Data computing resource allocation method considering privacy protection under edge-cloud collaborative framework

Abstract

The present invention discloses a data computing resource configuration method considering privacy protection under an edge-cloud collaborative framework, including constructing an edge-cloud collaborative computing framework, defining the data processing problems of the data processing network model of the computing framework; defining and setting various parameters of the network model; determining the computing resource configuration and privacy protection level according to the various parameters; constructing an edge-cloud data computing resource planning model considering privacy protection according to the various parameters of the data processing network model; solving the edge-cloud data computing resource planning model considering privacy protection to obtain the optimal edge-cloud computing resource configuration strategy and privacy protection level strategy. The present invention not only comprehensively considers the industrial production and operation and maintenance data processing costs under the edge-cloud collaborative framework, but also considers the privacy protection requirements when data is processed in the cloud computing layer, and achieves the optimal configuration of data computing resources by minimizing the total data processing cost.

Description

Technical Field

[0001] The present invention relates to the technical field of edge computing and cloud computing, and in particular to a data computing resource configuration method considering privacy protection under an edge-cloud collaborative framework. Background Art

[0002] The rapid adoption of Industry 4.0 technologies, such as the Internet of Things (IoT), cloud computing, edge computing, and artificial intelligence (AI), has provided companies with more opportunities to upgrade their manufacturing facilities. As digital business continues to evolve, more and more data generated by smart manufacturing companies will be created and processed outside of traditional centralized data centers or the cloud.

[0003] The exponential growth of industrial data poses enormous challenges to network, storage, and computing resources. Industrial equipment requires precise latency control, massive IoT connectivity, and the ability to handle exponentially growing data. Traditional cloud computing approaches inevitably lead to issues such as slow response times and uneven distribution of computing resources. Edge-cloud collaborative computing, which complements the strengths of cloud and edge computing, strategically leverages the computing and storage capabilities of the cloud layer and the real-time nature of the edge layer to better assist in monitoring the status of production equipment, thereby better supporting manufacturing enterprises' production operations and digital transformation.

[0004] Furthermore, cybersecurity is becoming a daily challenge for businesses worldwide. Smart manufacturing companies must consider privacy protection when processing industrial data, especially when computing and storing data in the cloud, where privacy is vulnerable. To prevent data theft, manufacturers must implement the right technologies to protect private data, at considerable cost.

[0005] Due to the respective shortcomings of edge computing and cloud computing in data computing and the urgent need to protect industrial data privacy, smart manufacturing companies have to solve the problem of how to effectively and reasonably configure the computing resources of the edge computing layer and cloud computing layer to minimize the total cost of data processing under the constraints of computing resources, storage capacity and latency, taking into account the privacy protection demands of data.

[0006] In order to understand the development status of the prior art, we searched, compared and analyzed existing patents and literature, and screened out the following technical information that is highly relevant to the present invention:

[0007] Patent number CN115242800A, "A Game Theory-Based Mobile Edge Computing Resource Optimization Method and Apparatus," discloses a game-theory-based mobile edge computing resource optimization method and apparatus. First, to address the growing computing demands in mobile edge computing, a Stackelberg competition model between user mobile devices and servers is established, taking into account constraints such as limited server resources, data transmission energy consumption, and server computing costs. Second, the computational utility maximization problem is solved to obtain the optimal user mobile device task offloading strategy and server pricing strategy. Taking into account limited server resources, data transmission energy consumption, multi-user competition for computing resources, and server computing costs, this approach simultaneously achieves efficient computation offloading for user mobile devices and maximizes server profits, thereby minimizing computational costs. This solution achieves efficient computation offloading for user mobile devices and maximizes server profits, thereby minimizing computational costs. However, it is limited to computing resource optimization at the edge layer and does not incorporate resource allocation at the cloud computing layer.

[0008] Patent number CN114285853A, "Task Offloading Method Based on End-Edge-Cloud Collaboration in Device-Intensive Industrial IoT," discloses a task offloading method based on end-edge-cloud collaboration in device-intensive industrial IoT. The specific steps are as follows: Step 1: Set the parameters of the system model; Step 2: Use the ISAC-DMDRL algorithm to make the optimal decision for each IIoT device. This algorithm first combines distributed RL with the SAC algorithm to address the overestimation or underestimation of Q values ​​in traditional SAC algorithms. Then, using the CTDE framework, the improved SAC algorithm is extended to multi-agent scenarios, addressing the non-stationarity and scalability issues inherent in multi-agent DRL. Furthermore, a value function decomposition approach is employed to address the centralized-distributed mismatch and multi-agent credit allocation issues inherent in traditional CTDE architectures. This method effectively reduces task execution latency and energy consumption across all devices in the IIoT, balances the workload of edge servers, and improves resource utilization, making it suitable for large-scale, device-intensive industrial IoT scenarios.

[0009] This solution can effectively reduce the task execution delay and energy consumption of all devices in the industrial Internet of Things, balance the workload of edge servers, and improve resource utilization. However, it cannot effectively control data processing costs and ignores the privacy protection requirements of data processing tasks.

[0010] In summary, in the scenario of edge-cloud collaborative processing of device-level industrial data, existing resource allocation models implement data computing resource allocation from the perspective of minimizing latency and energy consumption. However, data processing costs directly impact an enterprise's production and maintenance efficiency, and in practice, there are additional demands for data privacy protection. Considering both these factors can better facilitate the rational allocation of data computing resources. Therefore, while ensuring that device-level data is computed, the biggest problem with existing methods is that they cannot effectively allocate computing resources while also providing a certain degree of data privacy protection. Summary of the Invention

[0011] In order to solve the above-mentioned defects in the prior art, the purpose of the present invention is to provide a data computing resource configuration method that considers privacy protection under the edge-cloud collaborative framework. This method not only comprehensively considers the industrial production and operation and maintenance data processing costs under the edge-cloud collaborative framework, but also considers the privacy protection requirements when data is processed in the cloud computing layer, and achieves optimal configuration of data computing resources by minimizing the total data processing cost.

[0012] The present invention is achieved through the following technical solutions.

[0013] In one aspect, the present invention provides a data computing resource configuration method considering privacy protection in an edge-cloud collaborative framework, comprising:

[0014] Construct a data processing network model based on the edge-cloud collaborative computing framework and define the data processing problems of the data processing network model;

[0015] Define and set the parameters of the data processing network model, and determine the computing resource configuration and privacy protection level of the data processing network model;

[0016] Based on the parameters of the data processing network model, a privacy-preserving edge-cloud data computing resource planning model is constructed.

[0017] The edge cloud data computing resource planning model considering privacy protection is solved to obtain the optimal edge cloud computing resource configuration strategy and privacy protection level strategy.

[0018] Preferably, the data processing network model includes a device layer, an edge computing layer, and a cloud computing layer. The device layer is used to collect original data on production and device status; the edge computing layer deploys small base stations and edge servers to process simple original data from the device layer; and the cloud computing layer is used to process complex original data from the device layer.

[0019] Preferably, the data processing problems of the data processing network model are defined, including making decisions on device layer data of different data sizes, and processing them at the edge computing layer and cloud computing layer nodes with higher data privacy leakage risks under the constraints of latency, computing power, and storage space, thereby minimizing the total cost of data processing.

[0020] Preferably, the parameters of the edge-cloud collaborative computing framework include the coefficients of various costs generated by data processing, the rate and power of data upload and processing, the computing resources required for each edge computing layer node and cloud computing layer node to calculate 1 bit of data, computing power, maximum computing resources, latency limit, storage space limit and the possibility of data being attacked by network attacks.

[0021] Preferably, privacy protection methods include secure multi-party computing, federated learning, differential privacy protection, homomorphic encryption, and blockchain protection.

[0022] Preferably, the edge cloud data computing resource planning model considering privacy protection includes:

[0023] The sum of fixed costs, variable costs for processing data, energy costs for uploading and computing, storage costs, privacy protection costs, and data leakage loss costs.

[0024] Preferably, the edge-cloud data computing resource planning model considering privacy protection satisfies the following constraints:

[0025] Task allocation constraints at the device level;

[0026] The latency constraints, computing power constraints, storage space constraints, configuration of decision variable computing resources, and the range of privacy protection levels of the edge computing layer and cloud computing layer.

[0027] Preferably, the task allocation constraint is to ensure that all device layer data is processed and one piece of data can only be processed at one node.

[0028] Preferably, the edge-cloud data computing resource planning model considering privacy protection is solved by using CPLEX and the enumeration of decision variable privacy protection levels.

[0029] In another aspect, the present invention provides an electronic device, comprising a memory and a processor, wherein the memory and the processor are connected;

[0030] Memory, used to store programs;

[0031] A processor, configured to call a program stored in the memory to execute the method according to any one of claims 1 to 9.

[0032] The present invention adopts the above technical solution, which has the following beneficial effects:

[0033] 1. The present invention takes into account the data processing tasks of industrial equipment at the device layer, adopts the construction of a data processing network model based on the edge-cloud collaborative computing framework, defines the data processing problems of the data processing network model, and solves the problems of all data processing tasks at the effective computing device layer.

[0034] 2. This invention takes into account the processing cost of industrial production and operation and maintenance data and the potential leakage of privacy data. It adopts the parameters of the data processing network model to construct an edge-cloud data computing resource planning model that considers privacy protection, and solves the problem of formulating strategies that take into account both edge-cloud data computing resource configuration and privacy protection levels.

[0035] 3. This paper considers the requirements of practical industrial computing and solves a privacy-sensitive edge-cloud data computing resource planning model to obtain the optimal edge-cloud computing resource allocation strategy and privacy protection level strategy. This solves the problem of solving resource planning and configuration models in a relatively short time. This approach guides smart manufacturing enterprises in determining optimal computing resource allocation decisions and privacy protection strategies within the edge-cloud collaborative framework to control the total cost of data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings described herein are used to provide a further understanding of the present invention, constitute a part of this application, and do not constitute an improper limitation of the present invention. In the drawings:

[0037] Figure 1 A flowchart of a data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework provided by the present invention;

[0038] Figure 2 A schematic diagram of the device layer data processing network model based on the edge-cloud collaborative computing framework provided by the present invention;

[0039] Figure 3 A conceptual model diagram of the trade-off between the various data processing costs under the edge-cloud collaborative framework provided by the present invention;

[0040] Figure 4 Cost composition diagram corresponding to different data processing methods provided by the present invention;

[0041] Figure 5 This is a trend chart showing how the total cost of edge-cloud collaborative data processing varies with the privacy protection level β provided by the present invention. DETAILED DESCRIPTION

[0042] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The exemplary embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0043] refer to Figure 1, the data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework provided by an embodiment of the present invention includes the following steps:

[0044] Step S1: construct a data processing network model based on the edge-cloud collaborative computing framework and define the data processing problems of the data processing network model.

[0045] The data processing network model based on edge-cloud collaborative computing framework in intelligent manufacturing environment is as follows: Figure 2 As shown in the figure, it consists of three layers: device layer, edge computing layer, and cloud computing layer.

[0046] The equipment layer includes various terminal control devices contained in the production sites of intelligent manufacturing enterprises. Collectors and sensors can collect raw data such as production and equipment status to provide support for intelligent operation and maintenance and diagnosis.

[0047] The edge computing layer is close to the end-layer devices and is deployed with small base stations and edge servers with general computing and storage capabilities to process simple raw data from the device layer.

[0048] The cloud computing layer is far away from the end-layer devices and has powerful computing capabilities, which can process more complex data at the device layer.

[0049] To minimize the total cost of data processing, smart manufacturing companies need to decide how to process device-layer data of different sizes at the edge computing layer and cloud computing layer nodes under the constraints of latency, computing power, and storage space. In particular, when data is processed at cloud computing layer nodes with higher leakage risks, how to configure computing resources and decide the optimal level of privacy protection.

[0050] Take an automobile welding production line as an example, which includes various equipment layer control devices such as motors, cylinders, robots, etc., and the amount of data generated by each in a period of time is q i The data set N = {1, 2, ..., n}, n represents the amount of data q n These device-layer data can be collected by collectors and sensors and transmitted to edge computing layer nodes or cloud computing layer nodes for processing.

[0051] The set of computing nodes in the edge and cloud computing layers is represented as M = {1, 2…, m}, where the first m-1 nodes E = {1, 2,…, m-1} are edge computing layer nodes, and the mth node C = {m} is a cloud computing layer node. When device-layer data processing tasks are performed at the edge computing layer, the tasks are offloaded to the edge computing layer via the cellular network and then processed at the edge computing layer nodes. When data is computed at the cloud computing layer, the tasks are first offloaded to the edge computing layer via the cellular network and then transmitted via the core network to remote cloud computing layer nodes for processing.

[0052] Step S2: define and set the parameters of the data processing network model and determine the configuration of computing resources of the edge computing layer and the cloud computing layer. ij and privacy protection level β.

[0053] The parameters of the edge-cloud collaborative computing framework include the various cost coefficients generated by data processing, the rate and power of data upload and processing, the computing resources required for each edge computing layer node and cloud computing layer node to calculate 1 bit of data, computing power, maximum computing resources, latency limit, storage space limit and the possibility of data being attacked by network attacks.

[0054] The present invention takes into account the respective characteristics of the edge computing layer and the cloud computing layer. The data generated by the device layer will incur correspondingly different costs and time consumption when processed by computing nodes at different levels.

[0055] First, the processing time per unit of data volume varies among nodes at different levels. The edge computing layer needs to perform regular calculations within a certain period of time, and has stricter latency requirements; while the cloud computing layer can perform more complex data processing, and has relatively loose latency requirements. Since each node will generate latency when processing data, to simplify the problem, the latency of edge computing layer nodes processing data is unified as tl1, and the latency of cloud computing layer nodes processing data is unified as tl2. Here, tl1 <tl2。

[0056] Secondly, processing data at two different level nodes, the edge computing layer and the cloud computing layer, will generate different fixed costs and unit variable costs. Due to the powerful and rich computing power and expensive equipment costs of the cloud computing layer, its unit data processing cost is lower than that of the edge computing layer, while fixed costs such as depreciation and maintenance are higher than those of the edge computing layer.

[0057] Thirdly, the data will generate corresponding energy consumption costs during the process of uploading and computing. The unit energy consumption of the edge computing layer is less than that of the cloud computing layer. When further considering the problem of data storage after processing, storage costs will also be generated. The unit storage cost of the cloud computing layer is less than that of the edge computing layer. Based on this, it is assumed that the fixed cost and unit variable cost of data processing nodes in the edge computing layer and the cloud computing layer are f respectively. j and v h , then f j (j∈E′) <f j (j∈C′), v j (j∈E′)>v j (j∈C′). Assume that the computing power of the data processing node (edge ​​computing layer, cloud computing layer) is r j , and the computing power of the cloud computing layer is much greater than that of the edge computing layer, that is, r j (j∈E′)<<r j(j∈C′). Assume that the upload and processing power of data processing nodes (edge ​​computing layer and cloud computing layer) are and but The rate at which data is uploaded from the device layer to each node is The rate calculated by each node is The upload rate of the edge computing layer is higher than that of the cloud computing layer, and the calculation rate is lower than that of the cloud computing layer, that is,

[0058] Finally, when the data privacy level is high, the centralized distribution of nodes in the cloud computing layer creates a significant risk of privacy leakage when data is processed there. Network defense can be achieved by investing in privacy-preserving technologies such as federated learning and blockchain. However, it is important to determine the privacy protection level β (i.e., the success rate of detecting network attacks) in order to adopt appropriate privacy-preserving technologies.

[0059] This will generate a corresponding privacy protection cost f(β), which is a quadratic cost function. This means that as the marginal privacy protection cost increases, increasing the privacy protection level often means higher investment costs. In addition, the higher the level of data privacy protection, the lower the probability of data leakage ρ, which is a monotonically decreasing function of β.

[0060] To simplify the process, we can set ρ = α(1-β), where α represents the possibility of suffering a network attack.

[0061] Although investing in privacy protection technologies can further reduce the risk of data leaks, data leaks still result in losses. This is because competitors can analyze the leaked data and potentially obtain a company's core business secrets, putting the company's operations in trouble.

[0062] Table 1 shows several privacy protection methods. After the privacy protection level β is determined by the model of the present invention, an appropriate privacy protection method can be selected to protect the data according to the data protection level of the technology.

[0063] Table 1 Comparison of privacy protection related methods

[0064]

[0065] Note: *Defined as the security level of data privacy protection

[0066] Therefore, when processing data of the same size, the edge computing layer is superior to the cloud computing layer in terms of fixed costs, privacy protection costs, leakage loss costs, and energy consumption costs, but the cloud computing layer is superior in terms of variable costs and storage costs. In addition, the edge computing layer has lower latency, which means that the service quality of the computing nodes is higher, but the computing and storage resources of the cloud computing layer are more abundant. Therefore, in minimizing the total cost of data processing, due to the constraints of latency, computing power, and storage, we have to make trade-offs among various cost components, such as Figure 3 shown.

[0067] The decision variables include the configuration of computing resources x ij and privacy protection level β.

[0068] is a binary decision variable that represents the configuration of computing resources:

[0069]

[0070] When the i-th data of the device layer is processed in the data processing node j, its value is 1, otherwise it is 0. The data processing task allocation matrix is ​​X={x ij}∈{0,1} N×M β is a continuous decision variable with a value between [0,1], which represents the data privacy protection level or network defense level of the manufacturing enterprise.

[0071] Step S3: Based on the parameters of the data processing network model, a cloud-edge data computing resource planning model that takes privacy protection into consideration is constructed.

[0072] The total cost of device-layer data processing consists of the fixed costs of edge and cloud computing nodes, variable data processing costs, energy costs, storage costs, and privacy protection and data leakage costs at the cloud computing layer. Under the constraints of latency, computing resources, and storage resources, a privacy-preserving edge-cloud data computing resource planning model is constructed to address the cost optimization problem of device-layer data processing.

[0073] When device-layer data is calculated at the edge computing layer, it is uploaded to the edge computing layer for processing. If an edge node is overloaded, the data can be transferred to a nearby node for processing. The latency of uploading data to the edge node for processing consists of three components: upload time, calculation time at the node, and the time it takes to transmit the calculation results back to the device layer. Because the calculated result is much smaller than the original data and the downlink transmission rate is much higher than the uplink transmission rate, the time it takes to transmit the calculation result back to the device layer is ignored. Similarly, because the calculated result is much smaller than the original data, the storage space occupied by the result is also ignored.

[0074] When device-layer data is calculated at the cloud computing layer, it is uploaded to the cloud computing layer nodes for processing, and the results are transmitted back to the device layer. This time consumption is also divided into three parts: the time it takes to upload the device-layer data to the cloud computing layer, the time it takes to process the data at the cloud computing layer, and the time it takes to return the results. Again, the time it takes to return the data is ignored, as is the storage space occupied by the calculation results.

[0075] The various costs included in the objective function are expressed as:

[0076] 3.1) Fixed costs of nodes in the edge computing layer and cloud computing layer

[0077] The fixed cost incurred by computing node j is:

[0078]

[0079] Where n represents the number of data to be processed generated by the device layer, m represents the total number of computing nodes, of which the first m-1 nodes are edge computing layer nodes, the mth node is the cloud computing layer node, and x ij =1 means that the i-th data is processed at node j, f j is the fixed cost coefficient for owning node j, such as depreciation and maintenance fees;

[0080] 3.2) Variable data processing costs of nodes in the edge computing layer and cloud computing layer

[0081] The variable cost of node j is:

[0082]

[0083] Where: q i Indicates the amount of data generated by the i-th device in the device layer, v j The variable cost coefficient for data collection, cleaning, and preprocessing per unit data volume;

[0084] 3.3) Energy consumption cost of nodes in edge computing layer and cloud computing layer

[0085] The energy consumption cost of node j is:

[0086]

[0087] Where: ω represents the unit energy consumption cost coefficient, and are the power required for uploading data to node j and processing at node j, is the rate at which data is uploaded to node j, c j represents the CPU cycles required for node j to calculate 1 bit of data, r j represents the computing power of node j;

[0088] 3.4) Storage costs of edge computing and cloud computing nodes

[0089] The storage cost of node j is:

[0090]

[0091] Where: s j represents the unit storage cost of node j;

[0092] 3.5) Data privacy protection cost of cloud computing layer nodes

[0093]

[0094] Where: k represents the cost coefficient of enterprise data privacy protection, β represents the privacy protection level;

[0095] 3.6) Data leakage loss cost of cloud computing layer nodes

[0096] cd=α(1-β)F (6)

[0097] Where: α represents the possibility of data being attacked by a network attack, and F represents the loss cost of enterprise data leakage.

[0098] Based on the above settings and parameters, a privacy-protected edge-cloud data computing resource planning model is constructed.

[0099]

[0100] Among them, n represents the number of data to be processed generated by the device layer, q i represents the amount of data generated by the i-th device in the device layer, m represents the total number of computing nodes in the edge computing layer and the cloud computing layer, and f j is the fixed cost coefficient of depreciation and maintenance fees generated by owning data computing node j, v j The variable cost coefficient generated by data collection, cleaning, and preprocessing per unit data volume, is the rate at which device layer data is uploaded to node j, c j represents the CPU cycles required for node j to calculate 1 bit of data, q i c j Indicates that node j calculates q i The required computing resources, j j represents the computing power of node j, and They represent the power required for uploading device layer data to node j and calculating at node j, s jrepresents the cost required for node j to store a unit amount of data per unit time, ω represents the unit energy consumption cost coefficient, k represents the cost coefficient of enterprise data privacy protection, α represents the possibility of data being attacked by the network, F represents the loss cost of enterprise data leakage, and x ij represents the configuration of computing resources, and β represents the privacy protection level.

[0101] This is a mixed-integer nonlinear programming (MINLP) model. The first term is the fixed cost, the second is the variable cost of data processing, the third is the energy cost of uploading and computing, the fourth is the storage cost, and the last two terms are the privacy protection cost and the data leakage loss cost. Since the cloud computing layer is more open and has a greater risk of privacy leakage, while the edge computing layer is isolated by many network protection layers and is disconnected from the office network, it has higher security. Therefore, the main consideration here is the possibility of data leakage in the cloud computing layer.

[0102] The edge-cloud data computing resource planning model considering privacy protection must meet the following constraints:

[0103] 1) The task allocation constraints at the device layer are expressed as:

[0104]

[0105] Where: Indicates the number of computing nodes corresponding to the data of the i-th device in the device layer;

[0106] 2) The delay constraints of nodes in the edge computing layer and cloud computing layer are expressed as:

[0107]

[0108]

[0109] Where: Indicates the longest time required for edge computing layer nodes to process data. represents the time required for cloud computing layer nodes to process data, tl1 represents the delay constraint for edge computing layer nodes to process data, and tl2 represents the delay constraint for cloud computing layer nodes to process data;

[0110] 3) The computing power constraints of nodes in the edge computing layer and cloud computing layer are expressed as:

[0111]

[0112] Where: represents the CPU cycles required for all data processed at node j, cl h represents the maximum computing resources of node j;

[0113] 4) The storage space constraints of the nodes in the edge computing layer and the cloud computing layer are expressed as:

[0114]

[0115] Where: Indicates the total amount of data processed by node j, sl j represents the storage space constraint of node j.

[0116] 5) The configuration of decision variable computing resources and the value range constraints of the privacy protection level are expressed as:

[0117]

[0118] β∈[0,1] (14)

[0119] In step S4, the edge cloud data computing resource planning model considering privacy protection is solved to obtain the optimal edge cloud computing resource configuration strategy and privacy protection level strategy.

[0120] The edge-cloud data computing resource planning model considering privacy protection follows mixed integer nonlinear programming. The present invention adopts CPLEX and decision variable β enumeration to obtain the optimal solution in a relatively short time.

[0121] The advantages of the present invention can be further illustrated by the following simulation experiments:

[0122] 1. Build a data processing network model based on the edge-cloud collaborative computing framework and define the data processing problems of the data processing network model

[0123] In this example, data from a Chinese automobile body-in-white welding production line was selected. Automobile manufacturing is a typical discrete manufacturing industry. Non-standard equipment on the production line suffers from poor stability, impacting stable operation and resulting in wasted time. Pain points such as high production line quality control costs and low efficiency are becoming increasingly prominent. Intelligent operation and maintenance of welding production lines is a key development direction, with edge-cloud collaboration playing a key role. By deploying an edge computing layer close to the equipment layer, some data with high real-time processing requirements can be met, while large-scale data with lower real-time processing requirements can be uploaded to the cloud computing layer for processing.

[0124] The automobile body-in-white welding production line includes numerous industrial equipment. The production and operation maintenance data generated by the equipment layer during a certain production cycle needs to be processed promptly and effectively to judge the operation status of the production line site, such as beat balance, production efficiency, and equipment stability.

[0125] 2. Define and set the parameters of the data processing network model, and determine the computing resource configuration and privacy protection level of the data processing network model

[0126] The equipment layer of the automobile body-in-white welding production line contains 15 devices. The scale of data generated within a certain production cycle is shown in Table 2. It can be transmitted to the four computing nodes of the edge computing layer or the one computing node of the cloud computing layer for processing.

[0127] Table 2 Data size (MB) generated by each industrial device at the equipment layer

[0128]

[0129] Other corresponding parameter settings are shown in Table 3:

[0130] Table 3 Input parameter values ​​of the data processing network model

[0131]

[0132] 3. Based on the parameters of the data processing network model, build an edge cloud data computing resource planning model that considers privacy protection

[0133] Under the aforementioned parameter settings, the obtained edge-cloud data computing resource optimization model belongs to a mixed integer nonlinear programming problem (MINLP). The CPLEX solver can solve the model in a relatively short time, which meets the requirements of actual industrial computing.

[0134] 4. Solve the edge cloud data computing resource planning model considering privacy protection to obtain the optimal edge cloud computing resource allocation strategy and privacy protection level strategy

[0135] The model's decision results are shown in Table 4. Data generated by seven devices in the device layer are processed by compute nodes in the cloud computing layer, while data generated by eight devices are processed by compute nodes in the edge computing layer. This corresponds to an optimal privacy protection level β of 0.44. The total data processing cost for edge-cloud collaborative computing is 11,937.07 yuan.

[0136] Table 4 Computing resource configuration of each edge cloud node (MB)

[0137]

[0138] Note: * is defined as the amount of data generated by the i-th device in the device layer

[0139] Due to the limited computing power and storage space of the edge computing layer, it is impossible to perform all data processing tasks on the edge computing layer. However, the cloud computing layer has powerful computing and storage resources, which can accommodate all data processing in the cloud computing layer. Figure 4It can be seen that compared with all data calculations at the edge computing layer and all data calculations at the cloud computing layer, edge-cloud collaborative computing can not only ensure that all data processing tasks are calculated, but also reduce the total data processing cost by effectively weighing the costs of each component, reflecting the advantages of edge-cloud collaborative data processing.

[0140] Furthermore, when private data is leaked, it often results in losses for manufacturing companies. If competitors gain access to private data, they are likely to develop targeted business strategies to gain a competitive advantage. Furthermore, the leakage of commercial data can also cause losses to data providers, such as supply chain partners. Both parties often sign confidentiality agreements, requiring manufacturers to compensate their partners if the data is leaked. Under these circumstances, manufacturers are forced to adopt risk-averse behavior and invest in methods at the cloud computing layer to enhance data security and prevent private data leaks. Faced with varying degrees of privacy protection, manufacturers need to make independent decisions and select the appropriate level of privacy protection, or in other words, the appropriate privacy method. Since the optimal level of privacy protection is moderate, manufacturers should choose methods with moderate data protection levels, such as federated learning, differential privacy, and blockchain, to protect sensitive data computed at the cloud computing layer.

[0141] Table 5. Cost units (yuan) corresponding to different privacy protection levels in the computing resource planning model

[0142]

[0143] As can be seen from Table 5, the optimal privacy protection level is not the higher the better. If it is too high, the privacy protection cost will be too high, thereby increasing the total cost expenditure. If it is too low, the loss of data leakage will become a very high expenditure. Therefore, it is necessary to make an effective trade-off between the two. Figure 5 This further illustrates the importance of determining the appropriate level of privacy protection or choosing the right privacy protection method for cost control. Specifically, when faced with the risk of data privacy leakage, compared to completely ignoring privacy protection (β = 0), the total data processing cost is reduced by 7.84%, and the allocation decision of data computing resources is changed. Compared with fully protecting private data (β = 1), the allocation decision of data computing resources is also changed, and the total data processing cost is reduced by 10.33%.

[0144] 5. Analysis results

[0145] From the above experiments, it can be seen that the application of the present invention can guide smart manufacturing companies to better configure the computing resources of the edge computing layer and the cloud computing layer under the condition of privacy protection to minimize the data processing cost, and can also effectively determine a reasonable level of privacy protection, so as to adopt an appropriate privacy protection method. The optimal privacy protection level is not the higher the better, it needs to be balanced between the cost of data leakage loss and the cost of privacy protection. In general, excessive privacy protection does not necessarily make the total cost of data processing lower. Taking privacy protection into consideration, the configuration of edge cloud data computing resources is likely to change, and smart manufacturing companies have to examine their own data security environment in actual industrial production and operation.

[0146] An embodiment of the present invention further provides an electronic device comprising a memory and a processor, the memory and processor being connected; the memory being used to store computer programs, such as software function modules in a data processing network model, and raw production and device status data collected by the storage device layer; and the processor being used to call programs in the memory, such as small base stations and edge servers deployed in the edge computing layer, to process simple raw data from the device layer, and the cloud computing layer to process complex raw data from the device layer. The electronic device described above can be a computer or a server.

[0147] The present invention is not limited to the above-mentioned embodiments. On the basis of the technical solutions disclosed in the present invention, those skilled in the art can make some substitutions and modifications to some of the technical features therein according to the disclosed technical content without creative labor, and these substitutions and modifications are all within the protection scope of the present invention.

Claims

1. A data computing resource configuration method considering privacy protection in an edge-cloud collaborative framework, characterized by: include: Construct a data processing network model based on the edge-cloud collaborative computing framework and define the data processing problems of the data processing network model; Define and set the parameters of the data processing network model, and determine the computing resource configuration and privacy protection level of the data processing network model; Based on the parameters of the data processing network model, a privacy-preserving edge-cloud data computing resource planning model is constructed. Solve the edge cloud data computing resource planning model considering privacy protection and obtain the optimal edge cloud computing resource allocation strategy and privacy protection level strategy; The edge cloud data computing resource planning model considering privacy protection is: Among them, n represents the number of data to be processed generated by the device layer, q i represents the amount of data generated by the i-th device in the device layer, m represents the total number of computing nodes in the edge computing layer and the cloud computing layer, and f j is the fixed cost coefficient of depreciation and maintenance fees generated by owning data computing node j, v j The variable cost coefficient generated by data collection, cleaning, and preprocessing per unit data volume, is the rate at which device layer data is uploaded to node j, c j represents the CPU cycles required for node j to calculate 1 bit of data, q i c j Indicates that node j calculates q i The required computing resources, r j represents the computing power of node j, and They represent the power required for uploading device layer data to node j and calculating at node j, s j represents the cost required for node j to store a unit amount of data per unit time, ω represents the unit energy consumption cost coefficient, k represents the cost coefficient of enterprise data privacy protection, α represents the possibility of data being attacked by the network, F represents the loss cost of enterprise data leakage, and x ij represents the configuration of computing resources, and β represents the privacy protection level.

2. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 1 is characterized in that: The data processing network model includes: Equipment layer, used to collect raw data on production and equipment status; The edge computing layer deploys small base stations and edge servers to process simple raw data from the device layer; The cloud computing layer is used to process the complex raw data of the device layer.

3. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 2 is characterized in that: The data processing problems of the data processing network model are defined, including decision-making on device-layer data of different data sizes, and processing them at the edge computing layer and cloud computing layer nodes with higher data privacy leakage risks under the constraints of latency, computing power, and storage space, thereby minimizing the total cost of data processing.

4. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 1 is characterized in that: The parameters of the edge-cloud collaborative computing framework include the coefficients of various costs generated by data processing, the rate and power of data upload and processing, the computing resources required for each edge computing layer node and cloud computing layer node to calculate 1 bit of data, computing power, maximum computing resources, latency limit, storage space limit and the possibility of data being attacked by network attacks.

5. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 1 is characterized in that: Privacy protection methods include secure multi-party computing, federated learning, differential privacy protection, homomorphic encryption, and blockchain protection.

6. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 1 is characterized in that: The edge-cloud data computing resource planning model considering privacy protection satisfies the following constraints: Task allocation constraints at the device level; The latency constraints, computing power constraints, storage space constraints, configuration of decision variable computing resources, and the range of privacy protection levels of the edge computing layer and cloud computing layer.

7. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 6 is characterized in that: The task allocation constraint of the device layer is to ensure that all device layer data is processed and one piece of data can only be processed on one node.

8. The data computing resource configuration method considering privacy protection under the edge-cloud collaborative framework according to claim 1 is characterized in that: CPLEX and decision variable privacy protection level enumeration are used to solve the edge-cloud data computing resource planning model considering privacy protection.

9. An electronic device, characterized in that: comprising a memory and a processor, wherein the memory and the processor are connected; The memory is used to store programs; The processor is configured to call a program stored in the memory to execute the method according to any one of claims 1 to 8.

Images