Vehicle-mounted real-time road information acquisition system based on Internet of Things and edge computing

Abstract

The invention discloses a vehicle-mounted real-time road information acquisition system based on Internet of Things and edge computing, which comprises an integrated sensor module, a self-adaptive image processing module, an edge computing module, an intelligent task scheduling module, a distributed cooperative processing module, a resource allocation module, a hybrid optimization module and a resource reservation module. The system is decomposed into a plurality of loosely-coupled small services, each service is independently developed, deployed and expanded, the flexibility and maintainability of the system are enhanced, the system can quickly adapt to constantly-changing data processing requirements, an optimal resource allocation strategy is achieved through trial and error learning, and the resource allocation efficiency is improved. According to the distributed computing system and method, efficient utilization of resources is achieved by combining static optimization and dynamic adjustment, the response speed and the processing capacity of the system are improved, through a distributed cooperative processing module, when a certain edge device is insufficient in processing capacity, an adjacent edge device can be requested to assist processing, a distributed computing cluster is formed, and the processing capacity of the system is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and in particular to a vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing. Background Art

[0002] With the development of the Internet of Things (IoT) and edge computing technology, the Internet of Vehicles system plays an increasingly important role in the field of intelligent transportation. Traditional road information collection systems often rely on centralized data centers for data processing, which has high latency and bandwidth consumption problems. With the continuous increase in the number of vehicles, the traditional centralized processing method can no longer meet the needs of modern intelligent transportation systems, especially in terms of real-time performance and data processing capabilities.

[0003] After searching, the invention patent with Chinese patent number CN118230554A discloses a vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing. It first transmits the vehicle operation data of multiple vehicles to the edge computing node through the 5G Internet of Things communication network, then, at the edge computing node, the vehicle operation data of the multiple vehicles are respectively encoded to obtain multiple vehicle operation status fully connected encoding feature vectors, then, at the edge computing node, the multiple vehicle operation status fully connected encoding feature vectors are feature extracted to obtain a sequence of vehicle operation status association feature vectors, then, at the edge computing node, the sequence of vehicle operation status association feature vectors is filtered through the gated feature essential feature network to obtain the road-wide vehicle operation status association characterization feature, and finally, based on the road-wide vehicle operation status association characterization feature, the road congestion status level label is determined.

[0004] However, in the actual use of the above invention, there are specific problems in the following aspects. First, there is a lack of flexible data exchange mechanism between data processing modules, which makes it difficult to adapt to the ever-changing data processing needs; second, in terms of dynamic resource allocation and optimization, there is a lack of the ability to dynamically adjust resources to cope with load changes; in addition, in the specific implementation of data encryption and privacy protection, there is a lack of detailed encryption methods and privacy protection measures; finally, in terms of task scheduling and offloading strategies, there is a failure to describe in detail how to perform effective task offloading according to load conditions.

[0005] Therefore, a vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing is proposed. Summary of the invention

[0006] The purpose of the present invention is to solve the problems in the prior art, such as the lack of a flexible data exchange mechanism between data processing modules, which makes it difficult to adapt to the ever-changing data processing needs; secondly, in terms of dynamic resource allocation and optimization, there is a lack of the ability to dynamically adjust resources to cope with load changes; in addition, in the specific implementation of data encryption and privacy protection, there is a lack of detailed encryption methods and privacy protection measures; finally, in terms of task scheduling and unloading strategies, there is a failure to describe in detail how to perform effective task unloading according to the load conditions, and a vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing is proposed.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing includes:

[0009] Integrated sensor module for collecting data about the vehicle's operating environment, including GPS location information, image data, humidity data, acceleration data, and OBD-II interface data;

[0010] Adaptive image processing module, which is used to split high-resolution images into multiple small blocks and select image blocks containing high-activity areas for processing to reduce the computational load;

[0011] An edge computing module is used to process the data collected by the sensor module in real time and identify traffic conditions through a built-in deep learning model; the edge computing module receives data from the integrated sensor module and the adaptive image processing module, and after preprocessing, identifies traffic conditions using a deep learning model, and passes the processing results to the intelligent task scheduling module;

[0012] An intelligent task scheduling module is used to dynamically adjust the task priority according to the load of the edge device, and decide whether to offload the task to the cloud for processing according to the importance and urgency of the task; the intelligent task scheduling module receives the data processing results from the edge computing module, adjusts the task priority according to the current system load, and offloads the task to the cloud for processing when necessary;

[0013] A distributed collaborative processing module is used to request neighboring edge devices to assist in processing when the processing capacity of an edge device is insufficient, forming a distributed computing cluster to jointly complete the task; the distributed collaborative processing module receives task requests from the intelligent task scheduling module and coordinates resources between edge devices;

[0014] A resource allocation module, used to learn the optimal resource allocation strategy through trial and error; the resource allocation module predicts future resource requirements based on system status and historical data, and optimizes the resource allocation strategy through trial and error learning;

[0015] A hybrid optimization module is used to design a hybrid optimization method by combining traditional optimization methods and deep learning technology; the hybrid optimization module receives the optimization strategy from the resource allocation module, and generates a final resource allocation solution by combining a static optimization method and a dynamic adjustment algorithm;

[0016] The resource reservation module is used to use historical data and machine learning technology to predict future resource requirements and reserve computing resources in advance; the resource reservation module receives the resource allocation plan from the hybrid optimization module, and predicts future resource requirements based on historical data and machine learning technology, and reserves computing resources in advance.

[0017] The above technical solution further includes:

[0018] Preferably, the integrated sensor module specifically includes:

[0019] GPS positioning module, used to obtain vehicle location information in real time;

[0020] A camera module for capturing high-definition images of the vehicle's surroundings;

[0021] Humidity sensor module, used to detect the wetness of the road surface;

[0022] Accelerometer module to measure vehicle acceleration and vibration;

[0023] OBD-II interface module, used to read vehicle internal system data.

[0024] Preferably, the adaptive image processing module specifically includes:

[0025] Use sliding window technique to divide the image into multiple small blocks;

[0026] Identify dynamic objects using motion vector analysis;

[0027] Only image patches containing dynamic objects are processed, static patches are ignored.

[0028] Preferably, the edge computing module specifically includes:

[0029] Lightweight virtualization technology is used to enable rapid deployment of computing resources;

[0030] Use Docker or Kubernetes tools to quickly deploy computing resources;

[0031] Built-in deep learning model for real-time target detection, and use online learning technology to continuously optimize model performance.

[0032] Preferably, the intelligent task scheduling module specifically includes:

[0033] Regularly check device CPU and memory usage;

[0034] Prioritize critical tasks during high-load situations;

[0035] Handle non-critical tasks under low load conditions;

[0036] Decide whether to offload the task to the cloud based on its importance and urgency.

[0037] Preferably, the distributed collaborative processing module specifically includes:

[0038] Evaluate current equipment processing capabilities;

[0039] Send assistance requests to nearby edge devices that are not fully loaded;

[0040] Share computing resources and jointly process tasks.

[0041] Preferably, the resource allocation module specifically includes:

[0042] Define the current system state;

[0043] Choose the best action based on the current state;

[0044] Update the strategy using the Ql earning algorithm.

[0045] Preferably, the hybrid optimization module specifically includes:

[0046] Use convex optimization methods to develop basic resource allocation plans;

[0047] Use deep learning technology to dynamically adjust resource allocation based on real-time data;

[0048] Combining static optimization and dynamic adjustment results, resource allocation is performed.

[0049] Preferably, the resource reservation module specifically includes:

[0050] Analyze historical traffic patterns to predict future resource needs;

[0051] Reserve computing resources in advance based on prediction results;

[0052] Dynamically adjust reserved resources based on actual conditions.

[0053] Preferably, the system further comprises:

[0054] Security and privacy protection module, which is used to protect data transmission using encryption algorithms and record data processing logs;

[0055] Microservice architecture module, which is used to decompose the system into multiple loosely coupled small services, each of which is independently developed, deployed and scaled;

[0056] The cloud collaboration module is used to regularly synchronize non-sensitive data processed by edge computing to the cloud and mine information through big data analysis technology.

[0057] The present invention has the following beneficial effects:

[0058] 1. In the present invention, by adopting the microservice architecture module, the system is decomposed into multiple loosely coupled small services, each of which is independently developed, deployed and expanded, which greatly enhances the flexibility and maintainability of the system, enabling it to quickly adapt to changing data processing needs. Secondly, the optimal resource allocation strategy is realized through trial and error learning, and static optimization and dynamic adjustment are combined to achieve efficient utilization of resources, thereby improving the response speed and processing capacity of the system. In addition, the AES encryption algorithm is used to encrypt the transmission data, and blockchain technology is used to record data processing logs to ensure that the data cannot be tampered with, and user privacy is protected by data desensitization and the principle of least privilege to ensure data security and compliance. At the same time, the task priority is dynamically adjusted according to the importance and urgency of the task, and it is decided whether to offload the task to the cloud for processing according to the load of the edge device, ensuring that key tasks can be processed first and improving the overall efficiency and response speed of the system.

[0059] 2. In the present invention, through the resource reservation module, historical data and machine learning technology are used to predict future resource requirements, and computing resources are reserved in advance according to the prediction results, which improves the response speed and resource utilization of the system and ensures that the system can run smoothly during peak hours. In addition, the distributed collaborative processing module can request neighboring edge devices to assist in processing when the processing capacity of a certain edge device is insufficient, forming a distributed computing cluster, which enhances the processing capacity and reliability of the system and ensures the efficiency and consistency of data processing. Finally, through the cloud collaboration module, non-sensitive data is regularly synchronized to the cloud, and big data analysis technology is used to mine information to provide users with more personalized services. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a system architecture diagram of the vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing proposed in the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] like Figure 1As shown, the vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing proposed by the present invention includes:

[0063] Integrated sensor module for collecting data about the vehicle's operating environment, including GPS location information, image data, humidity data, acceleration data, and OBD-II interface data;

[0064] Adaptive image processing module, which is used to split high-resolution images into multiple small blocks and select image blocks containing high-activity areas for processing to reduce the computational load;

[0065] The edge computing module is used to process the data collected by the sensor module in real time and identify traffic conditions through the built-in deep learning model. The edge computing module receives data from the integrated sensor module and the adaptive image processing module, and after preprocessing, uses the deep learning model to identify traffic conditions and pass the processing results to the intelligent task scheduling module.

[0066] The intelligent task scheduling module is used to dynamically adjust the task priority according to the load of the edge device, and decide whether to offload the task to the cloud for processing according to the importance and urgency of the task; the intelligent task scheduling module receives the data processing results from the edge computing module, adjusts the task priority according to the current system load, and offloads the task to the cloud for processing when necessary;

[0067] The distributed collaborative processing module is used to request neighboring edge devices to assist in processing when the processing capacity of an edge device is insufficient, forming a distributed computing cluster to complete tasks together; the distributed collaborative processing module receives task requests from the intelligent task scheduling module and coordinates resources between edge devices;

[0068] Resource allocation module, used to learn the optimal resource allocation strategy through trial and error; the resource allocation module predicts future resource demand based on system status and historical data, and optimizes resource allocation strategy through trial and error learning;

[0069] The hybrid optimization module is used to design a hybrid optimization method by combining traditional optimization methods and deep learning technology. The hybrid optimization module receives the optimization strategy from the resource allocation module and generates the final resource allocation plan by combining the static optimization method and the dynamic adjustment algorithm.

[0070] The resource reservation module is used to use historical data and machine learning technology to predict future resource requirements and reserve computing resources in advance; the resource reservation module receives the resource allocation plan from the hybrid optimization module, and predicts future resource requirements based on historical data and machine learning technology, and reserves computing resources in advance.

[0071] In one embodiment, the integrated sensor module specifically includes:

[0072] GPS positioning module, used to obtain vehicle location information in real time;

[0073] A camera module for capturing high-definition images of the vehicle's surroundings;

[0074] Humidity sensor module, used to detect the wetness of the road surface;

[0075] Accelerometer module to measure vehicle acceleration and vibration;

[0076] OBD-II interface module, used to read vehicle internal system data.

[0077] In one embodiment, the adaptive image processing module specifically includes:

[0078] Use sliding window technique to divide the image into multiple small blocks;

[0079] Identify dynamic objects using motion vector analysis;

[0080] Only image patches containing dynamic objects are processed, static patches are ignored.

[0081] In one embodiment, the edge computing module specifically includes:

[0082] Lightweight virtualization technology is used to enable rapid deployment of computing resources;

[0083] Use Docker or Kubernetes tools to quickly deploy computing resources;

[0084] Built-in deep learning model for real-time target detection, and use online learning technology to continuously optimize model performance.

[0085] In one embodiment, the intelligent task scheduling module specifically includes:

[0086] Regularly check device CPU and memory usage;

[0087] Prioritize critical tasks during high-load situations;

[0088] Handle non-critical tasks under low load conditions;

[0089] Decide whether to offload the task to the cloud based on its importance and urgency.

[0090] In one embodiment, the distributed collaborative processing module specifically includes:

[0091] Evaluate current equipment processing capabilities;

[0092] Send assistance requests to nearby edge devices that are not fully loaded;

[0093] Share computing resources and jointly process tasks.

[0094] In one embodiment, the resource allocation module specifically includes:

[0095] Define the current system state;

[0096] Choose the best action based on the current state;

[0097] Update the strategy using the Ql earning algorithm.

[0098] In one embodiment, the hybrid optimization module specifically includes:

[0099] Use convex optimization methods to develop basic resource allocation plans;

[0100] Use deep learning technology to dynamically adjust resource allocation based on real-time data;

[0101] Combining static optimization and dynamic adjustment results, resource allocation is performed.

[0102] In one embodiment, the resource reservation module specifically includes:

[0103] Analyze historical traffic patterns to predict future resource needs;

[0104] Reserve computing resources in advance based on prediction results;

[0105] Dynamically adjust reserved resources based on actual conditions.

[0106] In one embodiment, the system further comprises:

[0107] Security and privacy protection module, which is used to protect data transmission using encryption algorithms and record data processing logs;

[0108] Microservice architecture module, which is used to decompose the system into multiple loosely coupled small services, each of which is independently developed, deployed and scaled;

[0109] The cloud collaboration module is used to regularly synchronize non-sensitive data processed by edge computing to the cloud and mine information through big data analysis technology.

[0110] In one embodiment, the integrated sensor module:

[0111] Integrated sensor modules include but are not limited to the following modules:

[0112] GPS positioning module: obtains the vehicle's location information in real time and determines the vehicle's precise geographic location through satellite signals. This module obtains location data once a second and sends the location coordinates (latitude, longitude) to the edge computing module for further processing.

[0113] Camera module: captures high-definition images of the vehicle's surroundings for traffic condition recognition. This module captures images at 30 frames per second and passes the image data to the adaptive image processing module.

[0114] Humidity sensor module: detects the wetness of the road surface to determine whether there is a potential driving risk. This module continuously monitors the road surface humidity and transmits humidity data to the edge computing module.

[0115] Accelerometer module: measures the acceleration and vibration of the vehicle and is used to monitor the vehicle's driving status. This module works continuously and transmits acceleration data and vibration data to the edge computing module.

[0116] OBDI I interface module: reads the vehicle's internal system data, including but not limited to engine status, fuel level, etc. This module reads the data once a second and sends the data vehicle_data to the edge computing module.

[0117] In one embodiment, the adaptive image processing module:

[0118] This module processes high-resolution images through the following steps:

[0119] Image segmentation: The image is divided into multiple small blocks using the sliding window technique, each of which is 64×64 pixels in size.

[0120] Dynamic object recognition: Use motion vector analysis technology to identify dynamic objects in image blocks. Specifically, the module compares the changes between two consecutive frames. If the pixel change of a small block exceeds the threshold T = 10%, the small block is considered to contain a dynamic object.

[0121] Among them, I t Represents the pixel intensity of the current frame, I t-1 It represents the pixel intensity of the previous frame, and block_size represents the side length of the block.

[0122] Image block filtering: Only image blocks containing dynamic objects are processed, and static blocks are ignored. Small blocks containing dynamic objects are filtered out and passed to the edge computing module for further processing.

[0123] In one embodiment, the edge computing module:

[0124] This module is responsible for real-time processing of data collected by the sensor module and identifying traffic conditions through the built-in deep learning model:

[0125] Data collection and preprocessing: Receive data from various sensors and perform necessary preprocessing. The preprocessed data will serve as input for the deep learning model.

[0126] Model application: Use the pre-trained deep learning model YOLO to classify or detect data. The model outputs the recognition result y, such as whether there are pedestrians or obstacles ahead.

[0127] y=Model(x)

[0128] Among them, x represents the input data and y represents the output result.

[0129] Online learning: Use online learning technology to continuously optimize model performance. Specifically, if the edge computing module discovers some new traffic conditions, the model will adjust the parameter w through the online gradient descent method, that is:

[0130]

[0131] Where η is the learning rate, L is the loss function, and x t and t is the current sample and its label.

[0132] In one embodiment, the intelligent task scheduling module:

[0133] This module dynamically adjusts the task priority according to the load of the edge device, and decides whether to offload the task to the cloud for processing based on the importance and urgency of the task:

[0134] Load detection: Regularly detect the CPU and memory usage of the device. If the CPU usage exceeds the preset threshold T CPU =80%, it means the equipment is in a high load state.

[0135] CPU_usage indicates the current CPU usage, and CPU_capacity indicates the total CPU capacity.

[0136] Task classification: Tasks are classified according to their importance and urgency. For example, the task level of emergency event detection (such as accident detection) is higher than that of regular traffic flow statistics.

[0137]

[0138] Task Scheduling: Prioritize critical tasks under high load conditions and non-critical tasks under low load conditions. If the edge device load is too high, the module will prioritize critical tasks, such as emergency event detection.

[0139]

[0140] Task offloading decision: Determine whether to offload tasks to the cloud based on their importance and urgency. If the edge device cannot handle all tasks, non-critical tasks will be offloaded to the cloud.

[0141] In one embodiment, the distributed collaborative processing module:

[0142] When the processing capacity of an edge device is insufficient, the module will request neighboring edge devices to assist in processing and form a distributed computing cluster:

[0143] Capacity assessment: Evaluate the processing capacity of the current device. If the processing capacity of the current device is not sufficient to complete the task, proceed to the next step.

[0144] C current <C threshold

[0145] Among them, C current Indicates the current device processing capability, C threshold Indicates the minimum capability required to handle a task.

[0146] Request sending: Send a request for assistance to nearby edge devices that are not fully loaded. The request contains the task to be processed and its importance level.

[0147] send_request(N,P)

[0148] Where N represents the number of tasks and P represents the importance level of the tasks.

[0149] Resource sharing: Share computing resources and jointly process tasks. After receiving a request, the neighboring device will decide whether to accept the request and participate in task processing based on its own load.

[0150]

[0151] In one embodiment, the resource allocation module:

[0152] This module implements the optimal resource allocation strategy through trial and error learning:

[0153] Status definition: defines the current system status s, including but not limited to CPU usage U CPU 、Memory usage U MEM 、The number of tasks currently to be processed N tasks and its types.

[0154] s=(U CPU ,U MEM ,N tasks )

[0155] Action selection: Select the best action based on the current state, such as increasing / decreasing resources. If the current device load is high, you may need to increase computing resources.

[0156]

[0157] Policy update: Use the Q-learning algorithm or other deep reinforcement learning algorithms to update the policy. Through trial and error learning, the module will gradually learn to take the most appropriate actions in different states to achieve optimal resource allocation. The Q-learning update rules are as follows:

[0158]

[0159] Among them, α is the learning rate, R(s ′ ,a ′ ) is the reward function and γ is the discount factor.

[0160] In one embodiment, the hybrid optimization module:

[0161] This module combines traditional optimization methods and deep learning technology to design a hybrid optimization method:

[0162] Static optimization: Use convex optimization methods to develop a basic resource allocation plan. Static optimization is performed when the system starts or when the system state is relatively stable. Assuming that the resource allocation function f(x) is a convex function, the optimal allocation can be found by solving the minimum problem:

[0163]

[0164] Wherein, x represents the resource allocation vector.

[0165] Dynamic adjustment: Use deep learning technology to dynamically adjust resource allocation based on real-time data. When the system detects load changes or other influencing factors, the module will dynamically adjust resource allocation based on real-time data.

[0166] x ′ =Adjust(x,Δt)

[0167] Here, Δt represents the time step.

[0168] Comprehensive optimization: Combine static optimization and dynamic adjustment results to allocate resources. The final resource allocation plan is the result of combining static optimization and dynamic adjustment to ensure that the system can operate efficiently under different conditions.

[0169] x final =Combine(x,x ′ )

[0170] In one embodiment, the resource reservation module:

[0171] This module uses historical data and machine learning techniques to predict future resource requirements and reserve computing resources in advance:

[0172] Historical data analysis: Analyze historical traffic patterns and predict future resource needs. This module will build a prediction model based on historical data.

[0173]

[0174] Among them, x t Indicates current resource demand and historical data.

[0175] Resource reservation: Reserve computing resources in advance based on the forecast results. The amount of resources R reserved is determined by the forecast demand.

[0176]

[0177] Among them, α is the reservation coefficient.

[0178] Dynamic adjustment: Dynamically adjust reserved resources based on actual conditions. When actual demand changes, the module will readjust reserved resources.

[0179]

[0180] Among them, β is the adjustment coefficient.

[0181] In one embodiment, the security and privacy protection module:

[0182] This module ensures the security of data transmission and protects user privacy:

[0183] Data encryption: Use encryption algorithms to protect data transmission.

[0184] Encryption algorithm selection: AES (Advanced Encryption Standard) algorithm is used for data encryption. The encryption key length is 256 bits.

[0185] Data encryption implementation: Before data transmission, the data is encrypted using an encryption key. The encrypted data format is:

[0186] E k (data)

[0187] Among them, E k Indicates the encryption operation using key k, and data indicates the original data.

[0188] Data decryption implementation: Use the same key to decrypt the encrypted data at the receiving end and restore the original data:

[0189] Dk (E k (data))=data

[0190] Among them, D k Represents a decryption operation using key k.

[0191] Privacy protection: Adopt privacy design principles to consider user privacy.

[0192] Data desensitization: Data involving personal privacy is desensitized. For example, sensitive data such as license plate numbers and personal information will be replaced with randomly generated identifiers.

[0193] Principle of least privilege: Only necessary data is collected and the use of data is ensured to comply with the principle of least privilege, that is, only authorized applications and services are allowed to access relevant data.

[0194] Data recording: Use blockchain technology to record data processing logs to ensure that the data cannot be tampered with.

[0195] Blockchain implementation: Use the Hyper ledgerFabric framework to build a private blockchain network. After each data processing is completed, the data processing log is recorded on the blockchain.

[0196] Data record format: Log records contain information such as timestamp, operator ID, operation type, operation result, etc. Each record is linked to the previous record through a hash function to form a chain structure.

[0197] Hash link: Use the SHA256 hash function to generate a hash value for each record, and link the hash value to the hash value of the previous record to form an unalterable chain structure:

[0198] H i =SHA256(H i-1 ,Record i )

[0199] In one embodiment, the microservice architecture module:

[0200] This module decomposes the system into multiple loosely coupled small services, each of which is independently developed, deployed, and scaled:

[0201] Service partitioning: Divide the system into multiple small services.

[0202] Service definition: Split the entire system into multiple independent services based on system functions, such as GPS positioning service, image processing service, task scheduling service, etc.

[0203] Service interface: defines the communication interface between services to ensure that services can interact through standardized messaging mechanisms.

[0204] Independent development: Each service is developed independently and does not affect each other.

[0205] Development process: The development process of each service is independent of other services, including stages such as requirements analysis, design, coding, and testing.

[0206] Development tools: Use different development tools and technology stacks for service development, such as SpringBoot, Docker, Kubernetes, etc.

[0207] Independent deployment and expansion: Each service is deployed independently and can be expanded independently according to demand.

[0208] Deployment method: Each service can be independently deployed through containerization technologies such as Docker, and managed and scheduled through container orchestration tools such as Kubernetes.

[0209] Scaling mechanism: Dynamically adjust the processing capacity of the service through horizontal scaling (increasing the number of instances) or vertical scaling (increasing the resources of a single instance) based on the actual load of the service.

[0210] In one embodiment, the cloud collaboration module:

[0211] This module synchronizes non-sensitive data processed by edge computing to the cloud regularly, and mines information through big data analysis technology:

[0212] Data synchronization: Regularly synchronize non-sensitive data processed by edge computing to the cloud.

[0213] Synchronization frequency: Set the frequency of data synchronization, which can be synchronized once every hour.

[0214] Synchronization protocol: Use HTTPS protocol for data transmission to ensure the security of data transmission.

[0215] Data analysis: mining more valuable information through big data analysis technology.

[0216] Analysis tools: Use big data processing frameworks such as Hadoop and Spark to analyze data synchronized to the cloud.

[0217] Analysis algorithms: Apply machine learning algorithms to conduct in-depth analysis of data and extract valuable information.

[0218] Personalized service: Provide users with more personalized services.

[0219] User portrait: Build a user portrait based on the user's driving habits, route preferences and other information.

[0220] Personalized recommendations: Based on user portraits and analysis results, provide users with personalized route suggestions, traffic condition warnings and other services.

[0221] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing is characterized by: include: Integrated sensor module for collecting data about the vehicle's operating environment, including GPS location information, image data, humidity data, acceleration data, and OBD-II interface data; Adaptive image processing module, which is used to split high-resolution images into multiple small blocks and select image blocks containing high-activity areas for processing to reduce the computational load; An edge computing module, used for processing the data collected by the sensor module in real time and identifying traffic conditions through a built-in deep learning model; The edge computing module receives data from the integrated sensor module and the adaptive image processing module, identifies traffic conditions using a deep learning model after preprocessing, and passes the processing results to the intelligent task scheduling module; Intelligent task scheduling module, which is used to dynamically adjust the task priority according to the load of edge devices and decide whether to offload the task to the cloud for processing based on the importance and urgency of the task; The intelligent task scheduling module receives the data processing results from the edge computing module, adjusts the task priority according to the current system load, and offloads the task to the cloud for processing when necessary; A distributed collaborative processing module is used to request neighboring edge devices to assist in processing when the processing capacity of an edge device is insufficient, forming a distributed computing cluster to jointly complete the task; the distributed collaborative processing module receives task requests from the intelligent task scheduling module and coordinates resources between edge devices; A resource allocation module, used to learn the optimal resource allocation strategy through trial and error; the resource allocation module predicts future resource requirements based on system status and historical data, and optimizes the resource allocation strategy through trial and error learning; A hybrid optimization module is used to design a hybrid optimization method by combining traditional optimization methods and deep learning technology; the hybrid optimization module receives the optimization strategy from the resource allocation module, and generates a final resource allocation solution by combining a static optimization method and a dynamic adjustment algorithm; Resource reservation module, which uses historical data and machine learning techniques to predict future resource requirements and reserve computing resources in advance; The resource reservation module receives the resource allocation plan from the hybrid optimization module, and predicts future resource requirements based on historical data and machine learning technology, and reserves computing resources in advance.

2. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The integrated sensor module specifically comprises: GPS positioning module, used to obtain vehicle location information in real time; A camera module for capturing high-definition images of the vehicle's surroundings; Humidity sensor module, used to detect the wetness of the road surface; Accelerometer module to measure vehicle acceleration and vibration; OBD-II interface module, used to read vehicle internal system data.

3. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The adaptive image processing module specifically includes: Use sliding window technique to divide the image into multiple small blocks; Identify dynamic objects using motion vector analysis; Only image patches containing dynamic objects are processed, static patches are ignored.

4. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The edge computing module specifically includes: Lightweight virtualization technology is used to enable rapid deployment of computing resources; Use Docker or Kubernetes tools to quickly deploy computing resources; Built-in deep learning model for real-time target detection, and use online learning technology to continuously optimize model performance.

5. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The intelligent task scheduling module specifically includes: Regularly check device CPU and memory usage; Prioritize critical tasks during high-load situations; Handle non-critical tasks under low load conditions; Decide whether to offload the task to the cloud based on its importance and urgency.

6. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The distributed collaborative processing module specifically includes: Evaluate current equipment processing capabilities; Send assistance requests to nearby edge devices that are not fully loaded; Share computing resources and jointly process tasks.

7. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The resource allocation module specifically includes: Define the current system state; Choose the best action based on the current state; Update the policy using the Q-learning algorithm.

8. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The hybrid optimization module specifically includes: Use convex optimization methods to develop basic resource allocation plans; Use deep learning technology to dynamically adjust resource allocation based on real-time data; Combining static optimization and dynamic adjustment results, resource allocation is performed.

9. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The resource reservation module specifically includes: Analyze historical traffic patterns to predict future resource needs; Reserve computing resources in advance based on prediction results; Dynamically adjust reserved resources based on actual conditions.

10. The vehicle-mounted real-time road information collection system based on the Internet of Things and edge computing according to claim 1 is characterized in that: The system further comprises: Security and privacy protection module, which is used to protect data transmission using encryption algorithms and record data processing logs; Microservice architecture module, which is used to decompose the system into multiple loosely coupled small services, each of which is independently developed, deployed and scaled; The cloud collaboration module is used to regularly synchronize non-sensitive data processed by edge computing to the cloud and mine information through big data analysis technology.

Images