A cloud-edge-end collaborative scheduling optimization method and system based on dynamic resource profiling

Abstract

This invention discloses a cloud-edge-device collaborative scheduling optimization method and system based on dynamic resource profiling. The method includes: constructing a dynamic resource profile of the terminal device and performing spatiotemporal sparse field modeling to generate sparse data; constructing a hierarchical differential privacy protection based on the dynamic resource profile and sparse data, and performing zero-knowledge scheduling verification on the differentially privacy-protected data; for tasks that pass the zero-knowledge scheduling verification, executing cloud-edge-device collaborative scheduling operations to optimize the hierarchical differential privacy protection and achieve efficient resource allocation. This invention reduces communication latency and improves task response efficiency by discretizing continuous spatiotemporal data streams into a three-dimensional grid. Furthermore, the hierarchical differential privacy protection mechanism balances the inefficiency of privacy protection, and the introduction of a zero-knowledge scheduling verification protocol reduces the risk of privacy leakage, making it suitable for various scheduling application scenarios.

Description

Technical Field

[0001] This invention relates to the field of cloud-based collaborative resource scheduling technology, and in particular to a cloud-edge-device collaborative scheduling optimization method and system based on dynamic resource profiling. Background Technology

[0002] In recent years, with the rapid development of the Internet of Things (IoT), edge computing, and cloud computing, cloud-edge-device collaborative scheduling technology has received widespread attention in fields such as intelligent transportation, industrial IoT, and smart cities. This technology aims to optimize resource allocation and task scheduling through the collaborative work of terminal devices, edge nodes, and cloud servers to address the demands of large-scale, dynamic, and heterogeneous distributed systems. In the field of intelligent transportation systems, connected vehicles collect location, speed, and task request data in real time through onboard units. Edge nodes (such as roadside units) perform regional-level data processing and task allocation, while cloud servers are responsible for global optimization and strategy formulation. Various scheduling methods have been developed, such as cloud-based scheduling based on centralized optimization, distributed scheduling based on edge computing, and hybrid scheduling methods combining machine learning and heuristic algorithms. Furthermore, privacy protection technologies and zero-knowledge proof technologies have been introduced to protect sensitive data and meet data security and compliance requirements. These technological advancements have significantly improved system performance, driven improvements in real-time task processing and resource utilization efficiency, and demonstrated strong adaptability, especially in highly dynamic scenarios.

[0003] However, existing technologies still have significant shortcomings in dynamic resource management and privacy protection. First, traditional scheduling methods rely heavily on static resource models, making it difficult to effectively capture real-time status changes of terminal devices (such as connected vehicles), such as dynamic changes in battery power, computing power, and movement trajectory. This results in scheduling decisions being unable to adapt to rapid fluctuations in traffic flow or task demands. Second, existing privacy protection mechanisms typically employ uniform differential privacy strategies, ignoring the spatiotemporal heterogeneity of data distribution. This makes it difficult to retain sufficient data utility in high-density areas (such as city centers) while providing strong privacy protection in low-density areas (such as suburbs), leading to an inefficient privacy-utility balance. Furthermore, traditional scheduling verification methods often require exposing device resource details, increasing the risk of privacy leaks, especially when dealing with sensitive tasks (such as emergency rescue), where an efficient privacy protection verification mechanism is lacking. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the aforementioned existing problems, this invention is proposed. Therefore, this invention provides a cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling to solve the problems mentioned in the background section.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling, comprising:

[0008] Construct dynamic resource profiles of terminal devices and perform spatiotemporal sparse field modeling to generate sparse data;

[0009] Based on the dynamic resource profile and sparse data, a hierarchical differential privacy protection is constructed, and the data after differential privacy protection is verified by zero-knowledge scheduling.

[0010] For tasks that pass the zero-knowledge scheduling verification, cloud-edge-device collaborative scheduling operations are performed to optimize the hierarchical differential privacy protection in order to achieve efficient resource allocation.

[0011] As a preferred embodiment of the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling described in this invention, the step of constructing a dynamic resource profile of the terminal device and performing spatiotemporal sparse field modeling to generate sparse data includes:

[0012] The system collects spatiotemporal data streams containing timestamps, geographic coordinates, and task attributes through terminal devices, and discretizes the spatiotemporal data streams into a three-dimensional spatiotemporal grid.

[0013] The privacy strength index of each three-dimensional spatiotemporal grid is calculated based on the amount of data, grid volume, information entropy of data distribution, and total number of terminal devices within the three-dimensional spatiotemporal grid.

[0014] As a preferred embodiment of the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling described in this invention, it further includes:

[0015] Preprocess the collected spatiotemporal data stream;

[0016] The size of the geographic grid in the 3D spatiotemporal grid is dynamically adjusted based on the data density of the spatiotemporal data stream.

[0017] The spatiotemporal grid is subjected to adaptive wavelet transform to generate sparse data.

[0018] As a preferred embodiment of the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling described in this invention, wherein: based on the dynamic resource profiling and sparse data, a hierarchical differential privacy protection is constructed, including:

[0019] At the terminal device layer, the original spatiotemporal data stream is protected by cropping and adding noise.

[0020] At the edge node layer, Gaussian noise is added to the aggregated region features;

[0021] At the cloud layer, the cluster resource scheduling system is disturbed through a probabilistic selection mechanism.

[0022] As a preferred embodiment of the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling described in this invention, it further includes:

[0023] The original spatiotemporal data stream is restricted to a preset range by pruning;

[0024] The probabilistic selection mechanism determines the perturbation intensity based on the reconstruction error and execution efficiency of the original spatiotemporal data stream.

[0025] As a preferred embodiment of the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling described in this invention, the zero-knowledge scheduling verification includes:

[0026] The scheduling constraints are transformed into arithmetic circuits, and the constraints include spatiotemporal constraints, resource constraints, and task priority constraints.

[0027] Zero-knowledge proofs are generated at edge nodes, and these proofs are generated based on private inputs of arithmetic circuits, hash values ​​of dynamic resource profiles, and three-dimensional spatiotemporal grid coordinates.

[0028] The validity of the zero-knowledge proof is verified via a smart contract in the cloud, and the verified task is added to the scheduling queue.

[0029] As a preferred embodiment of the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling described in this invention, the cloud-edge-device collaborative scheduling operation includes:

[0030] At the terminal device layer, task feasibility is assessed through local pre-screening, and the assessment is based on the task execution time.

[0031] At the edge node layer, task allocation within the region is optimized using a sliding window to minimize the weighted sum of load balancing metrics and resource allocation changes;

[0032] At the cloud layer, task allocation is optimized using genetic algorithms.

[0033] Secondly, this invention provides a cloud-edge-device collaborative scheduling and optimization system based on dynamic resource profiling, which includes:

[0034] The dynamic resource modeling and sparsification module is configured to construct a dynamic resource profile of the terminal device and perform spatiotemporal sparse field modeling to generate sparse data.

[0035] The layered privacy protection and zero-knowledge verification module is configured to construct layered differential privacy protection based on the dynamic resource profile and sparse data, and to perform zero-knowledge scheduling verification on the differentially privacy protected data;

[0036] The cloud-edge-device collaborative scheduling and optimization module is configured to perform cloud-edge-device collaborative scheduling operations on tasks that have passed the zero-knowledge scheduling verification, and optimize the hierarchical differential privacy protection to achieve efficient resource allocation.

[0037] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any step of the above-described method.

[0038] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any step of the above-described method.

[0039] Compared with existing technologies, the beneficial effects of the invention are as follows:

[0040] 1. By constructing a dynamic resource profile of terminal devices, this invention can capture the dynamic changes in device status in real time, and combined with spatiotemporal sparse field modeling, discretize the continuous spatiotemporal data stream into a three-dimensional grid, thereby reducing communication latency and improving task response efficiency.

[0041] 2. This invention uses a layered differential privacy protection mechanism to add strategies at the terminal, edge, and cloud respectively. It preserves data details in high-density areas to support high-precision tasks and enhances privacy protection in low-density areas, thus balancing the inefficiency of privacy protection.

[0042] 3. By introducing a zero-knowledge scheduling verification protocol, scheduling constraints are transformed into arithmetic circuits, generating zero-knowledge proofs at edge nodes and verifying them through smart contracts in the cloud. This ensures that task allocation meets spatiotemporal and resource constraints without exposing sensitive details of dynamic resource profiles, thus reducing the risk of privacy leaks. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0044] Figure 1 This is a flowchart illustrating the overall process of a cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling, as described in one embodiment of the present invention. Detailed Implementation

[0045] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0046] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0048] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0049] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0050] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0051] Example 1

[0052] Reference Figure 1 This is the first embodiment of the present invention, which provides a cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling, including:

[0053] S1. Construct a dynamic resource profile of the terminal device and perform spatiotemporal sparse field modeling to generate sparse data;

[0054] It should be explained that dynamic resource profiling involves extracting multi-dimensional features from the real-time status of terminal devices, edge nodes, and cloud resources to form a structured feature vector, which is used to support scheduling decisions.

[0055] Specifically, the resource profile includes dimensions such as computing power (CPU / GPU performance), storage capacity, communication bandwidth, energy consumption status, and spatiotemporal location;

[0056] Furthermore, to model the spatiotemporal distribution of resource profiles, this invention employs a spatiotemporal sparse field method, discretizing the continuous spatiotemporal data stream into a three-dimensional grid (time, longitude, latitude). Let the spatiotemporal data stream collected by the terminal device be... Where t is the timestamp, (x, y) are the geographic coordinates, i represents the number of terminal devices, n represents the total number of terminal devices, and s i For task attributes;

[0057] Specifically, the collected spatiotemporal data stream is preprocessed using Min-Max normalization to unify the dimensions of the spatiotemporal data stream and constrain its range to [0, 1]. Since this preprocessing operation is a conventional data processing technique, it will not be elaborated here.

[0058] Furthermore, to make the present invention clear and understandable, this embodiment uses an intelligent transportation system as an example, but is not limited to this intelligent transportation system. Assuming a city has thousands of connected vehicles, hundreds of roadside units and a central cloud server, and vehicle data is collected through the vehicle-mounted units, the process of constructing the dynamic resource profile is as follows: the vehicles upload data to the roadside units every second. The data includes time, location and task type (such as emergency rescue or regular navigation). The urban area is divided into spatial grids and time slices. Each spatial grid is a three-dimensional grid. Then the resource profile of each vehicle includes static features (such as task type) and dynamic features (such as time and location).

[0059] Specifically, the formula for discretizing a continuous spatiotemporal data stream into a three-dimensional mesh is as follows:

[0060]

[0061] Among them, G k,l,mRepresented as a three-dimensional grid, it is used to organize and store the set of data points generated by terminal devices in time and geographic regions. Each grid is a data container that contains multiple spatiotemporal data streams. (also called data points within a grid), for example, in intelligent transportation systems, G k,l,m This represents all vehicle data (such as location, speed, task requests, etc.) within a 5-minute time period (e.g., 18:00:00 to 18:05:00) and a 500m × 500m geographical area (e.g., a city center intersection), while s i (t,x,y) is then specified as vehicle data; k represents the discrete index of the time dimension, used to define the time interval [kΔt, (k+1)Δt). Data points within the same time period can be aggregated through time intervals to form the time dimension of the spatiotemporal grid. Δt is the preset time slice length (e.g., 5 minutes). For example, in an intelligent transportation system, if Δt = 300s (5 minutes), then k = 1 represents the first time slice (0 to 300 seconds), and k = 2 represents the second time slice (300 to 600 seconds); (x,y)∈C l,m This indicates that the geographic coordinates of the data points belong to geographic grid C. l,m The defined area, the geographic grid, is formed by dividing a geographic area into grids of fixed size, such as the 500m × 500m geographic area mentioned above. The geographic grid defines the spatial range of data points, facilitating the analysis of vehicle distribution and task requirements within the geographic area; U represents the union operator, which combines the elements of multiple sets into one set, containing all unique elements. In the formula, firstly... Geographic grid C l,m All data points within the set are merged into one set, and then, ∪ t∈[kΔt,(k+1)Δt) The set of all data points within the time interval [kΔt, (k+1)Δt) is further merged, ultimately forming a data point G that includes this time slice and all terminal devices within the geographic grid. k,l,m ;

[0062] It should be noted that discretizing continuous time through the time slice index k makes it easier to organize data by time period to analyze the dynamic changes of the data stream;

[0063] Furthermore, based on the amount of data, grid volume, information entropy of data distribution, and total number of terminal devices within the three-dimensional spatiotemporal grid, the privacy strength index Φ(G) of each three-dimensional spatiotemporal grid is calculated:

[0064]

[0065] Where ||G||1 is the amount of data within the three-dimensional spatiotemporal grid, V is the grid volume, and H(G) is the information entropy of the data distribution;

[0066] It needs to be explained that, This reflects the concentration of data points within the grid. High-density areas (such as city centers) typically have higher privacy sensitivity because with more data points, individual information is more easily inferred; while This reflects the diversity of data attributes measured by information entropy. The higher the entropy, the more uniform the data distribution, the greater the difficulty in inferring individual information, and the higher the privacy strength. Therefore, the two play a role in balancing the privacy strength.

[0067] Furthermore, the size of the geographic grid in the three-dimensional spatiotemporal grid is dynamically adjusted based on the data density of the spatiotemporal data stream;

[0068] Specifically, data density is obtained by the ratio of the total number of data points in a geographic grid to the area of ​​that geographic grid within a specific time slice. This data density changes dynamically with the time slice and is recalculated every minute to reflect real-time changes in traffic flow.

[0069] It should be noted that this time slice is not usually used directly for data density calculation, but the time slice length can be considered in the calculation of privacy strength index to obtain the data privacy strength at a specific time.

[0070] Specifically, the size of the geographic grid is dynamically adjusted using a quadtree algorithm based on the dynamic changes in data density. In high-density areas (such as city centers), the current geographic grid size is refined (reduced) to capture more detailed vehicle distribution. In low-density areas (such as suburbs), the current geographic grid size is relaxed (expanded) to reduce the computational overhead of modeling resource profiles.

[0071] Furthermore, the spatiotemporal grid undergoes adaptive wavelet transform processing to generate sparse data:

[0072]

[0073] in, This represents the sparsed data after adaptive wavelet transform processing, which is based on G. k,l,m The compressed representation, where W(·) denotes the wavelet transform operation, transforms G... k,l,m The process is transformed from the time domain to the wavelet domain, decomposed into low-frequency components (such as the main trend of vehicle distribution) and high-frequency components (such as small fluctuations in vehicle trajectories), W -1 (·) represents the inverse wavelet transform operation, which transforms the thresholded coefficients back from the wavelet domain to the original domain, generating sparse data; Thresh(·,λ·Φ(G)) represents the thresholding function, which filters the wavelet-transformed coefficients, retaining coefficients with absolute values ​​greater than the threshold λ·Φ(G) and discarding coefficients with values ​​less than the threshold to achieve data compression. λ represents the threshold adjustment factor, which is inversely proportional to the privacy strength index Φ(G);

[0074] It should be noted that by using adaptive wavelet transform to compress the data volume and generate sparse data, the problems of edge nodes (such as roadside units) and communication bandwidth consumption can be reduced.

[0075] S2. Based on dynamic resource profiling and sparse data, a hierarchical differential privacy protection is constructed, and the data after differential privacy protection is verified by zero-knowledge scheduling.

[0076] It should be explained that, since the above only mentioned the balance between the strength of privacy, but privacy leaks may still occur, a privacy protection mechanism needs to be established on this basis.

[0077] Furthermore, a three-layer differential privacy protection mechanism is constructed. At the terminal device layer, the original spatiotemporal data stream is protected by pruning and adding noise. At the edge node layer, Gaussian noise is added to the aggregated regional features. At the cloud layer, the cluster resource scheduling system is disturbed through a probabilistic selection mechanism.

[0078] Specifically, taking vehicle location and task requests (such as emergency rescue) as an example, Laplace noise is added to the vehicle location and task type data. For example, higher noise is assigned to regular navigation tasks, and lower noise is assigned to emergency rescue tasks. In addition, considering that the vehicle location data represents the maximum possible movement distance and the task type data represents the maximum request variation, the data pruning range is set to ensure the rationality of the data. For example, vehicle locations exceeding the limit are pruned to the city boundary. Then, the vehicle location and task type data with added noise are uploaded to the roadside unit through an encrypted channel (such as TLS). The roadside unit aggregates vehicle data and generates regional features (such as average speed and traffic density). Gaussian noise is added to the features, and the noise level is adjusted according to Φ(G). For example, low noise is assigned to the city center area to preserve analysis accuracy, and high noise is assigned to the suburbs. The cloud assigns the optimal path to vehicles in multiple areas through a probabilistic selection mechanism. This probabilistic selection mechanism adopts an exponential mechanism, assigning probabilities based on the utility of the selected parameters. That is, high-utility parameters (such as the weights of optimized path planning) have a high probability of being selected and have small disturbances; low-utility parameters have a low probability and large disturbances. The utility is calculated through reconstruction error (i.e., comparing the disturbances). With G k,l,m The difference is determined by calculating the mean square error;

[0079] It should be noted that the perturbation operation is mainly carried out by randomly selecting parameters to introduce uncertainty, thereby preventing attackers from deducing the original data through the scheduling results of the cluster resource scheduling system.

[0080] It should be explained that the cluster resource scheduling system, as a distributed computing framework on the cloud layer, is mainly responsible for managing and optimizing global resource allocation and task scheduling. It coordinates multiple computing nodes through distributed clusters (such as Kubernetes and Apache Mesos) to handle cross-regional task allocation and resource management. In intelligent transportation systems, it mainly optimizes cross-regional tasks (such as route planning and emergency rescue dispatch) based on sparse data uploaded by terminal devices (such as connected vehicles) and edge nodes (such as roadside units).

[0081] Furthermore, the scheduling tasks in the differential privacy protection mechanism are verified using zero-knowledge scheduling.

[0082] Furthermore, the scheduling constraints are transformed into arithmetic circuits, including spatiotemporal constraints, resource constraints, and task priority constraints.

[0083] Specifically, scheduling constraints refer to the conditions that task allocation must meet, including: spatiotemporal constraints: time and space limitations for task execution, such as the vehicle must reach the target location within 5 minutes and the distance must not exceed 5 kilometers; resource constraints: equipment resources (such as computing power and power) must meet the task requirements, such as the path planning task requiring 1 TFLOPs of computing power; and task priority constraints: high-priority tasks (such as emergency rescue) take precedence over low-priority tasks (such as regular navigation).

[0084] It should be explained that an arithmetic circuit is a mathematical representation that transforms the above constraints into a combination of polynomial operations, consisting of addition and multiplication gates, which is suitable for zero-knowledge proof protocols (such as zk-SNARK).

[0085] Furthermore, zero-knowledge proofs are generated at edge nodes, proving the generation of private inputs based on the hash values ​​of arithmetic circuits and dynamic resource profiles, as well as three-dimensional spatiotemporal grid coordinates.

[0086] Specifically, the hash value of the dynamic resource profile is calculated using cryptographic hash functions such as SHA-256, and this hash value is used as public input to ensure privacy (the original data is not exposed) and verifiability (the hash value is unique).

[0087] Specifically, the coordinates (t, x, y) of the data points are treated as private information to represent the vehicle's location within the time slice and geographic grid.

[0088] It should be noted that private inputs can be used to avoid directly exposing data information and can be used only for generating zero-knowledge proofs;

[0089] Specifically, zero-knowledge proof generation uses the zk-SNARK protocol (such as libsnark), based on arithmetic circuits, to generate public parameters (i.e., proof keys and verification keys) at the cloud layer. The edge nodes input various constraint parameters in the scheduling constraints and their corresponding hash values ​​as public inputs, while the aforementioned private information is used as private inputs. Based on these public and private inputs, the zero-knowledge proof is generated in approximately 100–500 milliseconds (the number of milliseconds matches the processor of the edge node).

[0090] Furthermore, the validity of zero-knowledge proofs is verified in the cloud via smart contracts, and the verified tasks are added to the scheduling queue.

[0091] Specifically, the smart contract loads the verification key (obtained from the public parameters generated in the cloud layer in the aforementioned zk-SNARK protocol), inputs the zero-knowledge proof and the parameters in the public input, verifies whether the zero-knowledge proof satisfies the arithmetic circuit (this process does not require access to the private input), assigns a unique ID to the task that passes the proof verification, adds it to the scheduling queue, allocates cloud resources (such as GPU computing paths), and distributes it to edge nodes and actual vehicle data;

[0092] S3. For tasks that pass the zero-knowledge scheduling verification, perform cloud-edge-device collaborative scheduling operations and optimize hierarchical differential privacy protection to achieve efficient resource allocation.

[0093] Furthermore, at the terminal device layer, task feasibility is assessed through local pre-screening, and the assessment is based on the task execution time.

[0094] Specifically, terminal devices (such as connected vehicles) perform preliminary assessments locally to determine whether tasks (such as route planning) are feasible, reducing the upload of invalid tasks. The standard for preliminary assessment is determined by the task execution time (the time required to complete the task does not exceed the maximum allowed time).

[0095] It should be noted that the purpose of pre-screening is to ensure that tasks that meet the time requirements are uploaded to the edge nodes, thereby reducing communication overhead;

[0096] Furthermore, at the edge node layer, task allocation within the region is optimized using a sliding window.

[0097] Specifically, the length of the sliding window and the sliding interval in seconds are set. The task set and resource set within the region are used as the data representation within the sliding window. The uniformity of task allocation is measured by variance to determine the load balancing index. The L2 norm is used to measure the difference between the task allocation of the current window and the task allocation of the previous window to determine the changes in resource allocation.

[0098] Furthermore, the load balancing metric (uniformity of task allocation) and resource allocation change (stability of scheduling adjustment) are weighted and summed. The goal is to minimize the load balancing metric. A greedy algorithm or linear programming is used to solve the problem within a window. The solution is then distributed to vehicles to ensure that each vehicle has a balanced workload and avoids frequent adjustments.

[0099] Furthermore, at the cloud layer, task allocation is optimized through genetic algorithms;

[0100] Specifically, genetic algorithms are used to establish objectives and coordinate cross-regional task allocation to maximize global resource utilization and achieve the optimal allocation and scheduling scheme within the city.

[0101] Specifically, the genetic algorithm generates an initial population (allocation scheme), each allocation scheme being a mapping from task to resource. Then, through its own selection, crossover, and mutation operations, it generates segments (i.e. individuals) in the allocation scheme. When the maximum number of iterations of the genetic algorithm is reached, the optimal allocation and scheduling scheme is output.

[0102] Furthermore, this embodiment also provides a cloud-edge-device collaborative scheduling optimization system based on dynamic resource profiling, including:

[0103] The dynamic resource modeling and sparsification module is configured to construct a dynamic resource profile of the terminal device and perform spatiotemporal sparse field modeling to generate sparse data.

[0104] The layered privacy protection and zero-knowledge verification module is configured to construct layered differential privacy protection based on the dynamic resource profile and sparse data, and to perform zero-knowledge scheduling verification on the differentially privacy protected data;

[0105] The cloud-edge-device collaborative scheduling and optimization module is configured to perform cloud-edge-device collaborative scheduling operations on tasks that have passed the zero-knowledge scheduling verification, and optimize the hierarchical differential privacy protection to achieve efficient resource allocation.

[0106] This embodiment also provides a computer device applicable to the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling, including:

[0107] The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling proposed in the above embodiments.

[0108] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0109] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling proposed in the above embodiments.

[0110] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0111] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0112] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0113] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0114] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0115] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0116] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling, characterized in that, include: Construct dynamic resource profiles of terminal devices and perform spatiotemporal sparse field modeling to generate sparse data; The process of constructing a dynamic resource profile of the terminal device and performing spatiotemporal sparse field modeling to generate sparse data includes: The system collects spatiotemporal data streams containing timestamps, geographic coordinates, and task attributes through terminal devices, and discretizes the spatiotemporal data streams into a three-dimensional spatiotemporal grid. Based on the amount of data, grid volume, information entropy of data distribution, and total number of terminal devices within the three-dimensional spatiotemporal grid, calculate the privacy strength index of each three-dimensional spatiotemporal grid. Based on the dynamic resource profile and sparse data, a hierarchical differential privacy protection is constructed, and the data after differential privacy protection is verified by zero-knowledge scheduling. Based on the dynamic resource profile and sparse data, a hierarchical differential privacy protection mechanism is constructed, including: At the terminal device layer, the original spatiotemporal data stream is protected by cropping and adding noise. At the edge node layer, Gaussian noise is added to the aggregated region features; At the cloud layer, the cluster resource scheduling system is disturbed through a probabilistic selection mechanism; For tasks that pass the zero-knowledge scheduling verification, cloud-edge-device collaborative scheduling operations are performed to optimize the hierarchical differential privacy protection in order to achieve efficient resource allocation.

2. The cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling as described in claim 1, characterized in that, Also includes: Preprocess the collected spatiotemporal data stream; The size of the geographic grid in the 3D spatiotemporal grid is dynamically adjusted based on the data density of the spatiotemporal data stream. The spatiotemporal grid is subjected to adaptive wavelet transform to generate sparse data.

3. The cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling as described in claim 1, characterized in that, Also includes: The original spatiotemporal data stream is restricted to a preset range by pruning; The probabilistic selection mechanism determines the perturbation intensity based on the reconstruction error and execution efficiency of the original spatiotemporal data stream.

4. The cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling as described in claim 1, characterized in that, The zero-knowledge scheduling verification includes: The scheduling constraints are transformed into arithmetic circuits, and the constraints include spatiotemporal constraints, resource constraints, and task priority constraints. Zero-knowledge proofs are generated at edge nodes, and these proofs are generated based on private inputs of arithmetic circuits, hash values ​​of dynamic resource profiles, and three-dimensional spatiotemporal grid coordinates. The validity of the zero-knowledge proof is verified via a smart contract in the cloud, and the verified task is added to the scheduling queue.

5. The cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling as described in claim 1, characterized in that, Perform cloud-edge-device collaborative scheduling operations, including: At the terminal device layer, task feasibility is assessed through local pre-screening, and the assessment is based on the task execution time. At the edge node layer, task allocation within the region is optimized using a sliding window to minimize the weighted sum of load balancing metrics and resource allocation changes; At the cloud layer, task allocation is optimized using genetic algorithms.

6. A cloud-edge-device collaborative scheduling optimization system based on dynamic resource profiling, wherein the cloud-edge-device collaborative scheduling optimization method based on dynamic resource profiling as described in any one of claims 1 to 5 is characterized in that, include: The dynamic resource modeling and sparsification module is configured to construct a dynamic resource profile of the terminal device and perform spatiotemporal sparse field modeling to generate sparse data. The layered privacy protection and zero-knowledge verification module is configured to construct layered differential privacy protection based on the dynamic resource profile and sparse data, and to perform zero-knowledge scheduling verification on the differentially privacy protected data; The cloud-edge-device collaborative scheduling and optimization module is configured to perform cloud-edge-device collaborative scheduling operations on tasks that have passed the zero-knowledge scheduling verification, and optimize the hierarchical differential privacy protection to achieve efficient resource allocation.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images