding times corresponding to computing power tasks; the constraint conditions include delay constraints, resource constraints, computing power node decision constraints and forwarding path decision constraints;

[0008] The task scheduling method of the computing power network is obtained by solving a task scheduling optimization problem through a preset optimization algorithm.

[0009] In some embodiments of the present application, the time delay constraint is a constraint condition constructed based on a relationship between the forwarding time and the processing time delay of the computing power task and a tolerable time delay of the computing power network.

[0010] The resource constraint includes a constraint condition constructed based on a relationship between the current load and a preset load, a constraint condition constructed based on a relationship between the network energy consumption and a preset total energy consumption, and a constraint condition constructed based on a relationship between the cache data amount of the computing power node and a preset cache resource.

[0011] In some embodiments of the present application, the computing power task includes a first set of computing power tasks received in a current time slot and a second set of computing power tasks received before the current time slot; and the network energy consumption includes transmission energy consumption and execution energy consumption of each computing power task in the first set of computing power tasks and the second set of computing power tasks.

[0012] In some embodiments of the present application, the processing time delay includes a task execution time delay corresponding to the computing power task, a queuing waiting time delay, and a task transmission time delay; and the current load includes a current computing power network load and a current computing power node load.

[0013] In some embodiments of the present application, the task scheduling method of the computing power network is obtained by solving a task scheduling optimization problem through a preset optimization algorithm, and includes:

[0014] Initializing a current population; the current population includes a plurality of solutions;

[0015] Based on the current population, the network energy consumption and the network carbon emission are calculated to obtain a constraint condition violation condition corresponding to each solution based on the constraint condition;

[0016] Based on the constraint condition violation condition, each solution in the current population is non-dominantly sorted to obtain a non-dominant sorting result;

[0017] Based on the network energy consumption and the network carbon emission, a crowding distance corresponding to each solution is determined;

[0018] Based on the non-dominant sorting result and the crowding distance corresponding to each solution, a candidate solution is determined in the current population;

[0019] In the case of algorithm convergence, the candidate solution is determined as the task scheduling method.

[0020] In some embodiments of the present application, in the case where the algorithm does not converge, the method further comprises:

[0021] crossing and mutating the candidate solutions to generate a child population;

[0022] merging the current population and the child population, and iterating the current population;

[0023] performing, based on the current population, a step of calculating the objective function and obtaining, based on the constraint conditions, the constraint condition violation situation corresponding to each solution.

[0024] In some embodiments of the present application, before crossing and mutating the candidate solutions to generate a child population, the method further comprises: dynamically adjusting the crossing probability and the mutation probability corresponding to the current population based on the network energy consumption and the network carbon emission.

[0025] Another aspect of the present application provides a computing power network task scheduling device, comprising a processor, a memory and a computer program / instruction stored on the memory, wherein the processor is configured to execute the computer program / instruction, and when the computer program / instruction is executed, the device implements the steps of the foregoing computing power network task scheduling method.

[0026] Another aspect of the present application provides a computer readable storage medium, which stores a computer program / instruction, and when the computer program / instruction is executed by a processor, the steps of the foregoing computing power network task scheduling method are implemented.

[0027] Another aspect of the present application provides a computer program product, which comprises a computer program / instruction, and when the computer program / instruction is executed by a processor, the steps of the foregoing computing power network task scheduling method are implemented.

[0028] The computing power network task scheduling method, device, storage medium and program product of the present application can solve the problems of low scalability and management efficiency of task allocation; the computing power network task scheduling device deeply integrates the computing power of the computing power nodes and the network resources, including energy consumption and cache resources, etc.; by defining the optimization target as minimizing the total energy consumption and carbon emission of the computing power network, while meeting the constraint conditions of tolerable delay, processing delay and load balancing of the task, and combining the dynamic network load and delay demand, an adaptive scheduling strategy is provided for task allocation of big data application, so as to realize efficient scheduling of computing power tasks, optimize the utilization of renewable energy, improve the overall operation efficiency of the computing power network, and at the same time, improve the flexibility and management efficiency of task allocation, so as to ensure that the computing power network can still operate efficiently and stably under the condition of traffic growth and resource limitation.

[0029] Additional advantages, objects, and features of the application will be set forth in part by the description that follows, and will become apparent to those skilled in the art upon examination of the following detailed description and drawings in which:

[0030] Those skilled in the art will appreciate that the objects and advantages of the application can be obtained by specific structures as specifically described below and as illustrated in the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0031] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description, serve to explain the principles of the application.

[0032] Figure 1 A flow chart of a computing power network task scheduling method provided by an embodiment of the application.

[0033] Figure 2 A structural schematic diagram of a computing power network provided by an embodiment of the application.

[0034] Figure 3 A flow chart of a solution to a task scheduling optimization problem provided by an embodiment of the application. DETAILED DESCRIPTION

[0035] In order to make the objects, technical solutions and advantages of the application clearer, the following further describes the application with reference to the embodiments and drawings. Herein, the illustrative embodiments of the application and the descriptions thereof are used to explain the application, but are not used to limit the application.

[0036] It should be noted that, in order to avoid the application being obscured by unnecessary details, only the structures and / or processing steps closely related to the solutions according to the application are shown in the drawings, and other details not closely related to the application are omitted.

[0037] It should be emphasized that the term "comprises / comprising" as used herein is used to indicate the presence of the stated features, elements, steps or components, but does not preclude the presence or addition of one or more other features, elements, steps or components.

[0038] It should be noted that, if not specifically stated, the term "connected" as used herein can not only mean direct connection, but also indirect connection with an intermediate object.

[0039] In the following, embodiments of the application will be described with reference to the drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps. It should be noted that, in order to avoid the application being obscured by unnecessary details, only the structures and / or processing steps closely related to the solutions according to the application are shown in the drawings, and other details not closely related to the application are omitted.

[0040] Figure 1 is a flowchart of a computing power network task scheduling method according to an embodiment of the present application. The computing power network task scheduling method is applied to a computing power network task scheduling device in a computing power network.

[0041] Reference Figure 2 In some embodiments of the present application, in order to support the operation of the computing power network, the computing power network needs to obtain energy from an energy supply module. The computing power network includes, but is not limited to, a computing power network task scheduling device, a computing power node, a computing power network router, and a general-purpose router. The number of computing power network task scheduling devices, computing power nodes, computing power network routers, and general-purpose routers is at least one. The computing power network task scheduling device includes a server, a load balancer, or an edge computer, etc. The computing power node includes an edge computer, a server, or a tablet computer, etc. The number and type of devices in the computing power network are not limited in this embodiment.

[0042] In some embodiments of the present application, the general-purpose router and the computing power network router are used to receive the computing power task sent by the client and forward it to the computing power node; the computing power node is used to execute the computing power task; the computing power network task scheduling device obtains the resource information of the computing power network and the resource information of the energy supply module using discrete time slots, and executes the computing power network task scheduling method.

[0043] Each discrete time slot represents a fixed time unit The running time of the computing power network is divided into time periods of fixed length according to the time unit In each discrete time slot, the energy value is static, and the energy values corresponding to different discrete time slots are the same or different.

[0044] In some embodiments of the present application, the set of computing power nodes is represented as The set of routers is represented as wherein represents a general-purpose router responsible for receiving and forwarding data, represents a computing power network router responsible for data transmission and forwarding.

[0045] The network structure of the computing power network is represented by an undirected connected graph .

[0046] wherein represents the set of computing power nodes, general-purpose routers, and computing power network routers in the computing power network, which is represented by the following formula:

[0047]

[0048] wherein, denotes a set of computing power nodes; denotes a set of routers; denotes the number of computing power nodes or computing power network routers; denotes the number of general routers.

[0049] denotes a set of network links in the computing power network, which can be represented by the following formula:

[0050]

[0051] wherein, denotes a direct network connection between the computing power nodes, the general routers and the computing power network routers, and the network connection weight is non-negative; denotes a set of computing power nodes, general routers and computing power network routers in the computing power network.

[0052] For any discrete time slot , the maximum network link capacity can be represented by the following formula:

[0053]

[0054] wherein, denotes the maximum network link capacity of the network link ; is a matrix representing the network link capacity and network transmission rate corresponding to each node in the computing power network.

[0055] The network transmission rate can be represented by the following formula:

[0056]

[0057] wherein, denotes the network transmission rate of the network link ; is a matrix representing the network link capacity and network transmission rate corresponding to each node in the computing power network.

[0058] In the case of the network link , the maximum network link capacity of the network link is greater than 0, and the network transmission rate of the network link is greater than 0; otherwise, the maximum network link capacity of the network link is equal to 0, and the network transmission rate of the network link is equal to 0.

[0059] In some embodiments of the present application, by the existence of multiple transmission paths between the routers and the computing power nodes, the following formula can be used:

[0060]

[0061] wherein, represents a general router between the computing power nodes ; represents the first transmission path.

[0062] The computing power tasks received by the computing power network refer to tasks sent by users through clients, which are received and forwarded by routers in the computing power network. For any period of time ; at discrete time slot =0, the first set of computing power tasks (i.e., “new” tasks) arriving at the general router are represented as .

[0063] wherein, = , represents the set of new tasks arriving at the general router ; represents the th new task arriving at the general router . Wherein, represents the data volume (in bits) corresponding to the computing power task ; represents the computing volume (i.e., the number of CPU cycles required to execute the task) corresponding to the computing power task ; represents the tolerable delay corresponding to the computing power task . When the general router receives a computing power task, it can temporarily put the task into its cache, and the computing power network task scheduling device will ensure that the task is processed within [0, .

[0064] The second set of computing power tasks (i.e., the “historical” task set) arriving at the general router before discrete time slot =0. Although the computing power network has made scheduling decisions for historical tasks, due to the forwarding delay mechanism, historical tasks may be forwarded later.

[0065] The computing power network consumes a large amount of computing resources in big data processing, resulting in high energy consumption, greatly increasing the carbon footprint of the computing power network, and exacerbating global climate warming. Based on this, in order to reduce the negative impact on the environment, the computing power network urgently needs to reduce dependence on non-renewable energy and use renewable energy as the main energy supply, which has almost no carbon emissions. In this context, a task scheduling scheme that can improve energy efficiency and maximize the use of renewable energy will become a key technology to realize a sustainable and green computing power network and promote the development of big data.​

[0066] In the traditional computing network task scheduling system, the characteristics of the computing network and renewable energy are mostly ignored. Specifically, the computing network forced scheduling scheme jointly considers the influence of heterogeneous computing nodes and complex network paths on delay and energy consumption, but due to the geographical diversity and unpredictability of big data applications and renewable energy, the renewable energy computing network has high dynamics with time and location. In this case, the traditional computing network task scheduling system ignores the network consideration, and the space-time dynamics may not be suitable for green computing network.

[0067] To solve the above technical problems, the computing network task scheduling method provided by the present application fully considers the fusion of network and computing capacity, and the space-time dimension of big data application and renewable energy, aiming to minimize energy consumption and carbon emissions under the conditions of meeting delay constraints and long-term load balancing. The computing network task scheduling method provided by the embodiments of the present application is explained in detail below. The computing network task scheduling method can include the following steps S101 to S102:

[0068] Step S101, defining a target function between the network energy consumption of the computing network and the network carbon emission of the computing network and the constraint condition corresponding to the target function, obtaining a task scheduling optimization problem.

[0069] The network energy consumption of the computing network refers to the electric energy consumed for data transmission, processing and storage in the computing network. The network carbon emission of the computing network refers to the carbon dioxide (CO2) emission generated due to power consumption in the whole operation process of the computing network.

[0070] In some embodiments of the present application, the optimization target of the target function is to reduce the network energy consumption and the network carbon emission by optimizing the computing node, forwarding path and forwarding time corresponding to the computing task. Specifically, the target function can be represented by the following formula:

[0071]

[0072] wherein, represents the network energy consumption of the computing network; represents the network carbon emission of the computing network; represents the computing node decision, which is used to indicate the computing node executing the computing task; represents the forwarding path decision, which is used to indicate the forwarding path of the computing task in the computing network; represents the forwarding time of the computing task.

[0073] In some embodiments of the present application, the network energy consumption of the computing network includes the energy consumption of the computing task through the network path Transmission energy consumption and processing power task of a route of the power node of the execution energy consumption. Wherein, the power task includes a first set of power tasks received in the current time slot and a second set of power tasks received before the current time slot; accordingly, the network energy consumption includes the transmission energy consumption and the execution energy consumption of each power task in the first set of power tasks and the second set of power tasks.

[0074] Specifically, the transmission energy consumption is represented by the following formula:

[0075]

[0076] Wherein, represents the power task corresponding data volume; represents the set of routers along the network path ; represents the general router; represents the set of routers ; represents the energy consumption when the transponder of the general router performs photoelectric / electro-optical conversion; represents the energy consumption when the transponder of the power network router performs photoelectric / electro-optical conversion; represents the energy consumption of the electronic processing process of the general router .

[0077] Processing power task of the power node The execution energy consumption is represented by the following formula:

[0078]

[0079] In the formula, represents the effective capacitance parameter of the power node of the computing chip set; is the computing power resource (i.e. CPU computing frequency) of the power node ; represents the computing amount (i.e. the number of CPU cycles required for task execution).

[0080] The network carbon emission of the computing power network includes the carbon emission of the computing power network and the computing power nodes. The computing power nodes (such as data centers, servers, etc.) need to consume a large amount of power when performing calculations. The production mode of the power (such as coal, natural gas, renewable energy, etc.) directly affects the carbon emission. If the power comes from non-renewable energy sources such as fossil fuels, the carbon emission will increase significantly. At the same time, data transmission in the computing power network also consumes power, especially in high-load situations. The network equipment required for transmission (such as routers, switches, etc.) also needs power.

[0081] wherein the renewable energy includes solar energy and wind energy. For renewable energy, no carbon emission is considered. Based on this, the computing power network and the computing power nodes The carbon emission of the computing power network in the time slot may be represented as:

[0082]

[0083] wherein, represents the carbon intensity of non-renewable energy; represents the corresponding carbon emission of the computing power network in the time slot . and both represent the corresponding carbon emission of the computing power nodes in the time slot .

[0084] wherein, is represented by the following formula:

[0085]

[0086] In the formula, represents the transmission energy consumption of the historical tasks corresponding to the time slot . represents the transmission energy consumption of the new tasks corresponding to the time slot . represents the supply amount of renewable energy of the computing power network corresponding to the time slot , which is represented by the following formula:

[0087]

[0088] In the formula, represents the solar power generation of the solar panels of the renewable energy supplier matched with the computing power network in the time slot . Specifically, is represented by: , wherein, represents the area of the solar panels, represents the conversion rate of solar radiation into electric energy, represents the time slot solar irradiance of time slot t.

[0089] represents the wind power generation amount of the wind turbine group of the renewable energy supplier matched with the computing power node at time slot t. Specifically, it is represented by: ; wherein, represents the wind power conversion rate, represents the conversion rate of solar radiation into electric energy, represents the air density of the wind power site, represents the wind power rotor area, represents the wind speed of the wind turbine group at time slot t.

[0090] is represented by:

[0091]

[0092] wherein, represents the computing energy consumption of the corresponding historical task on the computing power node at time slot t. represents the computing energy consumption of the corresponding new task on the computing power node at time slot t. represents the supply amount of renewable energy of the computing power network at time slot t, which is represented by:

[0093] wherein,

[0094] represents the solar power generation amount of the solar panel of the renewable energy supplier matched with the computing power node at time slot t. Specifically, it is represented by: , wherein, represents the area of the solar panel, represents the conversion rate of solar radiation into electric energy, represents the solar irradiance of time slot t.

[0095] represents the wind power generation amount of the wind turbine group of the renewable energy supplier matched with the computing power node at time slot t. Specifically, it is represented by: ; wherein, represents the wind power conversion rate,​​​​​​​​​​​​ represents the air density of the wind power site, represents the rotor area of the wind power, represents the wind speed of the wind turbine at the time slot .

[0096] In some embodiments of the present application, the constraint conditions include a time delay constraint, a resource constraint, a computing node decision constraint, and a forwarding path decision constraint.

[0097] The time delay constraint is a constraint condition constructed based on the relationship between the forwarding time and processing time delay corresponding to the computing task and the tolerable time delay of the computing network. The resource constraint includes a constraint condition constructed based on the relationship between the current load and the preset load, a constraint condition constructed based on the relationship between the network energy consumption and the preset total energy, and a constraint condition constructed based on the relationship between the cache data amount corresponding to the computing node and the preset cache resource.

[0098] In the conventional task allocation and management method based on deep reinforcement learning and software defined network (SDN), the deep reinforcement learning algorithm can autonomously learn the optimal task allocation strategy through interaction with the environment, and realize optimization of energy consumption and delay. The centralized control characteristics of SDN support dynamic task scheduling and resource management. When a new task arrives, the SDN controller obtains the state information of all fog nodes, and the deep reinforcement learning algorithm selects the optimal node for processing according to the residual energy, current load and task execution demand of the node. At the same time, the SDN controller analyzes the network and node state in real time to ensure that the task is allocated to the fog node closest to the source node and with the lowest delay. In addition, in order to improve energy efficiency, the task is preferentially allocated to the node with sufficient energy.

[0099] However, due to the increasing dynamic and unpredictable traffic in the Internet of Things, the computing task scheduling becomes complex, especially in real-time scheduling, which requires a highly adaptive mechanism. In addition, the resources of the fog nodes are usually limited and are mostly battery-powered, and the computing and energy resources are limited, which brings challenges to task scheduling and energy management. Based on this, it is also very difficult to balance the relationship between network delay, scheduling accuracy, energy consumption and response time in the heterogeneous Internet of Things environment, and there are problems of low scalability and management efficiency of task allocation.

[0100] To solve the above problems, in some embodiments of the present application, the computing power network task scheduling device deeply integrates the computing power of the computing power node and the network resources, including energy consumption and cache resources, etc. By defining the optimization goal as minimizing the total energy consumption and carbon emissions of the computing power network, while meeting the constraints of tolerable delay, processing delay and load balancing of the task, combined with dynamic network load and delay demand, an adaptive scheduling strategy is provided for task allocation of big data application, to realize efficient scheduling of computing power tasks, optimize the use of renewable energy, improve the overall operation efficiency of the computing power network, and improve the flexibility and management efficiency of task allocation, ensuring that the computing power network can still run efficiently and stably under the condition of traffic growth and resource limitation.

[0101] In some embodiments of the present application, the processing delay includes the sum of the task execution delay corresponding to the computing power task, the queuing waiting delay and the task transmission delay.

[0102] Wherein, the task transmission delay is represented by the following formula:

[0103]

[0104] In the formula, represents the data volume corresponding to the computing power task . represents the network link; represents the s-th transmission path between the general router and the computing power node; represents the network transmission rate; represents the forwarding timestamp corresponding to the computing power task .

[0105]

[0106] In the formula, is a non-negative integer; represents the virtual delay. The computing power task arriving at time slot 0 will be forwarded at time slot .

[0107] The average queuing waiting delay of the computing power task can be represented as the sum of the average queuing waiting delay of the historical task arriving before the computing power task , the average queuing waiting delay of the historical task arriving at the same time as the computing power task , and the average queuing waiting delay caused by other new tasks. Specifically, the average queuing waiting delay of the computing power task can be represented by the following formula:

[0108]

[0109] In the formula, represents the computing power resource (i.e., CPU computing frequency) of the computing power node; represents the total computing amount of other new computing power tasks that arrive at the same time as the computing power task; represents the computing demand amount of historical tasks that arrive asynchronously with the computing power task; represents the computing demand amount of historical tasks that arrive synchronously with the computing power task.

[0110] The task execution delay refers to the delay generated by the computing power node in executing the computing power task, which can be represented by the following formula:

[0111]

[0112] In the formula, represents the computing power resource of the computing power node; represents the computing amount, i.e., the number of CPU cycles required to execute the task.

[0113] Specifically, the delay constraint constructed based on the relationship between the forwarding time and processing delay of the computing power task and the tolerable delay of the computing power network is represented by the following formula:

[0114]

[0115] In the formula, represents the forwarding timestamp corresponding to the computing power task; represents the processing delay corresponding to the computing power task; represents the tolerable delay corresponding to the computing power task; represents the first set of computing power tasks.

[0116] Maintaining long-term load balancing is crucial for improving resource utilization and ensuring system stability. In some embodiments of the present application, the current load includes network load balancing corresponding to the computing power network and node load balancing corresponding to the computing power node. Specifically, the current load can be represented by the following formula:

[0117]

[0118] In the formula, and represent weight factors; represents node load balancing; represents network load balancing. ​​​​​​​​​​

[0119] In some embodiments of the present application, the load balancing of all computing power nodes is evaluated by standard deviation, specifically, node load balancing which can be expressed by the following formula:

[0120]

[0121] In the formula, represents the computing power node The load at discrete time slot t is represented as wherein, represents the computing power node The maximum cache resource corresponding to the maintained cache queue, represents the discrete time slot The total amount of data cached in the computing power node at discrete time slot t; represents the average load of the computing power nodes in the entire computing power network.

[0122] Load balancing model of the computing power network which can be expressed by the following formula:

[0123]

[0124] In the formula, represents the network link The load at discrete time slot is represented as wherein, represents the network link The amount of data at discrete time slot , represents the average load of the entire computing power network; represents the set of network links in the computing power network.

[0125] Specifically, the constraint of the relationship between the current load and the preset load can be expressed by the following formula:

[0126]

[0127] In the formula, represents the discrete time slot The corresponding current load, including the network load balancing corresponding to the computing power network and the node load balancing corresponding to the computing power node; represents the number of discrete time slots; represents the preset load.

[0128] In a conventional task allocation method for real-time applications in a heterogeneous sensor network, tasks are divided into synchronous tasks and simple tasks according to the number of parent tasks and child tasks. After task division, the entire task graph is converted into a task partition graph, each branch of which is independently scheduled to reduce the complexity of scheduling. After task partitioning, the overall deadline is allocated to each task partition to ensure that the tasks in each partition can be completed within the respective sub-deadline, ensuring that the overall execution time of the final task does not exceed the global deadline, and minimizing energy consumption while ensuring that the deadline is met.

[0129] However, in the above conventional method, while reducing energy consumption, the algorithm must balance between energy consumption and real-time performance. It is very complex to find the best balance point without affecting either side, and it is difficult to balance energy consumption and service quality; at the same time, there are usually dependencies between tasks, and if the task assigned to a fast node needs to wait for a dependent task assigned to a slow node while minimizing energy consumption, unnecessary delays may occur, reducing efficiency.

[0130] To solve the above problems, in some embodiments of the present application, by introducing delay constraints and long-term load balancing strategies, the balance between energy consumption and real-time performance in the conventional method is effectively solved. This method not only ensures the reliability and efficiency of task processing, but also reduces the delay caused by dependencies through a dynamic scheduling mechanism, avoiding blocking fast node tasks on slow nodes. In addition, while reducing energy consumption, carbon emissions are effectively reduced, providing strong support for green computing. This solution not only applies to current complex big data application scenarios, but also provides important technical support for future intelligent energy management, promoting the goal of sustainable development.

[0131] In some embodiments of the present application, the constraint condition constructed based on the relationship between network energy consumption and the preset total energy amount can be represented by the following formula:

[0132]

[0133] wherein, represents network energy consumption; represents the preset total energy. In some embodiments of the present application, refers to the renewable energy that the energy supply module can provide.

[0134] The constraint condition constructed based on the relationship between the cache data amount corresponding to the computing power node and the preset cache resource is represented by the following formula:

[0135]

[0136] In the formula, represents the total amount of data cached in the computing power node m at discrete time slot t; represents the preset cache resource of the computing power node m.

[0137] The computing power node decision constraint and the forwarding path decision constraint are represented by the following formula:

[0138]

[0139] In the formula, represents the general router and the transmission path set between the computing power node ; represents the first transmission path in ; represents the computing power node decision, and in the case that is 1, it represents that the computing power node indicated by the computing power node decision is the computing power node ; represents the forwarding path decision, and in the case that is 1, it represents that the forwarding path is the first transmission path in ; represents the computing power task; computing power task set.

[0140] Step S102: solving the task scheduling optimization problem by a preset optimization algorithm to obtain a task scheduling method of the computing power network; the task scheduling method is used to indicate the forwarding path of the computing power task in the computing power network, the computing power node executing the computing power task, and the forwarding time corresponding to the computing power node.

[0141] The task scheduling optimization problem can be represented by the following formula:

[0142]

[0143]

[0144]

[0145]

[0146]

[0147]

[0148] In the formula, represents the network energy consumption of the computing power network; represents the network carbon emission of the computing power network; represents a computing power node decision, used to indicate a computing power node performing a computing power task; represents a forwarding path decision, used to indicate a forwarding path of a computing power task in a computing power network; represents a forwarding time of a computing power task; represents a computing power task corresponding forwarding timestamp; represents a computing power task corresponding processing delay; represents a computing power task corresponding tolerable delay; represents a first computing power task set; represents a discrete time slot corresponding current load, including network load balancing of the computing power network and node load balancing of the computing power node; represents the number of discrete time slots; represents a preset load; represents a preset total energy; represents a total data amount cached in the computing power node m at the discrete time slot t; represents a preset cache resource of the computing power node m; represents a general router and the computing power node between them; represents the th transmission path in .

[0149] In some embodiments of the present application, a preset optimization algorithm is used to solve the task scheduling optimization problem, where the preset optimization algorithm refers to a pre-set multi-objective optimization algorithm, including but not limited to Non-dominated Sorting Genetic Algorithm II (NSGA-II), Genetic Algorithm (GA), or Particle Swarm Optimization (PSO), etc.

[0150] In the case where the preset optimization algorithm is the non-dominated sorting genetic algorithm, the algorithm is optimized on the basis of the original non-dominated sorting genetic algorithm to adapt to its specific requirements and constraints, including initialization of population, non-dominated sorting, calculation of crowding distance, selection, crossover and mutation processing, and merging and reselection processing.

[0151] where the initialization of population refers to randomly generating a population of solutions, denoted as , where represents the population of the representing the first generation in the population , each individual, representing the total size, each individual contains the computing power node decision, forwarding path and forwarding time. At the same time, the initialization also includes the population related parameters, including the maximum number of generations , crossover probability , mutation probability .

[0152] Non-dominated sorting refers to non-dominated sorting of the population, and identifying different dominance levels. The calculation of the crowding distance refers to the calculation of the crowding degree of each solution to maintain the diversity of the solution.

[0153] In some embodiments of the present application, by calculating the objective function and , and the violation of the constraint condition, all individuals are assigned to different non-dominated queues. At the same time, in order to ensure the diversity of the population, according to and , the crowding distance of each solution is estimated, and the candidate solution is selected by non-dominated sorting and crowding distance.

[0154] The selection of crossover and mutation processing refers to the selection based on non-dominated level and crowding distance, and the crossover processing and mutation except to generate a new population. After that, by merging and reselecting, the parent and child are merged, and the optimal solution is selected by non-dominated sorting again.

[0155] Specifically, the task scheduling method of the computing power network is obtained by solving the task scheduling optimization problem by a preset optimization algorithm, including: initializing a current population; the current population includes a plurality of solutions; based on the current population, the network energy consumption and the network carbon emission are calculated by calculating the objective function, and based on the constraint condition, the constraint condition violation of each solution is obtained; based on the constraint condition violation, each solution in the current population is non-dominated sorted to obtain a non-dominated sorting result; based on the network energy consumption and the network carbon emission, the crowding distance corresponding to each solution is determined; based on the non-dominated sorting result and the crowding distance corresponding to each solution, the candidate solution is determined in the current population; in the case of algorithm convergence, the candidate solution is determined as the task scheduling method. Wherein, the candidate solution refers to the solution with higher dominance level and shorter crowding distance.

[0156] In the case of algorithm not converging, the candidate solution is subjected to crossover processing and mutation processing, the scheduling strategy of multiple solutions is combined through crossover, and the diversity of the solution is increased by randomly changing part of the scheduling strategy through mutation processing. After that, the above steps are repeated until the algorithm converges.

[0157] Specifically, in the case where the algorithm does not converge, further comprising: performing crossover processing and mutation processing on the candidate solutions to generate a child population; merging the current population and the child population, and then iterating the current population; performing, based on the current population, calculation of the objective function and, based on the constraint condition, obtaining the constraint condition violation condition corresponding to each solution.

[0158] In addition, in the computing power network, since the network energy consumption and carbon emissions may change over time, in order to improve the practicability of the task scheduling method, the crossover probability and the mutation probability are dynamically adjusted according to the network energy consumption and the network carbon emissions before the crossover processing and the mutation processing are performed; at the same time, unnecessary calculations can be reduced, and energy consumption can be reduced.

[0159] Specifically, before the crossover processing and the mutation processing are performed on the candidate solutions to generate a child population, further comprising: dynamically adjusting the crossover probability and the mutation probability corresponding to the current population based on the network energy consumption and the network carbon emissions.

[0160] Reference Figure 3 The solution process of the task scheduling optimization problem comprises at least the following steps:

[0161] Step S301, initializing a population P and population parameters, wherein the population parameters comprise population algebra, crossover probability and mutation probability;

[0162] Step S302, estimating an optimization objective function and a constraint condition of the population P, and performing non-dominated sorting on the population P;

[0163] Step S303, calculating a crowding distance corresponding to the population P;

[0164] Step S304, determining candidate solutions based on the crowding distance and the non-dominated sorting;

[0165] Step S305, dynamically updating the crossover probability and the mutation probability;

[0166] Step S306, performing crossover processing and mutation processing on the candidate solutions to generate a child population, and calculating a constraint condition;

[0167] Step S307, determining whether the current algebra is greater than the population algebra, and performing step S309 if the current algebra is greater than the population algebra, otherwise performing step S308;

[0168] Step S308, integrating and classifying the child population and the population P to obtain a new population P; performing step S302;

[0169] Step S309, outputting the candidate solutions as a task scheduling method.

[0170] In summary, the computing power network task scheduling method provided by the application defines the objective function between the network energy consumption of the computing power network and the network carbon emission of the computing power network and the constraint condition corresponding to the objective function, obtains a task scheduling optimization problem, the optimization target of the objective function is to reduce the network energy consumption and the network carbon emission by optimizing the computing power node, the forwarding path and the forwarding time corresponding to the computing power task, the constraint condition includes the delay constraint, the resource constraint, the computing power node decision constraint and the forwarding path decision constraint, the task scheduling optimization problem is solved by using a preset optimization algorithm, and the task scheduling method of the computing power network is obtained, the task scheduling method is used for indicating the forwarding path of the computing power task in the computing power network, the computing power node for executing the computing power task and the forwarding time corresponding to the computing power node, the problem of low expandability and management efficiency of task allocation can be solved, the computing capability and the network resources of the computing power node are deeply integrated by using the computing power network task scheduling device, the network resources include the energy consumption and the cache resource, the optimization target is defined as minimizing the total energy consumption and the carbon emission of the computing power network, the constraint conditions such as the tolerable delay, the processing delay and the load balancing of the task are met, the dynamic network load and the delay demand are combined, a scheduling strategy with strong adaptability is provided for task allocation of the big data application, the efficient scheduling of the computing power task is realized, the utilization of the renewable energy is optimized, the overall operation efficiency of the computing power network is improved, the flexibility and the management efficiency of the task allocation are improved, and it is ensured that the computing power network can still be efficiently and stably operated under the condition of traffic growth and resource limitation.

[0171] Corresponding to the above method, the application further provides a computing power network task scheduling device, which comprises a computer device, the computer device comprises a processor and a memory, the memory stores computer instructions, and the processor is used for executing the computer instructions stored in the memory, and the device realizes the steps of the computing power network task scheduling method as described above when the computer instructions are executed by the processor.

[0172] The embodiment of the application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by the processor to realize the steps of the foregoing computing power network task scheduling method. The computer readable storage medium can be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM or any other form of storage medium known in the technical field.

[0173] The embodiment of the application further provides a computer program product, which comprises a computer program / instruction, and the computer program / instruction is executed by the processor to realize the steps of the foregoing computing power network task scheduling method.

[0174] Those of ordinary skill in the art will appreciate that the various illustrative components, systems and methods described in connection with the embodiments disclosed herein can be implemented as hardware, software, or both. The particular implementation is dependent on the specific application and design constraints imposed on the overall system. Skilled persons can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application. When implemented in hardware, for example, the hardware can comprise an electronic circuit, an Application Specific Integrated Circuit (ASIC), a suitable firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the application are the program or code segments to perform a specific task. The program or code segments can be stored in a machine-readable medium, or transmitted by a carrier wave as data signals over a transmission medium or communication link.

[0175] It is to be understood that the application is not limited to the particular configurations and processes described herein and shown in the drawings, which can be varied in accordance with the particular needs of the application. For the sake of brevity, conventional techniques and methods related to making and using the application can not be described in detail herein. In the above embodiments, several specific steps are described and illustrated in order to provide a thorough understanding of the present application. However, the process of the present application can be practiced with less than all of the described specific steps, or with additional steps, or with different order of the steps, without departing from the spirit of the application.

[0176] In the present application, features described and / or illustrated in connection with one embodiment can be used in the same or a similar way or in conjunction with or in place of features of another embodiment.

[0177] The above description is intended to be illustrative and not restrictive. Many embodiments of the application will be apparent to those of skill in the art upon reviewing the above description. The scope of the application should, therefore, be determined not with reference to the above description, but instead with reference to the appended claims, along with their full scope of equivalents.

Claims

1. A method for scheduling tasks in a computing power network, characterized in that, Applied to a computing power network, the method comprises: defining a target function between network energy consumption of the computing power network and network carbon emission of the computing power network and a constraint condition corresponding to the target function, obtaining a task scheduling optimization problem; the optimization goal of the target function is to reduce the network energy consumption and the network carbon emission by optimizing the computing power nodes, forwarding paths and forwarding times corresponding to the computing power tasks; the constraint condition includes a delay constraint, a resource constraint, a computing power node decision constraint and a forwarding path decision constraint; solving the task scheduling optimization problem by a preset optimization algorithm to obtain a task scheduling method of the computing power network; the task scheduling method is used to indicate the forwarding path of the computing power task in the computing power network, the computing power node executing the computing power task and the forwarding time corresponding to the computing power node; The network energy consumption of the computing power network includes transmission energy consumption of computing power tasks routed through a network path and processing energy consumption of computing power tasks of the computing power node ​ the expression of the transmission energy consumption is: wherein, represents the computing power task corresponding data volume, represents the general router receiving the first new task; represents the router set along the network path , the s-th transmission path of router i and computing power node m; represents the energy consumption of the transponder of the general router when performing photoelectric or electro-optical conversion; represents the energy consumption of the transponder of the computing power network router when performing photoelectric or electro-optical conversion; represents the energy consumption of the electronic processing process of the general router ; the expression of the execution energy consumption is: In the formula, represents the effective capacitance parameter of the computing chip group; represents the computing power node of the computing chip group; represents the computing power node of the computing chip group; represents the computing power node of the computing chip group; represents the computing amount.

2. The method of claim 1, wherein, the delay constraint is a constraint condition constructed based on the relationship between the forwarding time and processing delay corresponding to the computing power task and the tolerable delay of the computing power network; the resource constraint includes a constraint condition constructed based on the relationship between the current load and the preset load, a constraint condition constructed based on the relationship between the network energy consumption and the preset total energy consumption, and a constraint condition constructed based on the relationship between the cache data amount corresponding to the computing power node and the preset cache resource.

3. The method of claim 2, wherein, The computing power task includes a first computing power task set received in a current time slot and a second computing power task set received before the current time slot; the network energy consumption includes the transmission energy consumption and the execution energy consumption corresponding to each computing power task in the first computing power task set and the second computing power task set.

4. The method of claim 2, wherein, The processing delay includes the sum of the task execution delay, the queuing waiting delay and the task transmission delay corresponding to the computing power task; the current load includes the current computing power network load and the current computing power node load.

5. The method of claim 1, wherein, The solving of the task scheduling optimization problem by the preset optimization algorithm to obtain the task scheduling method of the computing power network comprises: initializing a current population; the current population includes a plurality of solutions; based on the current population, calculating the target function to obtain the network energy consumption and the network carbon emission, and based on the constraint condition, obtaining the constraint condition violation condition corresponding to each solution; based on the constraint condition violation condition, performing non-dominated sorting on each solution in the current population to obtain a non-dominated sorting result; based on the network energy consumption and the network carbon emission, determining the crowding distance corresponding to each solution; based on the non-dominated sorting result and the crowding distance corresponding to each solution, determining a candidate solution in the current population; in the case of algorithm convergence, the candidate solution is determined as the task scheduling method.

6. The method of claim 5, wherein, In the case of algorithm non-convergence, the method further comprises: performing cross processing and mutation processing on the candidate solution to generate a child population; after merging the current population and the child population, iterating the current population; performing the step of calculating the target function based on the current population and obtaining the constraint condition violation condition corresponding to each solution based on the constraint condition.

7. The method of claim 6, wherein, Before the cross processing and the variation processing on the candidate solutions are performed to generate a child population, the method further includes dynamically adjusting a cross probability and a variation probability corresponding to the current population based on the network energy consumption and the network carbon emission.

8. A computing power network task scheduling apparatus, comprising a processor, a memory and a computer program / instruction stored on the memory, characterized in that, The processor is configured to execute the computer program / instructions which, when executed, cause the apparatus to perform the steps of the method of any one of claims 1 to 7.

9. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instructions, when executed by the processor, perform the steps of the method of any one of claims 1 to 7.

10. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions, when executed by the processor, perform the steps of the method of any one of claims 1 to 7. The computer program / instructions, when executed by the processor, perform the steps of the method of any one of claims 1 to 7.

Images