Carbon emission evaluation method and device for end-side cloud visual collaborative architecture

Abstract

The invention provides a carbon emission evaluation method and device for an end-side cloud vision collaborative architecture, and the method comprises the steps: building an end equipment carbon emission model, a side equipment carbon emission model and a cloud equipment carbon emission model according to the respective carbon emission parameters of a conventional vision terminal, an intelligent vision terminal, an edge server and a cloud server and functions; a network transmission carbon emission model is constructed according to carbon emission parameters generated due to network transmission during operation of the end-side cloud vision cooperative Internet of Things, and an unloading selection strategy matrix is solved by adopting a Hungary algorithm when the carbon emission of the network transmission carbon emission model is minimum and the computing resource occupied by the vision task is less than or equal to the computing resource upper limit of each device; as a task deployment strategy of a plurality of visual tasks, it is determined that each visual task is executed on the end side, the edge side or the cloud side, the carbon emission generated by executing the visual tasks directly on the cloud side is lower than that generated by executing the visual tasks directly on the cloud side, and the task processing rate is higher than that generated by processing the visual tasks directly on the edge and part of the visual tasks cannot be executed due to edge resource limitation.

Description

Technical Field

[0001] The present invention belongs to the technical field of carbon emission assessment, and specifically relates to a carbon emission assessment method for an end-edge-cloud collaborative architecture, and provides a specific Hungarian algorithm to achieve carbon emission reduction. Background Art

[0002] In recent years, with the continuous development of technologies such as deep learning and big data in the visual field, application fields such as smart transportation and smart security have more efficient visual perception, analysis, prediction and control capabilities. It has effectively improved the operating efficiency and management level of the system under the construction of smart cities. However, this change comes at the cost of increasing computing and communication resources. The higher the complexity of the deep learning model, the greater the amount of computing required, which will lead to huge energy consumption and carbon emissions. According to statistics, initially, the overall training and learning were mainly achieved through cloud servers, integrating terminal data, and the carbon emissions generated by cloud servers accounted for about 15% of the global emissions of the entire information and communication technology (ICT) ecosystem. From 2012 to 2018, the amount of computing used in machine learning training increased by more than 300 times. And centralized training is in cloud service centers, and cooling processing accounts for up to 40% of the energy consumption. And cloud service centers that realize deep learning and commercial operations use 200TWh of energy consumption each year, accounting for 0.3% of global carbon dioxide emissions, while the entire information system accounts for only 2%. Edge computing, as an emerging computing paradigm, is used to replace high-latency, high-carbon-emission cloud computing, which is conducive to offloading computing tasks to resource-rich infrastructure in real time with low carbon emissions. In addition, edge devices do not require a large amount of cooling equipment consumption, and usually choose GPU parallelism to process resource-intensive edge workloads. Compared with CPUs, it increases the operating speed and saves energy, thereby reducing the carbon footprint of cloud service centers. However, the current carbon emission assessment for edge-cloud systems is mainly the carbon emission assessment of cloud data centers, and no specific assessment method is given for the carbon emissions on the end and edge sides.

[0003] Regarding the carbon emission assessment of edge-edge cloud systems, existing methods mainly design the carbon efficiency of data centers through electricity consumption; design carbon-free energy score measurement methods for power plant carbon efficiency and power purchase contracts such as data operators; at the same time, considering the type of energy, give priority to renewable energy and introduce renewable energy indicators to obtain grid carbon intensity and emission source indicators. With the rise of edge computing, in order to meet the latency requirements of terminals, the large-scale computing requirements of edge-edge cloud systems are transferred to edge devices, and the carbon footprint of cloud server centers is also transferred to edge devices. The carbon emissions of edge devices and terminal devices are different from those of cloud devices. Only the energy consumption assessment of carbon emissions and the carbon emission assessment of data centers are used to measure carbon emissions and the granularity of carbon emission nodes. Specifically, the carbon emissions of edge devices and terminal devices are different from those of cloud devices in that:

[0004] Cloud devices: Carbon emissions generated by the operating energy consumption of IT computing equipment, refrigeration systems and basic system equipment, and carbon emissions caused by refrigerant leakage in the refrigeration system.

[0005] Edge devices: Carbon emissions generated by the energy consumption of IT computing equipment and basic system equipment;

[0006] Traditional terminal equipment: carbon emissions generated by the energy consumption of basic system equipment (video acquisition);

[0007] Intelligent visual terminal equipment: Carbon emissions generated by the operating energy consumption of IT computing equipment and basic operating equipment. Summary of the invention

[0008] In order to address the shortcomings of existing technologies, more accurately measure the carbon emissions of the end-edge-cloud collaborative system, and realize the unified measurement of carbon emissions of the end-edge-cloud visual Internet of Things operation in smart cities, an end-edge-cloud system model was constructed based on the visual task offloading strategy. The many-to-many Hungarian matching strategy was adopted to obtain the offloading task decision plan, and decide whether each visual task is executed on the end side, edge side or cloud side. The carbon emissions generated by executing it directly on the cloud side are lower than those generated by executing it directly on the cloud side, and the task processing rate is higher than that of directly processing it on the edge, where some tasks cannot be executed due to resource limitations on the edge.

[0009] In view of the shortcomings of existing technologies, such as Figure 4 As shown, the present invention proposes a carbon emission assessment method for an edge-cloud visual collaborative architecture, which includes:

[0010] The initial step is to obtain the edge-cloud vision collaborative IoT including traditional vision terminals, intelligent vision terminals, edge servers and cloud servers;

[0011] The model building step is to build the end device carbon emission model, edge device carbon emission model, and cloud device carbon emission model according to their functions according to the carbon emission parameters of traditional visual terminals, intelligent visual terminals, edge servers, and cloud servers. According to the carbon emission parameters generated by network transmission when the end-edge-cloud vision collaborative Internet of Things is running, a network transmission carbon emission model is built;

[0012] Task preparation step, obtaining multiple visual tasks, and constructing an offloading selection strategy matrix for the multiple visual tasks according to the deployment strategy of each visual task; constructing a carbon emission matrix corresponding to the offloading selection strategy matrix according to the network transmission carbon emission model, the end device carbon emission model, the edge device carbon emission model and the cloud device carbon emission model, wherein the carbon emission matrix represents the carbon emissions generated by each visual task on each server in the end-edge-cloud visual collaborative Internet of Things; constructing a computing resource load matrix according to the computing resource upper limit of each device in the end-edge-cloud visual collaborative Internet of Things;

[0013] In the task deployment step, the Hungarian algorithm is used to solve the offloading selection strategy matrix when the carbon emissions of the network transmission carbon emission model are minimized and the computing resources occupied by the visual task are less than or equal to the computing resource upper limit of each device. The matrix is ​​used as the task deployment strategy for the multiple visual tasks, and the task is deployed and executed.

[0014] The carbon emission assessment method for the edge-cloud visual collaborative architecture, wherein the task deployment step includes:

[0015] Step 1: Set the carbon emission matrix and computing resource load matrix based on the visual task and offloading selection strategy matrix. Each row in the carbon emission matrix increases in sequence according to the end, edge, and cloud order. When deciding to output the task deployment strategy, it is necessary to ensure that only one element in each row of the task deployment strategy is 1 and meets the computing resource load matrix requirements.

[0016] Step 2: For each visual task, select the edge server with the smallest carbon emissions, subtract the corresponding carbon emissions from the corresponding rows and columns of the carbon emissions matrix, and subtract the corresponding load from the corresponding server in the resource load matrix;

[0017] Step 3: assign the execution device of the visual task according to the 0 value of the carbon emission matrix and the computing resource load matrix. If the assigned 0 value is equal to the total number of the multiple visual tasks, the process ends; otherwise, execute step 2 again.

[0018] The carbon emission assessment method for the edge-cloud visual collaborative architecture, wherein the model building step includes:

[0019] The cloud server includes computing equipment for training, learning and data calculation, cooling system and basic equipment for cloud operation; the energy consumption of all equipment in the entire cloud center is calculated. Divided by the energy consumption of IT equipment operation Get the energy utilization efficiency pue of the cloud server c ;

[0020] The cooling factor of the cloud server is Where b, c and d are real numbers, Tem sup is the temperature set for the cooling equipment. The operating power consumption of the cloud server is p0 = (p c +p c / Cop(Tem sup )); Carbon emissions and carbon intensity of electricity consumption Where G = {1, 2, ... G} and D = {1, 2, ... D} represent the renewable energy source set and regional set respectively; gz represents the power consumption of renewable energy type g in area z∈D; c g is the carbon emission rate of renewable energy type g; the computing resource load rate of the cloud server is; in and are the current load and maximum load of the cloud server respectively; the carbon emissions generated by the operation of the cloud server are:

[0021] According to the carbon emissions caused by the refrigerant leakage of the cloud server and the quality of the refrigerant used, r , global warming potential GWP, refrigerant leakage rate ρ, and the overall operating time of the equipment t cool , building carbon emissions from the use of refrigerants

[0022] The carbon emission model of the cloud equipment is:

[0023] C c (pue,I,p,η,t,q)=pue c I c p0η c 0t0+q r GWPρt cool

[0024] The edge server includes IT equipment for edge computing and system basic operation equipment; the edge-edge-cloud visual collaborative Internet of Things includes K edge servers; the power energy utilization efficiency of each edge server is obtained

[0025] Get the computing power consumption of the edge server's own computing device and running time The computing power consumption of edge servers Power consumption at full load Idle power consumption Current required workload and maximum workload Related; The load rate of the edge server is; The power consumption is: The working and idle time of the edge server is divided into two periods t k e and Since different sites provide different energy mixes to edge servers, the carbon intensity of electricity at each edge Different, that is, the overall calculation model of carbon emissions at edge end k is: The carbon emission model of the edge equipment is:

[0026]

[0027] The traditional visual terminal uses visual perception to compress the captured video and upload it for storage, and performs analysis, recognition and processing on the edge and cloud. The intelligent visual terminal has the ability to extract visual features and store the compressed video stream locally, which can be retrieved on demand based on the needs of the edge and cloud. It also supports the migration and compression of models such as video encoding and feature analysis between the edge and cloud. m Equipment category o m , if it is the traditional visual terminal then o m =0, otherwise o m =1;

[0028] The traditional visual terminal is based on its power consumption and running time As well as the energy type g∈G and carbon intensity of electricity used in the region z∈D Based on the current required workload Energy consumption model of the traditional visual terminal

[0029] The intelligent visual terminal also has data analysis capabilities, and its power energy efficiency is lower than that of edge devices, i.e. According to the power consumption of the intelligent visual terminal Run time Power consumption at full load during operation Idle power consumption Current required workload and maximum workload The load rate of the intelligent visual terminal is; The power consumption of the intelligent visual terminal is The working and idle time of the intelligent visual terminal m is divided into two sections. and The electric carbon intensity of the intelligent visual terminal With the region z∈D and the energy type g∈G of electricity generation, the carbon emission model of the intelligent visual terminal is:

[0030] The carbon emission model of the terminal equipment is:

[0031]

[0032] Obtain the network transmission carbon emissions and communication transmission efficiency ee, the amount of data transmitted d, the energy consumption E and the electric carbon intensity I under the end-edge-cloud collaboration in the end-edge-cloud vision collaborative Internet of Things; obtain the energy consumption υ per bit of data in the end-edge-cloud vision collaborative Internet of Things;

[0033] The amount of data transmitted by the edge-cloud vision collaborative IoT is d m The resolution of the video frame m Related, that is Where σ is a fixed value; and the network transmission energy consumption is related to the energy consumption per bit of data, and the resolution of the video frame is r m The carbon emission model transmitted between node i and node j is: The carbon emissions of network transmission are related to the energy consumption per bit of data, the electric carbon intensity, the video frame resolution, the frame sampling rate, and the communication transmission efficiency. The carbon emission model of network transmission is:

[0034]

[0035] The carbon emission assessment method for the edge-cloud visual collaborative architecture is described, wherein

[0036] The offload selection strategy matrix x is 0 or 1. The decision task is executed locally on the smart vision terminal, on the edge server, or on the cloud server. m is the ID of the terminal device. V is the set of all visual tasks. j = 0 means local unloading on the smart vision terminal, j = [1, 2, ... k, ..., K] means unloading on the edge server, and j = K + 1 means unloading on the cloud server. For edge server k, the load rate of edge computing resources is; For cloud servers, the load rate of computing resources is: Overall carbon emission model:

[0037]

[0038] Among them, the first part is the carbon emission model of local unloading; the second part is the carbon emission model of edge unloading; the third part is the carbon emission model of cloud unloading; and the fourth part is the carbon emission model of network transmission from the terminal to the edge and the cloud.

[0039] like Figure 5 As shown, the present invention also proposes a carbon emission assessment device for the edge-cloud visual collaborative architecture, which includes:

[0040] The initial module obtains the edge-cloud vision collaborative IoT including traditional vision terminals, intelligent vision terminals, edge servers and cloud servers;

[0041] The model building module builds the end device carbon emission model, edge device carbon emission model, and cloud device carbon emission model according to their functions based on the carbon emission parameters of traditional visual terminals, intelligent visual terminals, edge servers, and cloud servers. It also builds the network transmission carbon emission model based on the carbon emission parameters generated by network transmission when the end-edge-cloud vision collaborative Internet of Things is running.

[0042] The task preparation module obtains multiple visual tasks, and constructs an offloading selection strategy matrix for the multiple visual tasks according to the deployment strategy of each visual task; constructs a carbon emission matrix corresponding to the offloading selection strategy matrix according to the network transmission carbon emission model, the end device carbon emission model, the edge device carbon emission model and the cloud device carbon emission model, and the carbon emission matrix represents the carbon emissions generated by each visual task on each server in the end-edge-cloud visual collaborative Internet of Things; constructs a computing resource load matrix according to the computing resource upper limit of each device in the end-edge-cloud visual collaborative Internet of Things;

[0043] The task deployment module adopts the Hungarian algorithm to solve the unloading selection strategy matrix when the carbon emissions of the network transmission carbon emission model are minimized and the computing resources occupied by the visual task are less than or equal to the computing resource upper limit of each device, as the task deployment strategy of the multiple visual tasks, and adopts the task deployment and executes the multiple visual tasks.

[0044] The carbon emission assessment device for the edge-cloud visual collaborative architecture, wherein the task deployment module includes:

[0045] Module 1: Set the carbon emission matrix and computing resource load matrix based on the visual task and offloading selection strategy matrix. Each row in the carbon emission matrix increases in sequence according to the end, edge, and cloud order. When deciding to output the task deployment strategy, it is necessary to ensure that only one element in each row of the task deployment strategy is 1 and meets the computing resource load matrix requirements.

[0046] Module 2: For each visual task, the edge server with the smallest carbon emissions is selected, the corresponding rows and columns of the carbon emissions matrix are subtracted from the corresponding carbon emissions, and the corresponding servers of the resource load matrix are calculated by subtracting the corresponding loads;

[0047] Module 3, according to the carbon emission matrix and the computing resource load matrix of the 0 value, assign the execution device of the visual task, if the assigned 0 value is equal to the total number of tasks of the multiple visual tasks, then end; otherwise, call the module 2 again.

[0048] The carbon emission assessment device for the edge-cloud visual collaborative architecture, wherein the model building module includes:

[0049] The cloud server includes computing equipment for training, learning and data calculation, cooling system and basic equipment for cloud operation; the energy consumption of all equipment in the entire cloud center is calculated. Divided by the energy consumption of IT equipment operation Get the energy utilization efficiency pue of the cloud server c ;

[0050] The cooling factor of the cloud server is Where b, c and d are real numbers, Tem sup is the temperature set for the cooling equipment. The operating power consumption of the cloud server is p0 = (p c +p c / Cop(Tem sup )); Carbon emissions and carbon intensity of electricity consumption Where G = {1, 2, ... G} and D = {1, 2, ... D} represent the renewable energy source set and regional set respectively; gz represents the power consumption of renewable energy type g in area z∈D; c g is the carbon emission rate of renewable energy type g; the computing resource load rate of the cloud server is; in and are the current load and maximum load of the cloud server respectively; the carbon emissions generated by the operation of the cloud server are:

[0051] According to the carbon emissions caused by the refrigerant leakage of the cloud server and the quality of the refrigerant used, r , global warming potential GWP, refrigerant leakage rate ρ, and the overall operating time of the equipment t cool , building carbon emissions from the use of refrigerants

[0052] The carbon emission model of the cloud equipment is:

[0053]

[0054] The edge server includes IT equipment for edge computing and system basic operation equipment; the edge-edge-cloud visual collaborative Internet of Things includes K edge servers; the power energy utilization efficiency of each edge server is obtained

[0055] Get the computing power consumption of the edge server's own computing device and running time The computing power consumption of edge servers Power consumption at full load Idle power consumption Current required workload and maximum workload Related; The load rate of the edge server is; The power consumption is: The time when the edge server is working and idle is divided into two periods and Since different sites provide different energy mixes to edge servers, the carbon intensity of electricity at each edge Different, that is, the overall calculation model of carbon emissions at edge end k is: The carbon emission model of the edge equipment is:

[0056]

[0057] The traditional visual terminal uses visual perception to compress the captured video and upload it for storage, and performs analysis, recognition and processing on the edge and cloud. The intelligent visual terminal has the ability to extract visual features and store the compressed video stream locally, which can be retrieved on demand based on the needs of the edge and cloud. It also supports the migration and compression of models such as video encoding and feature analysis between the edge and cloud. m Equipment category o m , if it is the traditional visual terminal then o m =0, otherwise o m =1;

[0058] The traditional visual terminal is based on its power consumption and running time As well as the energy type g∈G and carbon intensity of electricity used in the region z∈D Based on the current required workload Energy consumption model of the traditional visual terminal

[0059] The intelligent visual terminal also has data analysis capabilities, and its power energy efficiency is lower than that of edge devices, i.e. According to the power consumption of the intelligent visual terminal Run time Power consumption at full load during operation Idle power consumption Current required workload and maximum workload The load rate of the intelligent visual terminal is; The power consumption of the intelligent visual terminal is The working and idle time of the intelligent visual terminal m is divided into two sections. and The electric carbon intensity of the intelligent visual terminal With the region z∈D and the energy type g∈G of electricity generation, the carbon emission model of the intelligent visual terminal is:

[0060] The carbon emission model of the terminal equipment is:

[0061]

[0062] Obtain the network transmission carbon emissions and communication transmission efficiency ee, the amount of data transmitted d, the energy consumption E and the electric carbon intensity I under the end-edge-cloud collaboration in the end-edge-cloud vision collaborative Internet of Things; obtain the energy consumption υ per bit of data in the end-edge-cloud vision collaborative Internet of Things;

[0063] The amount of data transmitted by the edge-cloud vision collaborative IoT is d m The resolution of the video frame m Related, that is Where σ is a fixed value; and the network transmission energy consumption is related to the energy consumption per bit of data, and the resolution of the video frame is r m The carbon emission model transmitted between node i and node j is: The carbon emissions of network transmission are related to the energy consumption per bit of data, the electric carbon intensity, the video frame resolution, the frame sampling rate, and the communication transmission efficiency. The carbon emission model of network transmission is:

[0064]

[0065] The offload selection strategy matrix x is 0 or 1. The decision task is executed locally on the smart vision terminal, on the edge server, or on the cloud server. m is the ID of the terminal device. V is the set of all visual tasks. j = 0 means local unloading on the smart vision terminal, j = [1, 2, ... k, ..., K] means unloading on the edge server, and j = K + 1 means unloading on the cloud server. For edge server k, the load rate of edge computing resources is; For cloud servers, the load rate of computing resources is: Overall carbon emission model:

[0066]

[0067] Among them, the first part is the carbon emission model of local unloading; the second part is the carbon emission model of edge unloading; the third part is the carbon emission model of cloud unloading; and the fourth part is the carbon emission model of network transmission from the terminal to the edge and the cloud.

[0068] The present invention also proposes an electronic device, which includes the carbon emission assessment device for the end-edge-cloud visual collaborative architecture. The electronic device may be connected to an information display device, and the information display device is used to display the assessment results according to display parameters and attributes set by the user or through an artificial intelligence model.

[0069] The present invention also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the carbon emission assessment method for the end-edge-cloud visual collaborative architecture are implemented.

[0070] The present invention also proposes a computer program product, including a computer program, wherein when the computer program is executed by a processor, it implements the steps of any of the carbon emission assessment methods for the end-edge-cloud visual collaborative architecture. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is the end-edge-cloud collaborative carbon emission architecture diagram of the present invention;

[0072] Figure 2 This is a parameter diagram related to the end-edge-cloud collaborative carbon emission of the present invention;

[0073] Figure 3 This is a process diagram of carbon emission assessment under the end-edge-cloud collaboration of the present invention;

[0074] Figure 4 is a flow chart of the method of the present invention;

[0075] Figure 5 It is a module diagram of the device of the present invention;

[0076] Figure 6 This is a schematic diagram of the structure of a first electronic device of the present invention;

[0077] Figure 7 This is a schematic diagram of the application environment structure of the first electronic device of the present invention;

[0078] Figure 8 It is a schematic diagram of the structure of a second electronic device of the present invention.

[0079] Reference numerals:

[0080] A-First electronic device;

[0081] B-Carbon emission assessment device for edge-cloud visual collaborative architecture;

[0082] C-data acquisition equipment;

[0083] D-information display device;

[0084] 1000 - second electronic device;

[0085] Ⅰ-computational unit;

[0086] II-ROM;

[0087] III-RAM;

[0088] IV-bus;

[0089] V-interface;

[0090] VI - input unit;

[0091] VII-output unit;

[0092] VIII- Storage medium;

[0093] Ⅸ-Communication unit. DETAILED DESCRIPTION

[0094] It should be noted that, in this application, relational terms such as first and second, etc. 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", "comprises" 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.

[0095] Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0096] The processor described in the present invention is the control center of the electronic device, which can be a processor or a general term for multiple processing elements. For example, it can be one or more central processing units (CPUs), or application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs).

[0097] Optionally, the processor can perform various functions of the electronic device by running or executing a software program stored in the memory, and calling data stored in the memory.

[0098] In a specific implementation, as an embodiment, the processor may include one or more CPUs. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions). Electronic devices may include: servers, desktop computers, laptops, smart phones, tablet computers, embedded computers, etc., wherein the embedded computers include vehicles and robots, etc.

[0099] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment, which will not be repeated here.

[0100] It should be noted that the structure of the electronic device shown in the drawings of the present invention does not constitute a limitation thereto, and the actual knowledge structure recognition device may include more or fewer components than shown in the drawings, or a combination of certain components, or a different arrangement of components.

[0101] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0102] It should also be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0103] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0104] It should also be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0105] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0106] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0107] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0108] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program codes.

[0109] The present invention adopts the following technical solution:

[0110] First, based on the edge-cloud vision system, a carbon emission assessment model for the edge-cloud system is constructed. Figure 1The edge-cloud vision architecture shown. At the cloud level, the carbon emissions of the system are related to the energy consumption of cloud computing, the operating energy consumption of the refrigeration system, the carbon emissions released by the leakage of refrigerants, and the carbon emissions generated by the energy consumption of infrastructure operation; at the edge level, the carbon emissions of system operation are related to the energy consumption of edge computing and the energy consumption of infrastructure operation; and at the terminal level, it is divided into traditional vision terminals and intelligent vision terminals. Here, it is assumed that the system operation energy consumption of traditional vision terminals is determined by their own power consumption and time; and intelligent vision terminals have an additional end-computing operation energy consumption compared to traditional vision terminals. Therefore, the carbon emissions generated are evaluated through the cloud layer, edge layer, terminal layer, and network layer respectively.

[0111] Cloud carbon emission assessment: Figure 2 The parameters involved in cloud carbon emissions calculation include: carbon emissions assessment of cloud services and power usage effectiveness (PUE) of cloud devices. c 、Electricity Carbon Intensity of Cloud DevicesI c , evaluate the running power consumption p0 of the equipment during the calculation phase, the running time t0, and the load rate of the computing resources in the cloud service center and the quality q and usage time t of the refrigerant cool The calculation parameters of the cloud device's operating power consumption p0 include: the computing power consumption of the cloud server p c and the temperature Tem set by the refrigeration equipment, etc. sup The carbon intensity of electricity generated by cloud devices is related to the type of renewable energy g and the region z, as well as the carbon emission rate c of the renewable energy type g. g The carbon emissions caused by refrigerant leakage are related to the mass of refrigerant used (kg) q, global warming potential (GWP), refrigerant leakage rate (% / year) ρ, and working time period (year) t cool Related.

[0112] Based on the types of energy currently used for power generation: for example: coal, nuclear energy, coal, natural gas, oil, hydropower, wind power, etc. Cloud power generation energy: Due to the huge amount of electricity required, the energy used in this model includes coal, nuclear energy, coal, natural gas, oil, hydropower, wind power, etc. These all have statistics on the carbon intensity of carbon emissions. Edge side: It can use commercial electricity like the cloud, or it can be powered by small power generation equipment and use other green energy. End side: All commercial electricity energy is used.

[0113] Carbon emission assessment of marginal layer: Figure 2 ,The parameters involved in the calculation of carbon emissions at the edge layer include: the power energy efficiency of each edge end Carbon intensity of electricity at the edge Computing power consumption of edge devices Full load power consumption when the device is running Idle power consumption Run time Idle runtime Edge computing resource load Since the overall operating energy consumption of the edge server is related to the visual tasks being calculated and the operation of the basic equipment. Here, the computing power consumption of the edge equipment is The parameters include: full load power consumption when the device is running Idle power consumption and the load rate of edge computing resources The energy consumption of edge devices is provided by mixed energy from different sites. Unlike the cloud layer, the carbon intensity of electricity is obtained based on the type of energy used.

[0114] End-level carbon emissions assessment: Figure 2 , the parameters involved in the terminal layer carbon emission calculation include: the device category of the visual terminal is o m , terminal electricity carbon intensity Calculating Power Consumption Full load power consumption of smart terminal equipment during operation and idle power consumption Run time Idle runtime Intelligent terminal computing resource load rate The terminal carbon intensity Calculating Power Consumption and running time It is the carbon emission calculation parameter of traditional terminals. The computing power consumption of intelligent visual terminals Parameters are similar to edge devices, including full load power consumption Idle power consumption and running time Idle runtime Intelligent terminal computing resource load rate In addition, the electrical carbon intensity of the terminal layer Similar to the cloud.

[0115] Network layer carbon emission assessment: Figure 2 , the parameters involved in the calculation of carbon emissions at the network layer and the energy consumption per bit of data ij , communication transmission efficiency ee ij , the resolution of the video frame m , frame sampling rate fs m , and electrical carbon intensity I m The resolution of the video frame is rm Determines the amount of data on the transmission path.

[0116] The second aspect is the carbon emission assessment solution under the collaboration of end-edge-cloud. Figure 3 In the process shown, a carbon emission model scheme based on a single device\subsystem is constructed according to the carbon emission assessment model of the end-edge-cloud system; and an overall system carbon emission assessment model is constructed through the end-edge-cloud collaborative task offloading model, and a many-to-many Hungarian matching strategy is adopted to obtain a suitable end-edge-cloud collaborative offloading strategy.

[0117] The advantages and beneficial effects of the present invention are: based on the carbon footprint under the visual Internet of Things architecture, the carbon emissions of each subsystem device of the end-edge-cloud under the architecture are modeled and measured in a unified manner; at the same time, a network-level carbon emission assessment model for data transmission of each subsystem device under end-edge-cloud collaboration is provided; and taking into account the diversity of large-scale visual tasks, a carbon emission optimization model for the end-edge-cloud collaborative system is constructed, and a many-to-many Hungarian matching algorithm is provided to determine whether each visual task is executed locally, at the edge or in the cloud.

[0118] In order to make the above features and effects of the present invention more clearly and understandably described, embodiments are given below and described in detail with reference to the accompanying drawings. This specification discloses one or more embodiments that include the features of the present invention. The disclosed embodiments are only for illustration. The scope of protection of the present invention is not limited to the disclosed embodiments, and the present invention is defined by the attached claims.

[0119] The present invention mainly describes and illustrates the carbon emissions of different devices through four parts: cloud layer carbon emission assessment, edge layer carbon emission assessment, terminal layer carbon emission assessment and network layer carbon emission assessment, and proposes a terminal-edge-cloud collaborative carbon emission assessment technology based on the carbon emission assessment model.

[0120] 1. Carbon Emission Assessment Model for Edge-Cloud Systems

[0121] like Figure 1 The present invention is a specific edge-cloud collaborative visual IoT architecture. The system network consists of M traditional visual terminal devices, N intelligent visual terminal devices, a cloud server, K edge servers and several infrastructures. The terminal set is defined as V = {V1,...,V m ,...,V M ,V M+1 ,...V M+N}, the server set is defined as S = {S0,...,S k ,...S K}, where S0 is the cloud server. Terminals are divided into two categories: traditional visual terminals are responsible for video collection, acquisition, encoding and transmission; in addition to the above functions, smart terminals also include functions such as video data analysis, processing and control. The edge server is responsible for the analysis and processing of local data, while transmitting some unprocessed data to the cloud and returning relevant results to the client. In addition to analyzing and processing the data of the entire system globally, the cloud server also needs to implement control functions such as command distribution for the entire system. Due to the different functions and basic operating equipment contained in each node, the required energy consumption and corresponding carbon emissions are different. The following details the carbon emission assessment methods for different end, edge and cloud devices and network transmission between nodes, so as to achieve the overall carbon emission assessment and prediction under the operation of the entire system.

[0122] (1) Assessment of cloud carbon emissions

[0123] Cloud service equipment includes IT equipment for training, learning and data computing, cooling systems and basic equipment for cloud operation. Therefore, in the edge-cloud system, the carbon footprint of the cloud service center should first consider the power usage effectiveness (PUE). c Equal to the energy consumption of all devices in the entire cloud center Divided by the energy consumption of IT equipment operation The power consumption p0 and operating time t0 of the equipment in different evaluation time periods are considered to obtain the power energy utilization efficiency PUE.

[0124] In addition to considering the power consumption p0 and running time t0 of the IT equipment itself, the power consumption and running time of air conditioners and other equipment must also be considered. Here, the power consumption of cloud server IT equipment is defined to be related to the cooling coefficient. The cooling coefficient is Where b, c and d are heterogeneous real numbers, Tem sup The actual operating power consumption of the cloud server is p0 = (p c +p c / Cop(Tem sup )). The carbon emissions of cloud devices and the carbon intensity of electricity consumed Where G = {1, 2, ... G} and D = {1, 2, ... D} represent the renewable energy source set and regional set respectively; gz represents the power consumption of renewable energy type g in area z∈D; c g is the carbon emission rate of renewable energy type g. Let the load rate of computing resources in the cloud service center be; in and are the current required load and the maximum load of the cloud service center. Therefore, the carbon emissions generated by the energy consumption of cloud equipment operation are:

[0125] In addition, since the cloud uses cooling equipment such as air conditioners to maintain the operation of IT equipment such as servers and ensure the safety of cloud equipment, there is a risk of carbon emissions from refrigerant leakage in addition to the carbon emissions generated by electricity consumption. Carbon emissions generated by refrigerant leakage and the mass of refrigerant used (kg)q r , global warming potential (GWP), refrigerant leakage rate (% / year) ρ, and overall operating time of the equipment (year) t cool The calculation formula for carbon emissions caused by the use of refrigerants is:

[0126] Therefore, the assessment of cloud carbon emissions includes carbon emissions from energy consumption. Carbon emissions from the use of The parameters of power consumption / time / carbon intensity index / refrigerant usage / refrigerant leakage rate are used for measurement; among them, power consumption, time, carbon intensity index, etc. determine the carbon emission model of energy consumption. The usage of refrigeration equipment, the refrigerant leakage rate and the refrigerant usage period determine the carbon emission model caused by the use of refrigerants. The overall cloud carbon emissions assessment model is:

[0127]

[0128] (2) Carbon emission assessment of marginal layer

[0129] The edge device center mainly includes IT equipment for edge computing and basic system operation equipment. However, for the K edge devices under the edge-cloud collaborative visual IoT, the edge carbon emissions evaluated here are the sum of the carbon emissions of all edge devices. Similar to cloud servers, the first thing to consider is the power energy efficiency of each edge device. And because there is less related cooling equipment compared to the cloud service center, the PUE is reduced here, that is,

[0130] The overall carbon emissions of edge operations mainly consider the computing power consumption of its own IT equipment. and running time The computing power consumption of edge servers The device consumes power when fully loaded. Idle power consumption Current required workload and maximum workload Related. Let the load rate of edge computing resources be; That is, the current computing power consumption is: Therefore, the working and idle time of edge k is divided into two periods. and Since different sites provide different energy mixes to edge servers, the carbon intensity of electricity at each edge Different, that is, the overall calculation model of carbon emissions at edge end k is: The carbon emission calculation model of the entire edge is:

[0131]

[0132] (3) Carbon emission assessment at the terminal level

[0133] The terminal defined in this invention is a visual device, which is divided into two categories: traditional terminals and smart terminals. Traditional cameras use visual perception to compress the captured video and upload it for storage, and perform analysis, recognition and processing on the edge and cloud. Smart cameras have the ability to encode and extract visual features with high quality, store the compressed video stream locally, and retrieve it on demand based on the needs of the edge and cloud. They also support the migration and compression of models such as video encoding and feature analysis between the edge and cloud. Here, the visual terminal v m The device category is o m , if it is a traditional terminal o m =0, otherwise o m =1.

[0134] For traditional terminals, the electricity consumption is generally based on four categories: residents, agriculture, large industry, and general industry and commerce. Therefore, the power consumption here is mainly calculated based on the camera. and running time and the energy type g∈G and carbon intensity of electricity used in region z∈D Related, and the current required workload is Right now

[0135] In addition to visual perception, smart terminals also have certain data analysis capabilities. Therefore, they also have a certain power energy utilization efficiency (PUE), which is lower than that of edge devices. Therefore, the smart terminal can be regarded as a small computing node that can be analyzed. That is, the carbon emissions of the operation of the smart terminal mainly consider its own computing power consumption. and running time At the same time, similar to the edge, the computing power consumption of smart terminals The device consumes power when fully loaded. Idle power consumption Current required workload and maximum workload Related. Let the load rate of intelligent terminal computing resources be; That is, the current computing power consumption is: Therefore, the working and idle time of the intelligent terminal m can also be divided into two sections. and Similar to traditional terminals, smart terminals also use sold electricity, so the carbon intensity of electricity is It is related to the region z∈D and the energy type g∈G of electricity generation. That is, the carbon emission model is:

[0136] From the above, we can see that the carbon emission model of all terminals is:

[0137]

[0138] (4) Network-level carbon emissions assessment

[0139] The end-edge-cloud collaborative visual IoT architecture realizes the whole process of data transmission, cleaning, synchronization and retrieval through the collaboration between the end, edge and cloud. Since 2010, global Internet traffic has increased 25 times. There is no doubt that the energy consumption demand of the data transmission network has increased, and the resulting carbon emissions have also increased significantly. The carbon emissions of network transmission under the end-edge-cloud collaboration are related to the communication transmission efficiency ee, the amount of data transmitted d, the energy consumption E consumed by transmission and the electric carbon intensity I. The communication transmission efficiency ee and the energy consumption E consumed by transmission are both related to the transmission mode (such as wireless network transmission and core network transmission). The communication efficiency of wireless network transmission is worse than that of the core network, and the energy consumption is higher. Different transmission modes have different energy consumption υ (J / bit) per bit of data, and the carbon emissions caused are different.

[0140] The amount of data transmitted over the network under the visual Internet of Things m The resolution of the video frame m Related, that is Where σ is a fixed value. The network transmission energy consumption is related to the energy consumption per bit of data υ (J / bit). Depending on the node, the resolution of the video frame r m The carbon emission model transmitted between node i and node j is: The carbon emissions of network transmission are related to the energy consumption per bit of data, the electric carbon intensity, the video frame resolution, the frame sampling rate, and the communication transmission efficiency. The overall calculation model is:

[0141]

[0142] 2. Carbon Emission Assessment Solution under Device-Edge-Cloud Collaboration

[0143] Based on the constructed end-edge-cloud carbon emission assessment method, the present invention proposes a carbon emission assessment technology under end-edge-cloud collaboration, which is applied to the end-edge-cloud collaborative vision field. The specific steps are as follows:

[0144] (1) Carbon emission model measurement of a single subsystem or equipment

[0145] First, at the terminal, carbon emission calculation is based on local calculation and quantified conversion parameters to achieve the carbon emission model design at the terminal layer. m Equipment category o m , if it is a traditional terminal o m =0, otherwise it is an intelligent visual terminal, o m = 1. Analyze the parameters and characteristics of the terminal in each time period, that is, the task requirement fs m , r m and d m Etc., clarify the carbon emission model of the task executed locally Terminal tasks that cannot be processed locally are uploaded to the corresponding edge server or cloud server based on the nearest processing and latency requirements. The latency requirement is the task requirement of the terminal task itself, or the latency requirement is the maximum latency of the task in the current market obtained based on the task scenario and type.

[0146] Next, at the edge or cloud, based on the video / compressed video / video features data obtained from the terminal device, specific visual tasks and requirements are implemented, such as face recognition, vehicle counting statistics, deep learning training, etc. This involves the carbon emission model evaluation of edge and cloud devices and the carbon emission evaluation of the edge-cloud collaborative network. The carbon emission model C of the cloud service center and each edge service center at this stage is recorded according to formulas (1) and (2): c and At the same time, the carbon emission model under each transmission path is recorded according to formula (4): The time is adjusted according to actual needs.

[0147] (2) Carbon emission model of edge-cloud collaborative system

[0148] Based on the carbon emission model of a single subsystem and device, under the edge-cloud collaborative visual Internet of Things, a new paradigm for optimizing system carbon emissions is established based on the resource consumption under this architecture, and the overall deployment strategy of the edge-cloud system is reformed and innovated. First, set the two-dimensional offloading decision variable x is 0 or 1, and the decision task is executed locally, on edge 1, edge 2, or on the cloud; m is the id of the terminal device; V is the set of all terminal tasks. Among them, j = 0 means local offloading, j = [1, 2, ... k, ..., K] means offloading on the edge, and j = K + 1 means offloading on the cloud. Then for edge server k, the load rate of edge computing resources is; For cloud servers, the load rate of computing resources is: Under the condition that the network service quality and visual task latency are not affected, the collaborative optimization of visual task offloading is completed. Under the adaptive adjustment of different terminal visual task offloading schemes, the carbon emission target can be achieved. The overall carbon emission model is designed as follows:

[0149]

[0150] Among them, the first part is the carbon emission model of local unloading; the second part is the carbon emission model of edge unloading; the third part is the carbon emission model of cloud unloading; and the fourth part is the carbon emission model of network transmission from the terminal to the edge and the cloud.

[0151] (3) Carbon emission deployment plan for edge-cloud collaborative system

[0152] The above carbon emission model can be regarded as a many-to-many matching problem. The Hungarian algorithm, as a classic combinatorial optimization algorithm, can solve the allocation and deployment problem in polynomial time. Here, the Hungarian algorithm is used as a solution for the carbon emission deployment of the edge-cloud collaborative system. The details are as follows:

[0153] Step 1: First, consider the cloud as a matching node based on the near edge, and select the strategy matrix X based on the terminal vision task and offloading (M+N)·(K+2) Set different ID tags (carbon emissions and computing resource load) to generate the carbon emission matrix C (M+N)·(K+2) and compute resource load matrix The carbon emission matrix of each row increases in the order of end, edge and cloud. (M+N)·(K+2) When, it is necessary to ensure that each row of the carbon emission matrix C (M+N)·(K+2) Only one element is 0, the decision output matrix X (M+N)·(K+2) Each row of has only one element that is 1 and satisfies the computational resource load matrix Demand. It should be noted that the 0 in the carbon emission matrix corresponds to the 1 in the decision matrix x.

[0154] Step 2: Then according to the row and column reduction, the service node selected by the visual task m corresponding to the minimum carbon emission is j, and the minimum carbon emission is C min =C mj ; The carbon emission matrix as a whole minus C min, calculate the resource load matrix corresponding to the server minus the corresponding load, that is

[0155] Step 3: Based on the carbon emission matrix C (M+N)·(K+2) All 0 values ​​and computing resource load matrix If the assigned independent 0 value is equal to the number of visual tasks, the task ends. At this time, the carbon emission matrix C (M+N)·(K+2) The row and column where 0 is located is the decision matrix X (M+N)·(K+2) 1; otherwise, adjust the carbon emission matrix and computing resource load matrix and return to Step 2.

[0156] The following is a system embodiment corresponding to the above method embodiment. This embodiment can be implemented in conjunction with the above embodiment. The relevant technical details mentioned in the above embodiment are still valid in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiment.

[0157] like Figure 5 As shown, the present invention also proposes a carbon emission assessment device for the edge-cloud visual collaborative architecture, which includes:

[0158] The initial module obtains the edge-cloud vision collaborative IoT including traditional vision terminals, intelligent vision terminals, edge servers and cloud servers;

[0159] The model building module builds the end device carbon emission model, edge device carbon emission model, and cloud device carbon emission model according to their functions based on the carbon emission parameters of traditional visual terminals, intelligent visual terminals, edge servers, and cloud servers. It also builds the network transmission carbon emission model based on the carbon emission parameters generated by network transmission when the end-edge-cloud vision collaborative Internet of Things is running.

[0160] The task preparation module obtains multiple visual tasks, and constructs an offloading selection strategy matrix for the multiple visual tasks according to the deployment strategy of each visual task; constructs a carbon emission matrix corresponding to the offloading selection strategy matrix according to the network transmission carbon emission model, the end device carbon emission model, the edge device carbon emission model and the cloud device carbon emission model, and the carbon emission matrix represents the carbon emissions generated by each visual task on each server in the end-edge-cloud visual collaborative Internet of Things; constructs a computing resource load matrix according to the computing resource upper limit of each device in the end-edge-cloud visual collaborative Internet of Things;

[0161] The task deployment module adopts the Hungarian algorithm to solve the unloading selection strategy matrix when the carbon emissions of the network transmission carbon emission model are minimized and the computing resources occupied by the visual task are less than or equal to the computing resource upper limit of each device, as the task deployment strategy of the multiple visual tasks, and adopts the task deployment and executes the multiple visual tasks.

[0162] The carbon emission assessment device for the edge-cloud visual collaborative architecture, wherein the task deployment module includes:

[0163] Module 1: Set the carbon emission matrix and computing resource load matrix based on the visual task and offloading selection strategy matrix. Each row in the carbon emission matrix increases in sequence according to the end, edge, and cloud order. When deciding to output the task deployment strategy, it is necessary to ensure that only one element in each row of the task deployment strategy is 1 and meets the computing resource load matrix requirements.

[0164] Module 2: For each visual task, the edge server with the smallest carbon emissions is selected, the corresponding rows and columns of the carbon emissions matrix are subtracted from the corresponding carbon emissions, and the corresponding servers of the resource load matrix are calculated by subtracting the corresponding loads;

[0165] Module 3, according to the carbon emission matrix and the computing resource load matrix of the 0 value, assign the execution device of the visual task, if the assigned 0 value is equal to the total number of tasks of the multiple visual tasks, then end; otherwise, call the module 2 again.

[0166] The carbon emission assessment device for the edge-cloud visual collaborative architecture, wherein the model building module includes:

[0167] The cloud server includes computing equipment for training, learning and data calculation, cooling system and basic equipment for cloud operation; the energy consumption of all equipment in the entire cloud center is calculated. Divided by the energy consumption of IT equipment operation Get the energy utilization efficiency pue of the cloud server c ;

[0168] The cooling factor of the cloud server is Where b, c and d are real numbers, Tem sup is the temperature set for the cooling equipment. The operating power consumption of the cloud server is p0 = (p c +p c / Cop(Tem sup )); Carbon emissions and carbon intensity of electricity consumption Where G = {1, 2, ... G} and D = {1, 2, ... D} represent the renewable energy source set and regional set respectively; gz represents the power consumption of renewable energy type g in area z∈D; c gis the carbon emission rate of renewable energy type g; the computing resource load rate of the cloud server is; in and are the current load and maximum load of the cloud server respectively; the carbon emissions generated by the operation of the cloud server are:

[0169] According to the carbon emissions caused by the refrigerant leakage of the cloud server and the quality of the refrigerant used, r , global warming potential GWP, refrigerant leakage rate ρ, and the overall operating time of the equipment t cool , building carbon emissions from the use of refrigerants

[0170] The carbon emission model of the cloud equipment is:

[0171]

[0172] The edge server includes IT equipment for edge computing and system basic operation equipment; the edge-edge-cloud visual collaborative Internet of Things includes K edge servers; the power energy utilization efficiency of each edge server is obtained

[0173] Get the computing power consumption of the edge server's own computing device and running time The computing power consumption of edge servers Power consumption at full load Idle power consumption Current required workload and maximum workload Related; The load rate of the edge server is; The power consumption is: The working and idle time of the edge server is divided into two periods t k e and Since different sites provide different energy mixes to edge servers, the carbon intensity of electricity at each edge Different, that is, the overall calculation model of carbon emissions at edge end k is: The carbon emission model of the edge equipment is:

[0174]

[0175] The traditional visual terminal uses visual perception to compress the captured video and upload it for storage, and performs analysis, recognition and processing on the edge and cloud. The intelligent visual terminal has the ability to extract visual features and store the compressed video stream locally, which can be retrieved on demand based on the needs of the edge and cloud. It also supports the migration and compression of models such as video encoding and feature analysis between the edge and cloud. m Equipment category o m , if it is the traditional visual terminal then o m =0, otherwise o m =1;

[0176] The traditional visual terminal is based on its power consumption and running time As well as the energy type g∈G and carbon intensity of electricity used in the region z∈D Based on the current required workload Energy consumption model of the traditional visual terminal

[0177] The intelligent visual terminal also has data analysis capabilities, and its power energy efficiency is lower than that of edge devices, i.e. According to the power consumption of the intelligent visual terminal Run time Power consumption at full load during operation Idle power consumption Current required workload and maximum workload The load rate of the intelligent visual terminal is; The power consumption of the intelligent visual terminal is The working and idle time of the intelligent visual terminal m is divided into two sections. and The electric carbon intensity of the intelligent visual terminal With the region z∈D and the energy type g∈G of electricity generation, the carbon emission model of the intelligent visual terminal is:

[0178] The carbon emission model of the terminal equipment is:

[0179]

[0180] Obtain the network transmission carbon emissions and communication transmission efficiency ee, the amount of data transmitted d, the energy consumption E and the electric carbon intensity I under the end-edge-cloud collaboration in the end-edge-cloud vision collaborative Internet of Things; obtain the energy consumption υ per bit of data in the end-edge-cloud vision collaborative Internet of Things;

[0181] The amount of data transmitted by the edge-cloud vision collaborative IoT is d m The resolution of the video frame mRelated, that is Where σ is a fixed value; and the network transmission energy consumption is related to the energy consumption per bit of data, and the resolution of the video frame is r m The carbon emission model transmitted between node i and node j is: The carbon emissions of network transmission are related to the energy consumption per bit of data, the electric carbon intensity, the video frame resolution, the frame sampling rate, and the communication transmission efficiency. The carbon emission model of network transmission is:

[0182]

[0183] The offload selection strategy matrix x is 0 or 1. The decision task is executed locally on the smart vision terminal, on the edge server, or on the cloud server. m is the ID of the terminal device. V is the set of all visual tasks. j = 0 means local unloading on the smart vision terminal, j = [1, 2, ... k, ..., K] means unloading on the edge server, and j = K + 1 means unloading on the cloud server. For edge server k, the load rate of edge computing resources is; For cloud servers, the load rate of computing resources is: Overall carbon emission model:

[0184]

[0185] Among them, the first part is the carbon emission model of local unloading; the second part is the carbon emission model of edge unloading; the third part is the carbon emission model of cloud unloading; and the fourth part is the carbon emission model of network transmission from the terminal to the edge and the cloud.

[0186] like Figure 6 As shown, the present invention further proposes a first electronic device A in another embodiment, which includes the carbon emission assessment device for the end-edge-cloud visual collaborative architecture.

[0187] like Figure 7 As shown, the first electronic device A can also be connected to the data acquisition device C and the information display device D through a wired or wireless information transmission scheme. The data acquisition device C is used to collect and obtain multiple visual tasks, such as semantic segmentation tasks and image classification tasks. The information display device D is used to display the task deployment strategy and task execution results analyzed by the present invention.

[0188] The information display device D can sort and process the data output by the first electronic device A based on the information display mechanism to improve the readability of the data output by the first electronic device A. The information display mechanism can be manually preset, for example, the data output by the first electronic device A is visually displayed, which can be based on the display parameters and / or attributes set by the user. The display parameters can be, for example, the display data range, and the display attributes can be, for example, the display font, color, whether to scroll and play, etc. The user is presented with the key information specified by the user, and the user can understand the information more promptly without having to access the secondary page or scroll the page, saving the user's operation. Or the information display mechanism can be an artificial intelligence AI display model, which can learn the user's key information based on the user's previous usage habits, such as viewing time, number of clicks, number of edits, etc., and then automatically present rich and necessary key information to the user.

[0189] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a readable storage medium. When the computer program is executed by a processor, the computer can execute the carbon emission assessment method for the end-edge-cloud visual collaborative architecture provided by the above methods.

[0190] The present invention also proposes a storage medium VIII in another embodiment for storing a computer program for executing the carbon emission assessment method for the edge-cloud visual collaborative architecture. It should be understood that the storage medium in the embodiment of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus random access memory (DR RAM).

[0191] Figure 8 A schematic block diagram of a second electronic device 1000 that can be used to implement an embodiment of the present invention is shown. The second electronic device 1000 electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The second electronic device 1000 can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples, and are not intended to limit the implementation of the present invention described and / or required herein. The second electronic device 1000 may be the same or different from the first electronic device A.

[0192] The second electronic device 1000 includes a computing unit I, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory II (ROM) or a computer program loaded from a storage medium VIII into a random access memory (RAM) III. In RAM III, various programs and data required for the operation of the device 1000 can also be stored. The computing unit I, ROM II, and RAM III are connected to each other via a bus IV. An input / output (I / O) interface V is also connected to the bus IV.

[0193] A plurality of components in the second electronic device 1000 are connected to the I / O interface V, including: an input unit VI, such as a keyboard, a mouse, etc.; an output unit VII, such as various types of displays, speakers, etc.; a storage medium VIII, such as a disk, an optical disk, etc.; and a communication unit IX, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit IX allows the second electronic device 1000 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0194] The computing unit I may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of computing unit I include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit I performs the various methods and processes described above, such as method steps S1-S4. For example, in some embodiments, the method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage medium VIII. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 1000 via ROM II and / or a communication unit IX. When the computer program is loaded into RAM III and executed by the computing unit I, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit I may be configured to execute the method in any other appropriate manner (e.g., by means of firmware).

[0195] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A carbon emission assessment method for edge-cloud visual collaborative architecture, characterized in that: include: The initial step is to obtain the edge-cloud vision collaborative IoT including traditional vision terminals, intelligent vision terminals, edge servers and cloud servers; The model building step is to build the end device carbon emission model, edge device carbon emission model, and cloud device carbon emission model according to their functions according to the carbon emission parameters of traditional visual terminals, intelligent visual terminals, edge servers, and cloud servers. According to the carbon emission parameters generated by network transmission when the end-edge-cloud vision collaborative Internet of Things is running, a network transmission carbon emission model is built; Task preparation step, obtaining multiple visual tasks, and constructing an offloading selection strategy matrix for the multiple visual tasks according to the deployment strategy of each visual task; constructing a carbon emission matrix corresponding to the offloading selection strategy matrix according to the network transmission carbon emission model, the end device carbon emission model, the edge device carbon emission model and the cloud device carbon emission model, wherein the carbon emission matrix represents the carbon emissions generated by each visual task on each server in the end-edge-cloud visual collaborative Internet of Things; constructing a computing resource load matrix according to the computing resource upper limit of each device in the end-edge-cloud visual collaborative Internet of Things; In the task deployment step, the Hungarian algorithm is used to solve the offloading selection strategy matrix when the carbon emissions of the network transmission carbon emission model are minimized and the computing resources occupied by the visual task are less than or equal to the computing resource upper limit of each device. The matrix is ​​used as the task deployment strategy for the multiple visual tasks, and the task is deployed and executed.

2. The carbon emission assessment method for the edge-cloud visual collaborative architecture according to claim 1 is characterized in that: The task deployment steps include: Step 1: Set the carbon emission matrix and computing resource load matrix based on the visual task and offloading selection strategy matrix. Each row in the carbon emission matrix increases in sequence according to the end, edge, and cloud order. When deciding to output the task deployment strategy, it is necessary to ensure that only one element in each row of the task deployment strategy is 1 and meets the computing resource load matrix requirements. Step 2: For each visual task, select the edge server with the smallest carbon emissions, subtract the corresponding carbon emissions from the corresponding rows and columns of the carbon emissions matrix, and subtract the corresponding load from the corresponding server in the resource load matrix; Step 3: assign the execution device of the visual task according to the 0 value of the carbon emission matrix and the computing resource load matrix. If the assigned 0 value is equal to the total number of the multiple visual tasks, the process ends; otherwise, execute step 2 again.

3. The carbon emission assessment method for the edge-cloud visual collaborative architecture according to claim 1 or 2, characterized in that: The model building steps include: The cloud server includes computing equipment for training, learning and data calculation, cooling system and basic equipment for cloud operation; the energy consumption of all equipment in the entire cloud center is calculated. Divided by the energy consumption of IT equipment operation Get the energy utilization efficiency pue of the cloud server c ; The cooling factor of the cloud server is Where b, c and d are real numbers, Tem sup is the temperature set for the cooling equipment. The operating power consumption of the cloud server is p0 = (p c +p c / Cop(Tem sup )); Carbon emissions and carbon intensity of electricity consumption Where G = {1, 2, ... G} and D = {1, 2, ... D} represent the renewable energy source set and regional set respectively; gz represents the power consumption of renewable energy type g in area z∈D; c g is the carbon emission rate of renewable energy type g; the computing resource load rate of the cloud server is; in and are the current load and maximum load of the cloud server respectively; the carbon emissions generated by the operation of the cloud server are: According to the carbon emissions caused by the refrigerant leakage of the cloud server and the quality of the refrigerant used, r , global warming potential GWP, refrigerant leakage rate ρ, and the overall operating time of the equipment t cool , building carbon emissions from the use of refrigerants The carbon emission model of the cloud equipment is: The edge server includes IT equipment for edge computing and system basic operation equipment; the edge-edge-cloud visual collaborative Internet of Things includes K edge servers; the power energy utilization efficiency of each edge server is obtained Get the computing power consumption of the edge server's own computing device and running time The computing power consumption of edge servers Power consumption at full load Idle power consumption Current required workload and maximum workload Related; The load rate of the edge server is; The power consumption is: The time when the edge server is working and idle is divided into two periods and Since different sites provide different energy mixes to edge servers, the carbon intensity of electricity at each edge Different, that is, the overall calculation model of carbon emissions at edge end k is: The carbon emission model of the edge equipment is: The traditional visual terminal uses visual perception to compress the captured video and upload it for storage, and performs analysis, recognition and processing on the edge and cloud. The intelligent visual terminal has the ability to extract visual features and store the compressed video stream locally, which can be retrieved on demand based on the needs of the edge and cloud. It also supports the migration and compression of models such as video encoding and feature analysis between the edge and cloud. m Equipment category o m , if it is the traditional visual terminal then o m =0, otherwise o m =1; The traditional visual terminal is based on its power consumption and running time As well as the energy type g∈G and carbon intensity of electricity used in the region z∈D Based on the current required workload Energy consumption model of the traditional visual terminal The intelligent visual terminal also has data analysis capabilities, and its power energy efficiency is lower than that of edge devices, i.e. According to the power consumption of the intelligent visual terminal Run time Power consumption at full load during operation Idle power consumption Current required workload and maximum workload The load rate of the intelligent visual terminal is; The power consumption of the intelligent visual terminal is The working and idle time of the intelligent visual terminal m is divided into two sections. and The electric carbon intensity of this intelligent visual terminal With the region z∈D and the energy type g∈G of electricity generation, the carbon emission model of the intelligent visual terminal is: m∈{V M+1 ,...,V m ,...,V M+N }; The carbon emission model of the terminal equipment is: Obtain the network transmission carbon emissions and communication transmission efficiency ee, the amount of data transmitted d, the energy consumption E and the electric carbon intensity I under the end-edge-cloud collaboration in the end-edge-cloud vision collaborative Internet of Things; obtain the energy consumption υ per bit of data in the end-edge-cloud vision collaborative Internet of Things; The amount of data transmitted by the edge-cloud vision collaborative IoT is d m The resolution of the video frame m Related, that is Where σ is a fixed value; and the network transmission energy consumption is related to the energy consumption per bit of data, and the resolution of the video frame is r m The carbon emission model transmitted between node i and node j is: The carbon emissions of network transmission are related to the energy consumption per bit of data, the electric carbon intensity, the video frame resolution, the frame sampling rate, and the communication transmission efficiency. The carbon emission model of network transmission is:

4. The carbon emission assessment method for edge-cloud visual collaborative architecture according to claim 3 is characterized in that: The offload selection strategy matrix m∈V,j=[0,1,2,...,K+1], x is 0 or 1. The decision task is executed locally on the intelligent visual terminal, on the edge server or on the cloud server; m is the id of the terminal device; V is the set of all visual tasks; where j=0 means local unloading on the intelligent visual terminal, j=[1,2,...k,...,K] means unloading on the edge server, and j=K+1 means unloading on the cloud server; for edge server k, the load rate of edge computing resources is; For cloud servers, the load rate of computing resources is: Overall carbon emission model: Among them, the first part is the carbon emission model of local unloading; the second part is the carbon emission model of edge unloading; the third part is the carbon emission model of cloud unloading; and the fourth part is the carbon emission model of network transmission from the terminal to the edge and the cloud.

5. A carbon emission assessment device for edge-cloud visual collaborative architecture, characterized in that: include: The initial module obtains the edge-cloud vision collaborative IoT including traditional vision terminals, intelligent vision terminals, edge servers and cloud servers; The model building module builds the end device carbon emission model, edge device carbon emission model, and cloud device carbon emission model according to their functions based on the carbon emission parameters of traditional visual terminals, intelligent visual terminals, edge servers, and cloud servers. It also builds the network transmission carbon emission model based on the carbon emission parameters generated by network transmission when the end-edge-cloud vision collaborative Internet of Things is running. The task preparation module obtains multiple visual tasks, and constructs an offloading selection strategy matrix for the multiple visual tasks according to the deployment strategy of each visual task; constructs a carbon emission matrix corresponding to the offloading selection strategy matrix according to the network transmission carbon emission model, the end device carbon emission model, the edge device carbon emission model and the cloud device carbon emission model, and the carbon emission matrix represents the carbon emissions generated by each visual task on each server in the end-edge-cloud visual collaborative Internet of Things; constructs a computing resource load matrix according to the computing resource upper limit of each device in the end-edge-cloud visual collaborative Internet of Things; The task deployment module adopts the Hungarian algorithm to solve the unloading selection strategy matrix when the carbon emissions of the network transmission carbon emission model are minimized and the computing resources occupied by the visual task are less than or equal to the computing resource upper limit of each device, as the task deployment strategy of the multiple visual tasks, and adopts the task deployment and executes the multiple visual tasks.

6. The carbon emission assessment device for edge-cloud visual collaborative architecture according to claim 5, characterized in that: The task deployment module includes: Module 1: Set the carbon emission matrix and computing resource load matrix based on the visual task and offloading selection strategy matrix. Each row in the carbon emission matrix increases in sequence according to the end, edge, and cloud order. When deciding to output the task deployment strategy, it is necessary to ensure that only one element in each row of the task deployment strategy is 1 and meets the computing resource load matrix requirements. Module 2: For each visual task, the edge server with the smallest carbon emissions is selected, the corresponding rows and columns of the carbon emissions matrix are subtracted from the corresponding carbon emissions, and the corresponding servers of the resource load matrix are calculated by subtracting the corresponding loads; Module 3, according to the carbon emission matrix and the computing resource load matrix of the 0 value, assign the execution device of the visual task, if the assigned 0 value is equal to the total number of tasks of the multiple visual tasks, then end; otherwise, call the module 2 again.

7. The carbon emission assessment device for edge-cloud visual collaborative architecture according to claim 5 or 6, characterized in that: The model building blocks include: The cloud server includes computing equipment for training, learning and data calculation, cooling system and basic equipment for cloud operation; the energy consumption of all equipment in the entire cloud center is calculated. Divided by the energy consumption of IT equipment operation Get the energy utilization efficiency pue of the cloud server c ; The cooling factor of the cloud server is Where b, c and d are real numbers, Tem sup is the temperature set for the cooling equipment. The operating power consumption of the cloud server is p0 = (p c +p c / Cop(Tem sup )); Carbon emissions and carbon intensity of electricity consumption Where G = {1, 2, ... G} and D = {1, 2, ... D} represent the renewable energy source set and regional set respectively; gz represents the power consumption of renewable energy type g in area z∈D; c g is the carbon emission rate of renewable energy type g; the computing resource load rate of the cloud server is; in and are the current load and maximum load of the cloud server respectively; the carbon emissions generated by the operation of the cloud server are: According to the carbon emissions caused by the refrigerant leakage of the cloud server and the quality of the refrigerant used, r , global warming potential GWP, refrigerant leakage rate ρ, and the overall operating time of the equipment t cool , building carbon emissions from the use of refrigerants The carbon emission model of the cloud equipment is: The edge server includes IT equipment for edge computing and system basic operation equipment; the edge-edge-cloud visual collaborative Internet of Things includes K edge servers; the power energy utilization efficiency of each edge server is obtained Get the computing power consumption of the edge server's own computing device and running time The computing power consumption of edge servers Power consumption at full load Idle power consumption Current required workload and maximum workload Related; The load rate of the edge server is; The power consumption is: The time when the edge server is working and idle is divided into two periods and Since different sites provide different energy mixes to edge servers, the carbon intensity of electricity at each edge Different, that is, the overall calculation model of carbon emissions at edge end k is: The carbon emission model of the edge equipment is: The traditional visual terminal uses visual perception to compress the captured video and upload it for storage, and performs analysis, recognition and processing on the edge and cloud. The intelligent visual terminal has the ability to extract visual features and store the compressed video stream locally, which can be retrieved on demand based on the needs of the edge and cloud. It also supports the migration and compression of models such as video encoding and feature analysis between the edge and cloud. m Equipment category o m , if it is the traditional visual terminal then o m =0, otherwise o m =1; The traditional visual terminal is based on its power consumption and running time As well as the energy type g∈G and carbon intensity of electricity used in the region z∈D Based on the current required workload Energy consumption model of the traditional visual terminal The intelligent visual terminal also has data analysis capabilities, and its power energy efficiency is lower than that of edge devices, i.e. According to the power consumption of the intelligent visual terminal Run time Power consumption at full load during operation Idle power consumption Current required workload and maximum workload The load rate of the intelligent visual terminal is; The power consumption of the intelligent visual terminal is The working and idle time of the intelligent visual terminal m is divided into two sections. and The electric carbon intensity of this intelligent visual terminal With the region z∈D and the energy type g∈G of electricity generation, the carbon emission model of the intelligent visual terminal is: m∈{V M+1 ,...,V m ,...,V M+N }; The carbon emission model of the terminal equipment is: Obtain the network transmission carbon emissions and communication transmission efficiency ee, the amount of data transmitted d, the energy consumption E and the electric carbon intensity I under the end-edge-cloud collaboration in the end-edge-cloud vision collaborative Internet of Things; obtain the energy consumption υ per bit of data in the end-edge-cloud vision collaborative Internet of Things; The amount of data transmitted by the edge-cloud vision collaborative IoT is d m The resolution of the video frame m Related, that is Where σ is a fixed value; and the network transmission energy consumption is related to the energy consumption per bit of data, and the resolution of the video frame is r m The carbon emission model transmitted between node i and node j is: The carbon emissions of network transmission are related to the energy consumption per bit of data, the electric carbon intensity, the video frame resolution, the frame sampling rate, and the communication transmission efficiency. The carbon emission model of network transmission is: The offload selection strategy matrix m∈V,j=[0,1,2,...,K+1], x is 0 or 1. The decision task is executed locally on the intelligent vision terminal, on the edge server, or on the cloud server. m is the id of the terminal device. V is the set of all visual tasks. Among them, j=0 means local unloading on the intelligent vision terminal, j=[1,2,...k,...,K] means unloading on the edge server, and j=K+1 means unloading on the cloud server. For edge server k, the load rate of edge computing resources is; For cloud servers, the load rate of computing resources is: Overall carbon emission model: Among them, the first part is the carbon emission model of local unloading; the second part is the carbon emission model of edge unloading; the third part is the carbon emission model of cloud unloading; and the fourth part is the carbon emission model of network transmission from the terminal to the edge and the cloud.

8. An electronic device, characterized in that: It includes a carbon emission assessment device for an end-edge-cloud visual collaborative architecture as described in claims 5-7, and the electronic device is connected to an information display device, which is used to display the assessment result according to display parameters and attributes set by the user or through an artificial intelligence model.

9. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the carbon emission assessment method for an edge-cloud visual collaborative architecture as described in any one of claims 1-4.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the carbon emission assessment method for the edge-cloud visual collaborative architecture described in any one of claims 1-4 are implemented.

Images