Cloud edge cooperative control method

Abstract

The invention discloses a cloud edge cooperative control method, which comprises the following steps: establishing a cloud edge cooperative architecture, collecting data by edge nodes, and selecting a corresponding data compression strategy to compress the data; generating an edge-cloud collaborative uploading data stream; the cooperative control layer executes adaptive load balancing based on a cloud-side node state, task characteristics and a network condition which are monitored in real time so as to realize task allocation and scheduling between the cloud and the side; the cloud receives the data stream transmitted by the edge node, decompresses and processes the compressed data according to the metadata information, completes the task and feeds back an execution result; a node performance prediction model is constructed by using system state monitoring information, a task allocation strategy and a resource allocation and data compression strategy are adjusted in real time according to a prediction result, and cloud-side collaborative optimization is realized. Efficient data transmission, flexible task scheduling and self-adaptive resource management are realized, and the performance of the system in a complex and changeable environment is comprehensively improved.

Description

Technical Field

[0001] The present invention relates to the field of cloud-edge collaborative technology, and in particular to a cloud-edge collaborative control method. Background Art

[0002] With the rapid development of the Internet of Things, edge computing, and cloud computing technologies, various distributed systems (such as intelligent transportation, industrial Internet of Things, and smart home applications) have gradually shown the need for real-time collection, processing, and decision-making of large amounts of device data. The traditional centralized cloud computing architecture relies on the cloud for data processing, storage, and analysis. Although it has powerful computing and storage capabilities, the system has high network bandwidth pressure, data transmission delays, and real-time performance issues due to the large amount of data that needs to be transmitted from edge devices to the cloud. In addition, the cloud performs centralized analysis and processing of massive amounts of data, and cannot fully utilize the computing resources of edge nodes, resulting in low resource utilization and difficulty in improving the efficiency of processing tasks.

[0003] In existing methods, real-time data generated by edge devices (monitoring video streams, sensor alarm information, etc.) usually needs to be transmitted to the cloud first and then processed centrally by the cloud. This centralized processing mode results in long transmission links and high latency, which makes it difficult to meet the processing needs of real-time data. Especially in application scenarios that require rapid response, such as transportation and industrial production, transmission delays will have a significant impact on the overall performance and application effects of the system, resulting in an inability to quickly respond to emergencies or high-priority tasks.

[0004] In the cloud-edge architecture, since edge devices continuously generate a large amount of data, traditional transmission methods usually directly upload raw data or simply compressed data to the cloud. This method not only takes up a lot of bandwidth resources, but also easily causes network congestion and transmission bottlenecks, and cannot adapt to changes in network conditions. At the same time, existing technologies are difficult to flexibly select compression strategies and parameters for different data types (such as video, audio, and sensor readings), and cannot minimize the amount of transmitted data while maintaining data quality and meeting real-time requirements, resulting in low data transmission efficiency and inability to fully optimize bandwidth utilization. Summary of the invention

[0005] In order to address the shortcomings of the prior art, the present invention provides a cloud-edge collaborative control method, which achieves efficient data transmission, flexible task scheduling and adaptive resource management through intelligent compression, optimized load balancing, real-time performance prediction and feedback adjustment, and comprehensively improves the performance of the system in complex and changing environments.

[0006] The technical solution adopted by the present invention to solve the technical problem is: a cloud-edge collaborative control method, comprising the following steps:

[0007] Establish a cloud-edge collaborative architecture, and perform distributed deployment between the cloud, edge nodes, and collaborative control layer. The cloud is responsible for the management of global computing resources and complex task processing, and object model definition and configuration are performed in the cloud; the edge node is responsible for real-time data collection, preprocessing, and compression, and the collaborative control layer is responsible for resource allocation, task scheduling, and data transmission optimization between the cloud and the edge, to achieve collaborative control of the cloud-edge architecture;

[0008] The edge node collects data, performs data feature analysis and classification in real time, and selects a corresponding data compression strategy to compress the data; and implements data objectification according to the object model, and completes the calculation and assignment of object attribute values ​​according to the attribute logic, and generates an edge-cloud collaborative upload data stream;

[0009] The collaborative control layer performs adaptive load balancing based on the real-time monitored cloud-edge node status, task characteristics and network conditions to achieve task allocation and scheduling between cloud and edge;

[0010] The cloud receives the data stream transmitted by the edge node, decompresses and processes the compressed data according to the metadata information, completes the task and feeds back the execution result;

[0011] Utilize system status monitoring information to build a node performance prediction model, and make real-time adjustments to task allocation strategy, resource allocation, and data compression strategy based on the prediction results to achieve cloud-edge collaborative optimization.

[0012] Furthermore, the data feature analysis and classification includes:

[0013] The edge node extracts features from the collected data through a data identification module to identify the data type, importance and real-time requirements;

[0014] Based on the identified data features, the data is divided into high-priority data, ordinary data and low-priority data, and the data is classified into real-time data and non-real-time data according to real-time and importance.

[0015] Furthermore, the selection of the data compression strategy includes:

[0016] A data compression strategy library is set at the edge node, and the data compression strategy library contains a variety of compression algorithms for different data types; the optimal compression algorithm is selected according to the data type, and the compression parameters are adjusted to ensure that the data meets the real-time and bandwidth requirements of network transmission after compression.

[0017] Further, data transmission optimization includes:

[0018] Dynamically adjust the data compression level according to the transmission network status, task priority and data type, and select efficient compression ratio and data sampling method; adopt batch compression and transmission strategy for data with low real-time requirements to reduce transmission frequency and bandwidth occupancy.

[0019] Furthermore, the data decompression process includes:

[0020] When the edge node compresses data, the compression parameters and compression algorithm information are attached to the data packet header and transmitted to the cloud; the cloud decompresses the received data according to the attached compression parameters and compression algorithm information, and verifies the integrity and accuracy of the data after decompression.

[0021] Furthermore, the node performance prediction model includes:

[0022] Performance monitoring modules are deployed on the edge nodes and the cloud to monitor the system load, CPU and memory usage, network bandwidth, and task queue length in real time. Based on the monitored system status, the future status of the system is predicted using a time series prediction model or a long short-term memory network to obtain a prediction result.

[0023] Furthermore, the task allocation and scheduling includes:

[0024] Based on task characteristics, node status and prediction results, real-time tasks are assigned to edge nodes with low latency according to priority, and computing-intensive tasks are assigned to the cloud; the collaborative control layer uses a scheduling algorithm to balance the cloud-edge node load to ensure the timely completion of high-priority tasks and the sequential execution of low-priority tasks.

[0025] Furthermore, the adaptive load balancing includes: building a reinforcement learning model in the collaborative control layer, taking the system state and task queue as input features, and taking the system performance index as the reward function to train the reinforcement learning model; using the trained reinforcement learning model to optimize the task scheduling results and load distribution results in real time.

[0026] Furthermore, the cloud-edge collaborative optimization includes:

[0027] The collaborative control layer provides real-time feedback of system status, and dynamically adjusts data compression results and load balancing results based on task completion time, data transmission delay, and resource utilization;

[0028] If it is detected that the task delay time exceeds the threshold or the network condition deteriorates, the adjustment mechanism will be triggered to reset the transmission path, compression level and task allocation.

[0029] Furthermore, the performance indicators during the system operation are monitored and recorded, and the performance indicators include transmission efficiency, task completion time, resource utilization, and system energy consumption. Based on the system performance indicator evaluation results, parameter tuning and strategy updates are performed to improve the overall performance and resource utilization of the cloud-edge collaborative system.

[0030] The beneficial effects of the present invention are:

[0031] (1) By performing feature analysis, classification and compression strategy selection on the data at the edge node, a variety of compression algorithms and adjustable compression parameters are used for different types of data (real-time monitoring video streams, sensor data, and general status information) to optimize data transmission. According to network conditions such as network bandwidth, latency, and jitter, appropriate compression strategies and transmission methods are selected to ensure that data occupies the minimum network resources while meeting real-time and quality requirements. By dynamically adjusting the compression ratio, data sampling rate, and transmission flow, the network load is reasonably distributed and balanced, significantly improving the data transmission efficiency and bandwidth utilization of the cloud-edge system.

[0032] (2) By building an efficient cloud-edge collaborative architecture between the cloud, edge nodes, and collaborative control layer, adaptive control of data compression, transmission, and task scheduling is achieved based on data characteristics, task priorities, and network conditions. Edge nodes are responsible for data processing with high real-time requirements, and the cloud performs computationally intensive tasks. Through dynamic task allocation and adaptive resource allocation strategies, real-time response and computing efficiency are balanced to reduce data transmission delays and processing time. In this process, the reinforcement learning model is combined to optimize load balancing and resource allocation, realize real-time dynamic adjustment of task allocation strategies, and make full use of the respective advantages of edge computing and cloud computing to maximize global system performance and resource utilization.

[0033] (3) Introduce performance monitoring and prediction mechanisms in the cloud-edge collaborative architecture. The collaborative control layer monitors the system load, node status, task queue, and network status in real time, and uses time series models or long short-term memory networks (LSTM) to predict the future state of the system. Combining the prediction results and real-time feedback, dynamically optimize the compression strategy, task allocation, and transmission path to effectively cope with the variability of the network environment, node status, and task requirements. When it is detected that the task delay exceeds the threshold or the network condition deteriorates, the adjustment mechanism is triggered in time to reselect the optimal compression and scheduling strategy to ensure the system's adaptability and robustness, thereby maintaining the efficient and stable operation of the cloud-edge collaborative system and meeting the different needs of various application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0035] Figure 1It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0036] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0037] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0038] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0039] As shown in the figure, the cloud-edge collaborative control method of the present invention includes the following steps: establishing a cloud-edge collaborative architecture, performing distributed deployment between the cloud, edge nodes and collaborative control layer, the cloud is responsible for the management of global computing resources and complex task processing, and the object model is defined and configured in the cloud; the edge node is responsible for real-time data collection, preprocessing and compression, and the collaborative control layer is responsible for resource allocation, task scheduling and data transmission optimization between the cloud and the edge, so as to realize the collaborative control of the cloud-edge architecture. The edge node collects data, performs data feature analysis and classification in real time, and selects the corresponding data compression strategy to compress the data; and realizes data objectization according to the object model, and completes the calculation and assignment of object attribute values ​​according to the attribute logic, and generates edge-cloud collaborative upload data stream. The collaborative control layer performs adaptive load balancing based on the real-time monitored cloud-edge node status, task characteristics and network conditions to realize task allocation and scheduling between cloud and edge. The cloud receives the data stream transmitted by the edge node, decompresses and processes the compressed data according to the metadata information, completes the task and feeds back the execution result. Utilize system status monitoring information to build a node performance prediction model, and make real-time adjustments to task allocation strategy, resource allocation, and data compression strategy based on the prediction results to achieve cloud-edge collaborative optimization.

[0040] Specifically, first, build a cloud-edge collaborative architecture and perform distributed deployment between the cloud, edge nodes and collaborative control layer. The cloud is responsible for the management of global computing resources and complex task processing, the edge nodes are responsible for real-time data collection, preprocessing and compression, and the collaborative control layer is responsible for resource allocation, task scheduling and data transmission optimization between the cloud and the edge, so as to realize the collaborative control of the cloud-edge architecture.

[0041] Secondly, data collection is performed at the edge node, including obtaining raw data from IoT devices, sensors, and camera terminals, and real-time feature extraction and classification of the collected data through the feature analysis module on the edge node. Based on the data type, real-time requirements, and task priority, a suitable data compression strategy is selected. The compression strategy is selected from the compression strategy library pre-built in the edge node. Different compression algorithms are selected for different types of data (including text, images, video, and audio). The compression parameters (including compression ratio and frame rate) are dynamically adjusted according to the network status (including bandwidth and latency), system load, and task real-time requirements to generate a data stream with the optimal compression effect.

[0042] Subsequently, the collaborative control layer monitors the system status of the cloud-edge nodes in real time, including the computing resource status, task queue, bandwidth status, and network delay information of the edge nodes. Based on the task characteristics and network conditions, the collaborative control layer allocates tasks with high real-time requirements to edge nodes for processing through adaptive load balancing to ensure low latency and high response. At the same time, computing-intensive tasks or tasks with low real-time requirements are allocated to the cloud to balance the system load and improve overall performance.

[0043] After the cloud receives the compressed data stream transmitted by the edge node, it decompresses and processes the data through the cloud decompression module according to the accompanying metadata information (including compression algorithm and compression parameters), restores the original data, performs corresponding task processing, analysis or storage operations, and feeds back the processing results or analysis results to the edge node or other terminals.

[0044] Finally, using system status monitoring information (including task completion time, transmission delay, and node resource occupancy), the collaborative control layer builds a node performance prediction model. Based on historical data and real-time system status, the long short-term memory network (LSTM) or other time series prediction models are used to predict future system load, task execution status, and network status. According to the prediction results, task allocation, resource allocation, and data compression are adjusted in real time to achieve adaptive optimization of the system, ensuring the efficiency, low latency, and high resource utilization of cloud-edge collaborative control, so that the overall system can always maintain optimal performance under different network environments and task requirements.

[0045] Data feature analysis and classification include: edge nodes extract features from collected data through data identification modules to identify data types, importance, and real-time requirements; based on the identified data features, data is divided into high-priority data, ordinary data, and low-priority data, and data is classified into real-time data and non-real-time data based on real-time and importance.

[0046] Specifically, the data identification module is used to identify the data type, distinguish different types of data, including text, audio, image, video and sensor data, and identify the importance of the data in combination with context information, that is, to determine the degree of impact of the data on the current task, and analyze the data's requirements for system real-time performance. For example, control instruction data or video surveillance frame data are identified as data with high real-time requirements; on this basis, the data importance and real-time requirements are further used to divide the data into high priority, ordinary priority and low priority data, and classify them based on real-time requirements. Among them, high-priority data usually contains critical task instructions, real-time monitoring screens or emergency alarm information, which need to be transmitted and processed first to ensure the real-time response of the system; ordinary priority data is processing information with medium timeliness requirements, including ordinary monitoring data and status reports; low-priority data is log information, batch data or other data that can be processed later that has little impact on system real-time performance. Finally, combining the real-time and importance classification, the data is further divided into real-time data and non-real-time data. Real-time data requires transmission and processing within the specified time window to ensure the normal operation and rapid response of the system; non-real-time data can be appropriately buffered or processed in batches to reduce the occupancy of network bandwidth and edge computing resources, and realize efficient utilization and reasonable scheduling of system resources.

[0047] The selection of data compression strategy includes: setting up a data compression strategy library at the edge node, which contains a variety of compression algorithms for different data types; selecting the optimal compression algorithm based on the data type, and adjusting the compression parameters to ensure that the data meets the real-time and bandwidth requirements of network transmission after compression.

[0048] Specifically, the data compression strategy library pre-configures a variety of compression algorithms for different data types and application requirements. Different compression algorithms are suitable for different data features and application scenarios to optimize data transmission efficiency and compression effect. Compression algorithms include but are not limited to text data compression algorithms, audio data compression algorithms, image data compression algorithms, video data compression algorithms, and fast compression methods for real-time data and batch compression methods for large-scale data. After the edge node extracts and classifies the data features, it selects the optimal compression algorithm from the data compression strategy library based on the data type, real-time requirements, data size and importance characteristics to ensure that the compressed data minimizes the amount of data while meeting the network bandwidth and transmission real-time requirements. In this process, the edge node can also dynamically adjust the parameters of the compression algorithm according to the current network status and computing resource system status, such as adjusting the resolution and quality parameters of the image compression algorithm, the frame rate and bit rate of video compression, or the encoding length of text compression, to achieve the best compression ratio and transmission efficiency. Through this adaptive parameter adjustment, it can be ensured that data compression and transmission can be carried out in the best way under different network conditions and task priority conditions, which not only meets real-time requirements but also minimizes the use of bandwidth and computing resources, and realizes efficient compression and transmission of data flows between edge nodes and the cloud.

[0049] Data transmission optimization includes: dynamically adjusting the data compression level according to the transmission network status, task priority and data type, selecting efficient compression ratio and data sampling method; adopting batch compression and transmission strategy for data with low real-time requirements to reduce transmission frequency and bandwidth occupancy. The data decompression process includes: when compressing data at the edge node, attaching compression parameters and compression algorithm information to the data packet header and transmitting it to the cloud; the cloud decompresses the received data according to the attached compression parameters and compression algorithm information, and verifies the data integrity and accuracy after decompression.

[0050] The node performance prediction model includes: deploying performance monitoring modules on edge nodes and the cloud to monitor the system load, CPU and memory usage, network bandwidth, and task queue length in real time; based on the monitored system status, using a time series prediction model or a long short-term memory network to predict the future state of the system and obtain the prediction results.

[0051] Specifically, the data compression method and transmission strategy are dynamically adjusted according to the system network status, task priority and data type to ensure the high efficiency and real-time performance of data transmission. The network bandwidth, delay, jitter and other network conditions are monitored in real time at the edge node, and the appropriate compression level and compression ratio are selected according to the current task priority (including urgent tasks, ordinary tasks, low-priority tasks) and data type (including text, audio, image, video, sensor data); when the network condition is good and the real-time requirement is high, a lower compression ratio is selected to maintain data quality and fast transmission. When the bandwidth is limited or the delay is high, a higher compression ratio is selected to reduce the data volume and speed up the transmission. At the same time, combined with the characteristics of the data type, the compression parameters are dynamically adjusted. For example, the frame rate and resolution of video data can be reduced, the sampling rate of audio data can be reduced, and a stronger compression algorithm can be used for text data; for data with low real-time requirements (including log data and batch data), batch compression and batch transmission strategies are adopted. The data is first accumulated to a certain amount at the edge node and then compressed and transmitted at one time to reduce the bandwidth occupation and resource overhead caused by frequent data transmission, thereby improving the overall system transmission efficiency.

[0052] When the edge node performs data compression, the selected compression algorithm information and compression parameters (including compression ratio, frame rate, sampling rate) are attached to the header of the data packet to form metadata information. Metadata information is used to describe the compression method and decompression parameters to ensure that the data can be correctly decompressed when transmitted to the cloud. After the data is transmitted to the cloud, the cloud receiver selects the corresponding decompression algorithm and decompression parameters according to the metadata information in the header of the data packet to decompress the data stream and restore it to the original data. After decompression is completed, the cloud will check the integrity and accuracy of the decompressed data, including comparing checksums, verifying data hash values, etc., to ensure the correctness and integrity of the data during transmission, and avoid data damage or loss affecting task processing.

[0053] Performance monitoring modules can be deployed on edge nodes and the cloud to monitor and collect system status information in real time, including CPU and memory usage, network bandwidth availability, network latency and jitter, task queue length and system load, etc. Based on the monitored system status, time series prediction models (including ARIMA models) or deep learning models (including long short-term memory networks (LSTM)) are used to predict future system load conditions in order to predict the computing resources, storage resources, network conditions and task queues of edge nodes and the cloud in the future. The prediction results are used to assist task scheduling decisions. The collaborative control layer will dynamically adjust task allocation, data compression strategies and transmission strategies based on the output of the performance prediction model to ensure that in the future system state, cloud-edge collaboration can maintain optimal performance, achieve efficient resource utilization and timely completion of tasks.

[0054] Task allocation and scheduling include: allocating real-time tasks to edge nodes with low latency according to priority based on task characteristics, node status and prediction results, and allocating computing-intensive tasks to the cloud; the collaborative control layer uses scheduling algorithms to balance the cloud-edge node load to ensure the timely completion of high-priority tasks and the sequential execution of low-priority tasks.

[0055] Adaptive load balancing includes: building a reinforcement learning model in the collaborative control layer, using system status and task queue as input features, and system performance indicators (including task delay and energy consumption) as reward functions to train the reinforcement learning model; using the trained reinforcement learning model to optimize task scheduling results and load distribution results in real time.

[0056] Cloud-edge collaborative optimization includes: real-time feedback of system status through the collaborative control layer, dynamic adjustment of data compression results and load balancing results based on task completion time, data transmission delay and resource utilization; if it is monitored that the task delay time exceeds the threshold or the network condition deteriorates, the adjustment mechanism is triggered to reset the transmission path, compression level and task allocation.

[0057] System evaluation and optimization include: monitoring and recording performance indicators during system operation, including transmission efficiency, task completion time, resource utilization, and system energy consumption; based on the system performance indicator evaluation results, parameter tuning and strategy updates are performed to improve the overall performance and resource utilization of the cloud-edge collaborative system.

[0058] Specifically, the collaborative control layer monitors the system status, task queues and network conditions of the cloud and edge nodes in real time, and schedules and allocates tasks according to the priority of each task based on the characteristics of each task (including real-time requirements, computational complexity, and data volume), the resource status of the node (including CPU load, memory usage, and network bandwidth), and performance prediction results (including future load conditions and available resources). For tasks with high real-time requirements and latency sensitivity, they are preferentially assigned to edge nodes with lower latency for processing to ensure fast response and reduce data transmission latency. For tasks that require a large amount of computing resources (including deep learning reasoning, large-scale data analysis) or tasks with low real-time requirements, they are assigned to the cloud for processing to make full use of the high computing power and resource pool of the cloud. At the same time, the collaborative control layer uses advanced scheduling algorithms (including priority queues and dynamic programming) for load balancing to ensure that high-priority tasks can be completed on time and low-priority tasks are executed in sequence, thereby optimizing the overall load and performance of the system.

[0059] A reinforcement learning (RL) model is established in the collaborative control layer. The system status (including resource utilization of edge nodes and cloud, network status, task queue length, etc.) and task queue characteristics (task priority, data volume, computing amount) are used as input features of the model. System performance indicators such as task delay, energy consumption, and resource utilization are used as reward functions. The optimization goal of the reinforcement learning model is established, and the RL model is trained to learn the best task scheduling strategy and load distribution strategy. Through continuous learning and online updating, the RL model can adaptively adjust the task allocation and resource scheduling according to the real-time changes of the system status and the dynamic changes of task requirements, optimize the load balance of cloud-edge collaboration in real time, and improve the overall performance and response speed of the system.

[0060] The collaborative control layer monitors the system operation status in real time, including information such as task completion time, data transmission delay, network bandwidth utilization, resource occupancy of edge nodes and cloud, and continuously adjusts the compression strategy and load balancing strategy through the feedback mechanism. When it is detected that the task delay exceeds the preset threshold, the network is congested or the bandwidth is limited, or the system resource load is unbalanced, the collaborative control layer triggers the strategy adjustment mechanism to optimize and adjust the data compression level, transmission path, task allocation strategy, etc. in real time, and select the optimal parameter configuration to ensure that the cloud-edge collaborative system maintains efficient operation under different conditions and quickly responds to changing task requirements and network environments. During the operation of the system, various performance indicators are continuously monitored and data is recorded, including data transmission efficiency (bandwidth utilization, compression effect), task completion time (time from task allocation to result feedback), resource utilization (CPU and memory usage of edge nodes and cloud), system energy consumption (power consumption during task processing), etc. Based on the monitored performance data, regular system evaluations are carried out, and data analysis tools are used to comprehensively evaluate the system's operating status. Parameter configurations and strategies are optimized and adjusted based on the evaluation results, including parameter tuning of compression strategies, priority adjustment of task scheduling strategies, and updating and optimization of load balancing strategies, so as to improve the overall performance, resource utilization and stability of the cloud-edge collaborative system, and ensure that the system always maintains the best state under changing network conditions and task loads.

[0061] The following takes the intelligent traffic monitoring scenario as an example to illustrate the cloud-edge collaborative control method.

[0062] Step 1: Architecture deployment and initialization

[0063] In the urban intelligent traffic monitoring system, a cloud-edge collaborative architecture is deployed, with edge nodes deployed near cameras or sensors at each traffic intersection and the cloud deployed in the data center. The collaborative control layer is located between the edge and the cloud to achieve traffic data collection, analysis, task allocation and data transmission optimization.

[0064] Step 2: Real-time traffic data collection and feature analysis

[0065] Data collection: Edge nodes collect traffic information such as video streams and sensor data in real time, and cameras monitor vehicle and pedestrian traffic, changes in traffic light status, etc.

[0066] Feature extraction: The data recognition module on the edge node analyzes the video content and sensor data to identify the type and location of objects such as vehicles, pedestrians, and traffic lights. For image and video data, it extracts characteristics such as frame rate, resolution, and motion features; for sensor data, it identifies the timing characteristics and real-time requirements of the data.

[0067] Classification and prioritization: Based on the extracted features, emergency traffic events (including traffic jams, accidents) and real-time control information (including traffic light status) are classified as high-priority data; vehicle flow statistics, ordinary monitoring images, etc. are classified as ordinary priority; long-term traffic log data is classified as low priority.

[0068] Step 3: Compression strategy selection and transmission

[0069] Compression strategy selection: For traffic video stream data, select video compression algorithms with high real-time requirements (including H.265) and lower the frame rate and resolution. For sensor data, use low-latency compression algorithms (including Delta encoding) to ensure that compressed data can be transmitted quickly.

[0070] Transmission optimization: Dynamically adjust compression parameters according to network bandwidth and latency; reduce compression ratio when the network is good to ensure data quality; increase compression ratio to improve transmission speed when the network is congested.

[0071] Step 4: Adaptive load balancing and task scheduling

[0072] The collaborative control layer monitors the network latency, task queue length, and load status of edge nodes and the cloud, and assigns real-time events and high-priority tasks (including traffic accident analysis) to edge nodes for rapid processing. Tasks with large computational load but low real-time requirements (including traffic flow trend analysis) are assigned to the cloud to utilize the cloud's powerful computing resources.

[0073] Step 5: Decompression and processing of cloud data

[0074] The cloud receives the compressed data stream, selects the decompression algorithm based on the attached metadata information, recovers the video stream and sensor data, and conducts further analysis and processing. The processing results (including traffic flow prediction and optimization control scheme) are fed back to the edge node to achieve real-time traffic optimization.

[0075] Step 6: System status monitoring and prediction

[0076] The collaborative control layer obtains system status information such as real-time load and network status through the performance monitoring module, uses LSTM to predict future system load, and dynamically adjusts task scheduling and resource allocation strategies.

[0077] In intelligent traffic monitoring, the focus is on achieving efficient compression and transmission of real-time traffic data through cloud-edge collaboration, dynamically adjusting the load to optimize system response speed and data processing efficiency, and is suitable for monitoring and responding to traffic incidents with high real-time requirements.

[0078] The following takes the industrial Internet of Things scenario as an example to illustrate the cloud-edge collaborative control method.

[0079] Step 1: Deploy industrial edge computing nodes

[0080] In a factory environment, edge computing nodes are deployed near key equipment, including production line sensors, CNC machine tools, etc. The cloud is deployed in the factory's central control room or remote data center, responsible for global data storage and task optimization. The collaborative control layer is used to connect the edge nodes of industrial equipment with the cloud to achieve real-time data collection, analysis and resource management of equipment.

[0081] Step 2: Production system data collection and feature classification

[0082] Data collection: Edge nodes collect production data from industrial equipment, including sensor readings such as temperature, pressure, vibration, as well as the operating status of the equipment, alarm information, etc.

[0083] Feature extraction: The edge node analyzes the characteristics of various sensor data, including data sampling frequency, importance, and real-time requirements; for example, equipment fault alarms and key parameters (including temperature exceeding the limit) are of high priority; periodic status reports are of normal priority; and equipment operation logs are of low priority.

[0084] Step 3: Adaptive data compression and transmission strategy

[0085] Strategy selection: Use lossless compression algorithm for high-priority data (including equipment failure alarms) to ensure integrity and real-time performance; for periodic data, select lossy compression to reduce the amount of data; for low-priority log data, use batch compression algorithm with high compression ratio.

[0086] Transmission method: High-priority alarm information is quickly transmitted to the cloud in real time; ordinary status information is transmitted at a medium rate; log information is transmitted in batches during non-peak hours to reduce bandwidth usage.

[0087] Step 4: Edge task processing and cloud collaboration

[0088] Edge nodes process high-priority alarm information in real time and respond quickly to prevent equipment failures; general status information and low-priority log information are deeply analyzed in the cloud (including production efficiency optimization and predictive maintenance). The collaborative control layer monitors the cloud-edge resource status and dynamically allocates tasks based on task priority and real-time performance to ensure that edge nodes prioritize emergency events.

[0089] Step 5: Reinforcement Learning-Driven Load Balancing

[0090] A reinforcement learning model is built at the collaborative control layer, with equipment status, sensor data, task queues, etc. as input features. The reinforcement learning model is trained with the goal of minimizing task delay and energy consumption, and task scheduling and load balancing strategies are optimized in real time. The reinforcement learning model dynamically updates the task allocation strategy to achieve optimal cloud-edge collaborative control based on the working status of factory equipment and network conditions.

[0091] This method is applied to the industrial Internet of Things, and realizes the efficient collection, compression and processing of real-time data of the production line through cloud-edge collaborative control; optimizes task scheduling and load balancing through reinforcement learning, ensuring the real-time response and fault prediction capabilities of the factory under high-load working conditions.

[0092] The following takes the smart home security scenario as an example to illustrate the cloud-edge collaborative control method.

[0093] Step 1: Smart home edge node and cloud deployment

[0094] In the home environment, edge nodes are deployed near smart home devices, including smart access control, surveillance cameras, smoke sensors, etc. The cloud is deployed in the home network center or remote security service platform to provide storage and global monitoring management. The collaborative control layer is responsible for resource scheduling, task allocation and data transmission optimization of multiple devices in the home.

[0095] Step 2: Security data collection and classification

[0096] Data collection: The edge node collects home monitoring images and sensor data (including door and window status, smoke concentration) in real time, and analyzes the camera image stream, microphone recording and other data.

[0097] Feature analysis: Extract features from video, audio, and sensor data to identify possible abnormal behaviors (including intruders and fires). Based on the importance and real-time nature of the data, high-risk alarms and real-time monitoring images are classified as high-priority data, device status information is classified as normal priority, and low priority is environmental status information.

[0098] Step 3: Edge compression and data transmission optimization

[0099] Strategy selection: For real-time monitoring images and audio streams, select a low-latency compression algorithm suitable for real-time transmission (including H.265); for sensor data such as smoke sensors and door and window status, use simple encoding compression; for environmental status information, use lossy compression to reduce the amount of data.

[0100] Transmission optimization: High-priority alarm information is immediately transmitted to the cloud when anomalies are detected to achieve rapid response; normal priority data is uploaded regularly in a scheduled manner; low-priority data is uploaded in batches when the network is idle to reduce the use of network resources.

[0101] Step 4: Load balancing and collaborative processing based on reinforcement learning

[0102] The collaborative control layer uses a reinforcement learning model to optimize the task scheduling strategy based on the current device status, network status, and task queue to ensure that high-priority alarm information is processed quickly. The reinforcement learning model dynamically adjusts the strategy based on real-time feedback, assigning tasks with high real-time requirements to edge nodes for processing, and assigning analysis results or long-term storage tasks to the cloud to ensure overall system load balancing.

[0103] Step 5: Cloud-edge collaborative feedback and system optimization

[0104] The collaborative control layer monitors the status of home devices and task completion time in real time; if network latency is too high or task execution timeouts occur, the strategy adjustment mechanism is triggered to optimize the transmission path, compression strategy, and task allocation. The performance evaluation module is used to continuously optimize the cloud-edge collaborative strategy based on indicators such as transmission efficiency and task response time to ensure the stability and efficiency of the system in security monitoring scenarios.

[0105] In the smart home security scenario, real-time monitoring and event response of the home environment can be achieved through cloud-edge collaboration; by optimizing data compression and task allocation, efficient use of network resources can be achieved, and the real-time and reliability of home security can be guaranteed.

[0106] In summary, the cloud-edge collaborative control method of the present invention has the following advantages:

[0107] (1) By performing feature analysis, classification and compression strategy selection on the data at the edge node, a variety of compression algorithms and adjustable compression parameters are used for different types of data (real-time monitoring video streams, sensor data, and general status information) to optimize data transmission. According to network conditions such as network bandwidth, latency, and jitter, appropriate compression strategies and transmission methods are selected to ensure that data occupies the minimum network resources while meeting real-time and quality requirements. By dynamically adjusting the compression ratio, data sampling rate, and transmission flow, the network load is reasonably distributed and balanced, significantly improving the data transmission efficiency and bandwidth utilization of the cloud-edge system.

[0108] (2) By building an efficient cloud-edge collaborative architecture between the cloud, edge nodes, and collaborative control layer, adaptive control of data compression, transmission, and task scheduling is achieved based on data characteristics, task priorities, and network conditions. Edge nodes are responsible for data processing with high real-time requirements, and the cloud performs computationally intensive tasks. Through dynamic task allocation and adaptive resource allocation strategies, real-time response and computing efficiency are balanced to reduce data transmission delays and processing time. In this process, the reinforcement learning model is combined to optimize load balancing and resource allocation, realize real-time dynamic adjustment of task allocation strategies, and make full use of the respective advantages of edge computing and cloud computing to maximize global system performance and resource utilization.

[0109] (3) Introduce performance monitoring and prediction mechanisms in the cloud-edge collaborative architecture. The collaborative control layer monitors the system load, node status, task queue, and network status in real time, and uses time series models or long short-term memory networks (LSTM) to predict the future state of the system. Combining the prediction results and real-time feedback, dynamically optimize the compression strategy, task allocation, and transmission path to effectively cope with the variability of the network environment, node status, and task requirements. When it is detected that the task delay exceeds the threshold or the network condition deteriorates, the adjustment mechanism is triggered in time to reselect the optimal compression and scheduling strategy to ensure the system's adaptability and robustness, thereby maintaining the efficient and stable operation of the cloud-edge collaborative system and meeting the different needs of various application scenarios.

[0110] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A cloud-edge collaborative control method, characterized in that: The following steps are involved: Establish a cloud-edge collaborative architecture, and perform distributed deployment between the cloud, edge nodes, and collaborative control layer. The cloud is responsible for the management of global computing resources and complex task processing, and object model definition and configuration are performed in the cloud; the edge node is responsible for real-time data collection, preprocessing, and compression, and the collaborative control layer is responsible for resource allocation, task scheduling, and data transmission optimization between the cloud and the edge, to achieve collaborative control of the cloud-edge architecture; The edge node collects data, performs data feature analysis and classification in real time, and selects a corresponding data compression strategy to compress the data; and implements data objectification according to the object model, and completes the calculation and assignment of object attribute values ​​according to the attribute logic, and generates an edge-cloud collaborative upload data stream; The collaborative control layer performs adaptive load balancing based on the real-time monitored cloud-edge node status, task characteristics and network conditions to achieve task allocation and scheduling between cloud and edge; The cloud receives the data stream transmitted by the edge node, decompresses and processes the compressed data according to the metadata information, completes the task and feeds back the execution result; Utilize system status monitoring information to build a node performance prediction model, and make real-time adjustments to task allocation strategy, resource allocation, and data compression strategy based on the prediction results to achieve cloud-edge collaborative optimization.

2. The cloud-edge collaborative control method according to claim 1, characterized in that: The data feature analysis and classification include: The edge node extracts features from the collected data through a data identification module to identify the data type, importance and real-time requirements; Based on the identified data features, the data is divided into high-priority data, ordinary data and low-priority data, and the data is classified into real-time data and non-real-time data according to real-time and importance.

3. The cloud-edge collaborative control method according to claim 2, characterized in that: The selection of the data compression strategy includes: A data compression strategy library is set at the edge node, and the data compression strategy library contains a variety of compression algorithms for different data types; the optimal compression algorithm is selected according to the data type, and the compression parameters are adjusted to ensure that the data meets the real-time and bandwidth requirements of network transmission after compression.

4. The cloud-edge collaborative control method according to claim 3, characterized in that: Data transfer optimization includes: Dynamically adjust the data compression level according to the transmission network status, task priority and data type, and select efficient compression ratio and data sampling method; adopt batch compression and transmission strategy for data with low real-time requirements to reduce transmission frequency and bandwidth occupancy.

5. The cloud-edge collaborative control method according to claim 4, characterized in that: The data decompression process includes: When the edge node compresses data, the compression parameters and compression algorithm information are attached to the data packet header and transmitted to the cloud; the cloud decompresses the received data according to the attached compression parameters and compression algorithm information, and verifies the integrity and accuracy of the data after decompression.

6. The cloud-edge collaborative control method according to claim 5, characterized in that: The node performance prediction model includes: Performance monitoring modules are deployed on the edge nodes and the cloud to monitor the system load, CPU and memory usage, network bandwidth, and task queue length in real time. Based on the monitored system status, the future status of the system is predicted using a time series prediction model or a long short-term memory network to obtain a prediction result.

7. The cloud-edge collaborative control method according to claim 6, characterized in that: The task allocation and scheduling include: Based on task characteristics, node status and prediction results, real-time tasks are assigned to edge nodes with low latency according to priority, and computing-intensive tasks are assigned to the cloud; the collaborative control layer uses a scheduling algorithm to balance the cloud-edge node load to ensure the timely completion of high-priority tasks and the sequential execution of low-priority tasks.

8. The cloud-edge collaborative control method according to claim 7, characterized in that: The adaptive load balancing includes: building a reinforcement learning model in the collaborative control layer, taking the system state and task queue as input features, and taking the system performance index as the reward function to train the reinforcement learning model; using the trained reinforcement learning model to optimize the task scheduling results and load distribution results in real time.

9. The cloud-edge collaborative control method according to claim 8, characterized in that: The cloud-edge collaborative optimization includes: The collaborative control layer provides real-time feedback of system status, and dynamically adjusts data compression results and load balancing results based on task completion time, data transmission delay, and resource utilization; If it is detected that the task delay time exceeds the threshold or the network condition deteriorates, the adjustment mechanism will be triggered to reset the transmission path, compression level and task allocation.

10. The cloud-edge collaborative control method according to claim 9, characterized in that: Monitor and record performance indicators during system operation, including transmission efficiency, task completion time, resource utilization, and system energy consumption; Based on the system performance indicator evaluation results, parameter tuning and strategy updates are performed to improve the overall performance and resource utilization of the cloud-edge collaborative system.

Images