Cloud-side collaborative data service system for smart community

Abstract

The invention relates to the technical field of smart community data service, in particular to a smart community-oriented cloud edge collaborative data service system, which comprises an edge data acquisition node module, a region edge collaborative service module, a community center cloud platform and a cloud edge collaborative management module, the edge data acquisition node module comprises a multi-protocol adaptation unit, a heterogeneous data acquisition unit, a local preprocessing unit, a Beidou / GPS dual-mode positioning unit, a backup data acquisition unit and a resident outdoor equipment linkage unit; according to the invention, an end-side cloud cooperative elastic and reliable architecture is constructed, key information transmission is ensured not to be interrupted under an extreme condition through the Beidou positioning and backup acquisition unit, and whole-process redundancy design and rapid fault recovery from the terminal to the cloud are realized in combination with an edge-side disaster recovery backup and cloud-side data synchronization mechanism; and the overall robustness and service continuity of the system for various abnormal conditions are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of smart community data service technology, specifically to a cloud-edge collaborative data service system for smart communities. Background Technology

[0002] A smart community refers to a new type of community ecosystem that utilizes next-generation information technologies such as the Internet of Things, cloud computing, big data, and artificial intelligence to digitally and intelligently upgrade community infrastructure, management mechanisms, and public services, thereby building an efficient, convenient, safe, and green community ecosystem. The smart community data service system serves as its core support, integrating various terminal devices and diverse data within the community to achieve functions such as intelligent security, energy management, and convenient services, aiming to improve residents' quality of life and the efficiency of community governance.

[0003] For example, a big data service system for a grid-based smart community, application number CN202110868964.8 and publication date 20211029, includes a multi-source heterogeneous data acquisition module, a protocol and interface feature library, a protocol identification module, a data extraction module, a data standardization processing module, and a storage module. The protocol and interface feature library stores feature strategies corresponding to various protocols and interfaces. The protocol identification module, based on the multi-source heterogeneous data acquired by the acquisition module, calls the feature strategies in the protocol and interface feature library to perform protocol feature judgment and interface feature judgment sequentially, and selects the corresponding communication protocol based on the judgment results. This technical solution can solve the problem of information silos in the smart community application environment by cooperating with the protocol identification module and the protocol and interface feature library.

[0004] Existing technologies still have significant shortcomings in building smart community data service systems. The systems lack reliable emergency communication capabilities in extreme situations, and the weak disaster recovery architecture makes it difficult to guarantee service continuity. At the same time, the resource scheduling strategy is rigid, making it difficult to achieve dynamic optimization and efficient utilization of computing resources. In addition, there is a lack of enhancement mechanisms for data security and privacy protection throughout the entire process of transmission, processing and evidence storage, and energy efficiency management has not been deeply integrated with real-time data, which restricts the system's reliability, intelligence level and the achievement of green and energy-saving goals.

[0005] In view of this, there is an urgent need to design a cloud-edge collaborative data service system for smart communities to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide a cloud-edge collaborative data service system for smart communities to address the aforementioned shortcomings in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A cloud-edge collaborative data service system for smart communities includes an edge data acquisition node module, a regional edge collaborative service module, a community center cloud platform, and a cloud-edge collaborative management module, wherein: The edge data acquisition node module includes a multi-protocol adaptation unit, a heterogeneous data acquisition unit, a local preprocessing unit, a BeiDou / GPS dual-mode positioning unit, a backup data acquisition unit, and a resident outdoor equipment linkage unit, wherein: The multi-protocol adaptation unit supports multiple IoT protocols such as MQTT, HTTP, CoAP, Modbus, and OPC-UA, enabling unified access for heterogeneous terminals such as smart cameras, environmental sensors, and access control devices. The multi-protocol adaptation unit achieves unified access and data parsing for heterogeneous terminals through protocol conversion middleware. The protocol conversion middleware adopts data standardization output based on JSONSchema to eliminate data format differences between heterogeneous devices. The heterogeneous data acquisition unit collects real-time multi-dimensional heterogeneous raw data, including community security video streams, environmental monitoring data, equipment status data, and public facility status data. The local preprocessing unit performs preliminary cleaning, filtering, and format standardization on the raw data to reduce redundant data uploads. In extreme situations, the BeiDou / GPS dual-mode positioning unit transmits critical alarm information via BeiDou short message communication, providing emergency communication support. The BeiDou / GPS dual-mode positioning unit is configured to automatically switch to BeiDou short message mode in the event of network interruption or extreme situations, and send encrypted alarm messages containing geographic coordinates, timestamps, and anomaly codes via BeiDou RDSS service. The encrypted alarm messages are encrypted using national cryptographic algorithms, and the transmission rate is limited to a preset range to ensure channel availability. The backup data acquisition unit automatically activates the backup sensor to acquire data when the main acquisition device fails, ensuring continuous data acquisition. The resident outdoor equipment linkage unit securely accesses data from residents' own smart doorbells, security cameras, and other outdoor equipment, achieving the integration of public and private resources. The resident outdoor equipment linkage unit securely accesses data from residents' own outdoor equipment through the OAuth2.0 protocol, establishes a device capability abstract model library, implements JSON Schema-based format standardization for heterogeneous device data, and sets a data traffic threshold to limit the maximum bandwidth usage of a single device to no more than a preset proportion of the public channel capacity. The regional edge collaborative service module includes an edge data gateway unit, a lightweight AI inference engine unit, an edge cache database unit, a local service bus unit, a disaster recovery backup cluster unit, and a people flow prediction and resource scheduling unit, wherein: The edge data gateway unit acts as a data aggregation node, realizing multi-protocol data conversion and unified encapsulation; The lightweight AI inference engine unit runs TensorFlow Lite or PyTorch Mobile lightweight models to perform real-time analysis of face recognition and abnormal behavior detection. The response latency is less than a preset threshold of 200ms. The lightweight AI inference engine unit is based on the TensorFlow Lite or PyTorch Mobile framework, integrates a model quantization and compression module, uses fixed-point quantization technology to compress the models delivered from the cloud, and improves the inference speed through layer fusion optimization technology. It supports real-time analysis of face recognition, abnormal behavior detection, and high-altitude object throwing detection, and realizes real-time video stream analysis on edge computing devices. The edge cache database unit uses a time-series database or an in-memory database to store hot data and intermediate processing results. The local service bus unit provides low-latency data query and control services to the property management system and resident APP through RESTful API or message queue interface; The disaster recovery backup cluster unit consists of at least two edge servers. When the primary server fails, it automatically switches to the backup server to ensure high service availability. The disaster recovery backup cluster unit consists of at least two edge servers forming a primary-backup architecture. It adopts an active-passive redundancy architecture based on heartbeat detection. When a preset number of heartbeat packets are lost consecutively, it automatically switches to the backup server. The service interruption time during the switching process is controlled within a preset threshold. It also uses memory mirroring synchronization technology to achieve hot migration of service status and controls data synchronization latency within the transaction cycle. The pedestrian flow prediction and resource scheduling unit predicts future pedestrian flow in each area based on a deep learning LSTM model, and dynamically allocates computing resources according to the prediction results, tilting more AI computing power in densely populated areas. The unit analyzes historical pedestrian flow data and real-time video stream information based on a deep learning LSTM model, with input dimensions including timestamp, area ID, historical pedestrian flow, and environmental factors, and outputs a predicted pedestrian flow value for the future time period; and dynamically adjusts GPU resource allocation according to the prediction results, tilting more AI computing power resources in peak pedestrian flow areas. The community center cloud platform includes a data lake warehouse storage unit, an AI model training and optimization unit, a global operation and maintenance management unit, a multi-source data fusion and analysis unit, and a blockchain evidence storage service unit, wherein: The data lake warehouse storage unit integrates data lake and data warehouse technologies to store structured and unstructured historical data for a long time. The AI ​​model training and optimization unit trains complex LSTM risk prediction and energy consumption optimization models based on historical data, and converts them into lightweight versions through model compression and distributes them to the edge. The global operation and maintenance management unit provides a visual portal to monitor system resource utilization, security events, and edge node health status. The multi-source data fusion analysis unit integrates environmental, security, and energy consumption data to generate a community health index, supporting refined governance decisions. The multi-source data fusion analysis unit generates the community health index through a weighted algorithm, and the index formula is shown below: ; Eenv is the environmental score (including 6 indicators such as PM2.5 and noise), Ssec is the security coefficient (including alarm response speed, event handling rate, intrusion alarm, and fire incident parameters), Penergy is the energy efficiency score (calculated using the energy consumption ratio per unit area), and the weighting coefficients α, β, and γ are dynamically calculated using the entropy weight method or dynamically adjusted according to the community type. The index is updated every 30 minutes and a graded early warning signal is generated. The blockchain-based evidence storage service unit establishes immutable evidence for key transactions such as property fee payment and community voting. Based on smart contracts, it automatically executes community transaction rules, enabling automatic transfer of property fees, automatic release of maintenance funds, and automatic effectiveness of community voting results. Specifically, it generates Merkle tree hash values ​​for property fee payment records and stores them on the blockchain. When the payment rate reaches a certain threshold, the smart contract is automatically triggered to release maintenance funds. Community voting uses zero-knowledge proof technology to protect voting privacy, and the voting results automatically take effect after on-chain consensus. A dual-chain architecture is constructed: the transaction chain stores the property fee payment hash values ​​and generates the Merkle tree root using the SHA-256 algorithm; the audit chain records smart contract execution logs, and the PBFT consensus mechanism ensures multi-party verification. The block generation interval is set to 120 seconds, and a single block can hold a maximum of 200 transactions. The cloud-edge collaborative management module includes a dynamic task offloading decision unit, a resource scheduling and bandwidth allocation unit, a cloud-edge data synchronization unit, an encrypted communication management unit, and an energy consumption optimization control unit, wherein: The dynamic task offloading decision unit dynamically allocates tasks to the edge or cloud based on task attribute computational complexity, real-time requirements, privacy sensitivity, and system status. The resource scheduling and bandwidth allocation unit uses a Markov decision process model to optimize bandwidth allocation, ensuring priority transmission of critical data streams during peak periods. The cloud-edge data synchronization unit manages bidirectional data flow between the edge and the cloud, ensuring edge autonomy during network interruption and automatically synchronizing key data after recovery; The encrypted communication management unit employs national cryptographic algorithms for end-to-end encryption of cloud-edge communication and implements homomorphic encryption processing for sensitive data to enhance data privacy and security. The encrypted communication management unit implements a layered encryption strategy, wherein: Video stream data transmission is encrypted using national cryptographic algorithms; User identity information is processed using homomorphic encryption to ensure that the original data is not decrypted during the cloud analysis process; Structured data uses homomorphic encryption algorithms, supporting aggregation operations in ciphertext state; Key management employs a threshold signature scheme to achieve distributed key escrow; The encrypted communication management unit also uses an identifier cryptography algorithm to establish a device identity authentication system, implements segmented encrypted transmission of video stream data, and realizes hierarchical access to data according to permissions at the edge through a key derivation mechanism. The energy consumption optimization and control unit automatically adjusts the operating parameters of public area lighting and air conditioning based on pedestrian flow prediction data to achieve a green and energy-saving mode of supply on demand.

[0008] In the above technical solution, the cloud-edge collaborative data service system for smart communities provided by the present invention has the following beneficial effects: (1) This invention constructs an elastic and reliable architecture for end-edge-cloud collaboration. Through Beidou positioning and backup acquisition units, it ensures that the transmission of critical information is not interrupted under extreme conditions. Combined with edge-side disaster recovery backup and cloud-edge data synchronization mechanism, it realizes full redundancy design and rapid fault recovery from terminal to cloud, which significantly improves the overall robustness and service continuity of the system in the face of various abnormal situations.

[0009] (2) This invention realizes dynamic resource optimization driven by data and empowered by artificial intelligence. It uses long short-term memory network to accurately predict the dynamic flow of people in the community, and intelligently allocates edge computing resources accordingly and reasonably unloads tasks between the edge and the cloud. At the same time, it automatically adjusts the operating parameters of public facilities based on the predicted data, achieving dual refined management and control of computing resources and energy consumption, effectively improving the system's intelligence level and green energy-saving benefits.

[0010] (3) This invention establishes a multi-layered security protection system that runs through the entire process. At the communication level, it implements hierarchical encryption for different types of data. At the evidence storage level, it uses blockchain and smart contracts to ensure the immutability and automatic execution of key transactions. It constructs a full-link security barrier from data transmission and processing to transaction execution, providing a solid and reliable security foundation for smart community applications.

[0011] (4) This invention breaks through the technical bottleneck of integrating heterogeneous devices and multi-source data. Through unified protocol adaptation and secure access mechanism, it realizes seamless access and collaborative integration of public devices and residents' private devices. It also uses multi-source data fusion analysis to generate an intuitive community health index, providing powerful data insights and decision support for refined community governance. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0013] Figure 1 This is a schematic diagram of the system flow for an embodiment of a cloud-edge collaborative data service system for smart communities according to the present invention.

[0014] Figure 2 This is a flowchart illustrating the operation of an embodiment of a cloud-edge collaborative data service system for smart communities according to the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0016] like Figure 1-2 As shown in the figure, an embodiment of the present invention provides a cloud-edge collaborative data service system for smart communities, including an edge data acquisition node module, a regional edge collaborative service module, a community center cloud platform, and a cloud-edge collaborative management module, wherein: The edge data acquisition node module includes a multi-protocol adaptation unit, a heterogeneous data acquisition unit, a local preprocessing unit, a BeiDou / GPS dual-mode positioning unit, a backup data acquisition unit, and a residential outdoor equipment linkage unit, among which: The multi-protocol adapter unit supports multiple IoT protocols such as MQTT, HTTP, CoAP, Modbus, and OPC-UA, enabling unified access for heterogeneous terminals such as smart cameras, environmental sensors, and access control devices. The multi-protocol adapter unit achieves unified access and data parsing for heterogeneous terminals through protocol conversion middleware. The protocol conversion middleware adopts data standardization output based on JSON Schema to eliminate data format differences between heterogeneous devices. The heterogeneous data acquisition unit collects diverse and heterogeneous raw data in real time, including community security video streams, environmental monitoring data, equipment status data, and public facility status data. The local preprocessing unit performs preliminary cleaning, filtering, and format standardization on the raw data to reduce redundant data uploads. In extreme situations, the BeiDou / GPS dual-mode positioning unit transmits critical alarm information via BeiDou short message communication, providing emergency communication support. The BeiDou / GPS dual-mode positioning unit is configured to automatically switch to BeiDou short message mode in the event of network interruption or extreme situations, and send encrypted alarm messages containing geographic coordinates, timestamps, and anomaly codes through BeiDou RDSS service. The encrypted alarm messages are encrypted using national cryptographic algorithms, and the transmission rate is limited to a preset range to ensure channel availability. The backup data acquisition unit automatically activates the backup sensor to acquire data when the main acquisition device fails, ensuring continuous data acquisition. The resident outdoor equipment linkage unit securely accesses data from residents' own smart doorbells, security cameras, and other outdoor equipment, achieving the integration of public and private resources. The resident outdoor equipment linkage unit securely accesses data from residents' own outdoor equipment through the OAuth2.0 protocol, establishes an abstract model library of equipment capabilities, implements JSON Schema-based format standardization for heterogeneous equipment data, and sets a data traffic threshold to limit the maximum bandwidth usage of a single device to no more than a preset proportion of the public channel capacity. The regional edge collaborative service module includes an edge data gateway unit, a lightweight AI inference engine unit, an edge cache database unit, a local service bus unit, a disaster recovery backup cluster unit, and a people flow prediction and resource scheduling unit, among which: The edge data gateway unit acts as a data aggregation node, realizing multi-protocol data conversion and unified encapsulation; The lightweight AI inference engine unit runs TensorFlow Lite or PyTorch Mobile lightweight models to perform real-time analysis of face recognition and abnormal behavior detection. The response latency is less than a preset threshold of 200ms. The lightweight AI inference engine unit is based on the TensorFlow Lite or PyTorch Mobile framework and integrates a model quantization and compression module. It uses fixed-point quantization technology to compress the models delivered from the cloud and improves the inference speed through layer fusion optimization technology. It supports real-time analysis of face recognition, abnormal behavior detection, and high-altitude object throwing recognition, and realizes real-time video stream analysis on edge computing devices. The edge cache database unit uses a time-series database or an in-memory database to store hot data and intermediate processing results; The local service bus unit provides low-latency data query and control services to the property management system and resident APP through RESTful API or message queue interface; The disaster recovery backup cluster unit consists of at least two edge servers. When the primary server fails, it automatically switches to the backup server to ensure high service availability. The disaster recovery backup cluster unit consists of at least two edge servers forming a primary-backup architecture. It adopts an active-passive redundancy architecture based on heartbeat detection. When a preset number of heartbeat packets are lost consecutively, it automatically switches to the backup server. The business interruption time during the switching process is controlled within a preset threshold. It also uses memory mirroring synchronization technology to achieve hot migration of service status and control the data synchronization latency within the transaction cycle. The pedestrian flow prediction and resource scheduling unit predicts future pedestrian flow in each area based on a deep learning LSTM model, and dynamically allocates computing resources according to the prediction results, giving more AI computing power to densely populated areas. The unit analyzes historical pedestrian flow data and real-time video stream information based on a deep learning LSTM model. The input dimensions include timestamp, area ID, historical pedestrian flow, and environmental factors, and the output is a predicted value of pedestrian flow for the future time period. It also dynamically adjusts the allocation of GPU resources according to the prediction results, giving more AI computing power to areas with peak pedestrian flow. The community center cloud platform includes a data lake warehouse storage unit, an AI model training and optimization unit, a global operation and maintenance management unit, a multi-source data fusion and analysis unit, and a blockchain evidence storage service unit, among which: The data lake warehouse storage unit integrates data lake and data warehouse technologies to store structured and unstructured historical data for long periods of time. The AI ​​model training and optimization unit trains complex LSTM risk prediction and energy consumption optimization models based on historical data, and then converts them into lightweight versions through model compression and distributes them to the edge. The global operations and maintenance management unit provides a visual portal to monitor system resource utilization, security events, and edge node health status. The multi-source data fusion analysis unit integrates environmental, security, and energy consumption data to generate a community health index, supporting refined governance decisions. The multi-source data fusion analysis unit generates the community health index through a weighted algorithm, and the index formula is shown below: ; Eenv is the environmental score (including 6 indicators such as PM2.5 and noise), Ssec is the security coefficient (including alarm response speed, event handling rate, intrusion alarm, and fire incident parameters), Penergy is the energy efficiency score (calculated using the energy consumption ratio per unit area), and the weighting coefficients α, β, and γ are dynamically calculated using the entropy weight method or dynamically adjusted according to the community type. The index is updated every 30 minutes and a graded early warning signal is generated. The blockchain-based evidence storage service unit establishes immutable evidence for key transactions such as property fee payments and community voting. Based on smart contracts, it automatically executes community transaction rules, enabling automatic transfer of property fees, automatic release of maintenance funds, and automatic effectiveness of community voting results. Specifically, it generates Merkle tree hash values ​​for property fee payment records and stores them on the blockchain. When the payment rate reaches a certain threshold, the smart contract is automatically triggered to release maintenance funds. Community voting uses zero-knowledge proof technology to protect voting privacy, and the voting results automatically take effect after on-chain consensus. A dual-chain architecture is constructed: the transaction chain stores the property fee payment hash values ​​and uses the SHA-256 algorithm to generate the Merkle tree root; the audit chain records smart contract execution logs. The PBFT consensus mechanism ensures multi-party verification, with a block generation interval of 120 seconds and a maximum of 200 transactions per block. The cloud-edge collaborative management module includes a dynamic task offloading decision unit, a resource scheduling and bandwidth allocation unit, a cloud-edge data synchronization unit, an encrypted communication management unit, and an energy consumption optimization control unit, among which: The dynamic task offloading decision unit dynamically allocates tasks to the edge or cloud based on task attributes, computational complexity, real-time requirements, privacy sensitivity, and system status. The resource scheduling and bandwidth allocation unit uses a Markov decision process model to optimize bandwidth allocation, ensuring priority transmission of critical data streams during peak periods. The cloud-edge data synchronization unit manages bidirectional data flow between the edge and the cloud, ensuring edge autonomy during network outages and automatically synchronizing critical data upon recovery. The encrypted communication management unit employs national cryptographic algorithms for end-to-end encryption of cloud-edge communication and implements homomorphic encryption for sensitive data to enhance data privacy and security. The encrypted communication management unit implements a layered encryption strategy, including: Video stream data transmission is encrypted using national cryptographic algorithms; User identity information is processed using homomorphic encryption to ensure that the original data is not decrypted during the cloud analysis process; Structured data uses homomorphic encryption algorithms, supporting aggregation operations in ciphertext state; Key management employs a threshold signature scheme to achieve distributed key escrow; The encrypted communication management unit also uses an identifier cryptography algorithm to establish a device identity authentication system, implements segmented encrypted transmission of video stream data, and enables hierarchical access to data according to permissions at the edge through a key derivation mechanism; The energy consumption optimization control unit automatically adjusts the operating parameters of public area lighting and air conditioning based on pedestrian flow prediction data to achieve a green and energy-saving mode of supplying energy on demand.

[0017] This invention provides a cloud-edge collaborative data service system for smart communities, such as... Figure 1As shown in the attached figures (but those skilled in the art can understand from the description), the system includes an edge data acquisition node module, a regional edge collaborative service module, a community central cloud platform, and a cloud-edge collaborative management module. These four modules are organically integrated through a cloud-edge collaborative mechanism, forming an integrated edge-cloud data service architecture. The edge data acquisition node module is deployed near various terminal devices in the community, responsible for raw data acquisition and preprocessing; the regional edge collaborative service module is located in the community edge data center, providing localized computing and storage services; the community central cloud platform centrally manages global data and models; and the cloud-edge collaborative management module coordinates edge-cloud resources and task allocation to ensure efficient and reliable system operation.

[0018] Implementation method of edge data acquisition node module: The edge data acquisition node module serves as the system's data entry point, comprising a multi-protocol adaptation unit, a heterogeneous data acquisition unit, a local preprocessing unit, a BeiDou / GPS dual-mode positioning unit, a backup data acquisition unit, and a linkage unit for residential outdoor equipment. In specific implementation: Multi-protocol adaptation unit: This unit parses and converts various IoT protocols, including MQTT, HTTP, CoAP, Modbus, and OPC-UA, through a protocol conversion middleware. The middleware uses JSON Schema-based data standardization output, uniformly encapsulating data from heterogeneous devices (such as smart cameras, environmental sensors, and access control devices) into a standardized JSON format, eliminating data format differences. For example, when a Modbus device uploads temperature and humidity data, the middleware converts it into a JSON object containing a timestamp, device ID, and numerical value.

[0019] Heterogeneous data acquisition unit: This unit collects community security video streams, environmental monitoring data (such as PM2.5 and temperature), equipment status data (such as access control switch status), and public facility status data (such as street light energy consumption) in real time. Data is acquired through polling or event triggering, and the sampling frequency can be configured according to business needs.

[0020] Local preprocessing unit: Cleans, filters, and standardizes the format of raw data. For example, it removes redundant frames from video stream data, eliminates outliers (such as temperature and humidity readings that are outside the reasonable range) from environmental data, and converts all data to Avro or Parquet format to improve transmission efficiency.

[0021] BeiDou / GPS Dual-Mode Positioning Unit: Configured to automatically switch to BeiDou short message mode in the event of network interruption or extreme conditions (such as natural disasters). This unit sends encrypted alarm messages via BeiDou RDSS service. The message content includes geographic coordinates, timestamps, and anomaly codes (such as the fire code "F001"). Encryption uses the national cryptographic algorithm SM4, and the transmission rate is limited to within 1kbps to ensure channel availability.

[0022] Backup data acquisition unit: Automatically activates the backup sensor when the primary sensor fails. For example, when the primary temperature and humidity sensor fails, the unit switches to the backup sensor via redundant circuitry and records the fault event log.

[0023] Residential Outdoor Equipment Linkage Unit: Securely accesses residents' own smart doorbells, security cameras, and other devices via the OAuth2.0 protocol. The unit establishes a device capability abstract model library, standardizes heterogeneous device data based on JSON Schema format, and sets data traffic thresholds (e.g., the maximum bandwidth usage of a single device does not exceed 5% of the public channel capacity).

[0024] Implementation method of the regional edge collaborative service module: The regional edge collaborative service module includes an edge data gateway unit, a lightweight AI inference engine unit, an edge cache database unit, a local service bus unit, a disaster recovery backup cluster unit, and a people flow prediction and resource scheduling unit. In specific implementation: Edge data gateway unit: As a data aggregation node, it receives data from multi-protocol adapter units and performs protocol conversion and unified encapsulation. For example, it converts CoAP protocol data packets into HTTP RESTful format and adds metadata (such as data source and timestamp).

[0025] Lightweight AI Inference Engine Unit: Runs lightweight models based on TensorFlow Lite or PyTorch Mobile frameworks. This unit integrates a model quantization and compression module, using fixed-point quantization technology to compress FP32 models delivered from the cloud into INT8 models, and improves inference speed through layer fusion optimization techniques (such as merging convolutional layers and ReLU layers). The engine supports real-time analysis of face recognition, abnormal behavior detection (such as crowd gathering), and high-altitude object throwing detection, and performs real-time processing of video streams on edge computing devices (such as Jetson Nano) with latency controlled within 100ms.

[0026] Edge caching database unit: Uses time-series databases (such as InfluxDB) or in-memory databases (such as Redis) to store hot data (such as environmental data from the last 24 hours) and intermediate processing results (such as intermediate feature vectors for AI inference).

[0027] Local Service Bus Unit: Provides low-latency data query and control services to the property management system and resident apps via RESTful APIs or message queues (such as RabbitMQ). For example, the resident app can query real-time parking space information via API with a response time of less than 200ms.

[0028] Disaster recovery backup cluster unit: Composed of at least two edge servers in a primary-backup architecture, employing an active-passive redundancy architecture based on heartbeat detection. The primary and backup servers send heartbeat packets every second, and automatically switch to the backup server when three consecutive heartbeat packets are lost. The switchover process utilizes memory mirroring synchronization technology to achieve hot migration of service status, keeping business interruption time within 500ms and data synchronization latency within a transaction cycle (e.g., 1 second).

[0029] The pedestrian flow prediction and resource scheduling unit predicts future pedestrian flow in various areas based on a deep learning LSTM model. The model input dimensions include timestamps, area IDs, historical pedestrian flow data, and environmental factors (such as weather), outputting a predicted pedestrian flow value for the next 30 minutes. Based on the prediction results, the unit dynamically adjusts GPU resource allocation, for example, allocating more AI computing power resources to areas with high pedestrian flow (such as community squares), increasing GPU utilization from 50% to 80%.

[0030] Implementation methods of the community center cloud platform: The community center cloud platform includes a data lake warehouse storage unit, an AI model training and optimization unit, a global operation and maintenance management unit, a multi-source data fusion and analysis unit, and a blockchain evidence storage service unit. In specific implementation: Data Lake Warehouse Storage Unit: Integrates data lake and data warehouse technologies, using HDFS to store unstructured data (such as video recordings), using ClickHouse to store structured data (such as device logs), and retaining historical data for long periods (such as more than 5 years).

[0031] The AI ​​model training and optimization unit trains complex models such as LSTM risk prediction and energy consumption optimization based on historical data. After training, the model is compressed (e.g., pruning, quantization) to convert it into a lightweight version and then distributed to edge nodes. For example, a 100MB model trained in the cloud is compressed into a 10MB edge model.

[0032] Global Operations and Maintenance Management Unit: Provides a visual portal (based on Web Dashboard) to monitor system resource utilization (such as CPU and memory), security events (such as unauthorized access), and the health status of edge nodes (such as offline node alarms).

[0033] Multi-source data fusion analysis unit: Generates a community health index through a weighted algorithm. The index formula is: ; Eenv is the environmental score, Ssec is the security coefficient, Penergy is the energy efficiency score, and the weighting coefficients α, β, and γ are dynamically calculated using the entropy weighting method or adjusted according to the community type (such as commercial community or residential community). The index is updated every 30 minutes and a graded warning signal (such as green normal, yellow warning) is generated.

[0034] Blockchain-based evidence storage service unit: This unit automatically executes community affairs rules based on smart contracts. For example, property management fee payment contracts automatically transfer fees from residents' accounts on the 1st of each month; maintenance fund contracts automatically release funds after being approved by a homeowner vote; and community voting contracts automatically tally results and take effect by the deadline. Evidence storage data is recorded on the blockchain using SHA-256 hashing to ensure immutability.

[0035] Implementation method of cloud-edge collaborative management module: The cloud-edge collaborative management module includes a dynamic task offloading decision-making unit, a resource scheduling and bandwidth allocation unit, a cloud-edge data synchronization unit, an encrypted communication management unit, and an energy consumption optimization control unit. In practical implementation: Dynamic task offloading decision unit: Based on task attributes (such as computational complexity, real-time requirements, and privacy sensitivity) and system status (such as edge load and network latency), it determines the allocation of tasks between the edge and the cloud. For example, high real-time face recognition tasks are processed at the edge, while high-complexity model training is performed in the cloud.

[0036] Resource scheduling and bandwidth allocation unit: A Markov decision process model is used to optimize bandwidth allocation. The model considers factors such as network congestion and data priority, ensuring priority transmission of critical data streams (such as alarm videos) during peak periods. The bandwidth allocation strategy is dynamically adjusted every 5 minutes.

[0037] Cloud-edge data synchronization unit: manages bidirectional data flow between the edge and the cloud. In the event of a network outage, edge nodes continue to provide services based on local caching (edge ​​autonomy); after the network is restored, the unit automatically synchronizes critical data (such as event logs) to the cloud, using incremental synchronization to reduce bandwidth consumption.

[0038] Encrypted Communication Management Unit: Implements a layered encryption strategy. Video stream data transmission is encrypted using the national cryptographic algorithm SM4; User identity information is processed using homomorphic encryption (such as the Paillier algorithm) to ensure that the original data is not decrypted during the cloud analysis process; Structured data uses homomorphic encryption algorithms and supports performing aggregation operations (such as summation and averaging) in ciphertext. Key management employs a threshold signature scheme (such as Shamir secret sharing) to achieve distributed key escrow; The unit also uses an identifier cryptography algorithm (such as SM9) to establish a device identity authentication system, implements segmented encrypted transmission of video stream data, and implements hierarchical access to data according to permissions at the edge through a key derivation mechanism (such as property staff can access data in public areas, while residents can only access data from their own devices).

[0039] Energy optimization control unit: Automatically adjusts the operating parameters of public area lighting and air conditioning based on pedestrian flow forecast data. For example, when the predicted pedestrian flow is below a threshold, it dims corridor lights or raises the air conditioning temperature setpoint, achieving energy savings of 15%-20%.

[0040] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A cloud-edge collaborative data service system for smart communities, comprising an edge data acquisition node module, a regional edge collaborative service module, a community center cloud platform, and a cloud-edge collaborative management module, characterized in that: The edge data acquisition node module includes the following: The multi-protocol adapter unit supports multiple IoT protocols, enabling unified access for heterogeneous terminals; The heterogeneous data acquisition unit collects diverse and heterogeneous raw data from the community in real time. The local preprocessing unit cleans, filters, and standardizes the format of the raw data; The BeiDou / GPS dual-mode positioning unit transmits critical alarm information via BeiDou short message service; The backup data acquisition unit automatically activates the backup sensor in the event of a major equipment failure. The residential outdoor equipment linkage unit securely connects to data from residents' own outdoor equipment. The regional edge collaborative service module includes the following: The edge data gateway unit acts as a data aggregation node; The lightweight AI inference engine unit runs lightweight models to perform real-time analysis of facial recognition and abnormal behavior detection; The edge cache database unit uses a time-series database or an in-memory database; The local service bus unit provides low-latency data query and control services through RESTful APIs or message queue interfaces; A disaster recovery and backup cluster unit consists of at least two edge servers; The pedestrian flow prediction and resource scheduling unit predicts pedestrian flow based on an LSTM model; The community center cloud platform includes the following: Data lake warehouse storage units integrate data lake and data warehouse technologies; AI model training and optimization units train complex models and convert them into lightweight versions for distribution to the edge through model compression. The global operations and maintenance management unit provides a visual portal; The multi-source data fusion and analysis unit integrates environmental, security, and energy consumption data to generate a community health index; The blockchain-based evidence storage service unit establishes tamper-proof evidence for key transactions. The cloud-edge collaborative management module includes the following: The dynamic task unloading decision unit calculates the complexity, real-time requirements, privacy sensitivity, and system status based on task attributes. The resource scheduling and bandwidth allocation unit uses a Markov decision process model to optimize bandwidth allocation; The cloud-edge data synchronization unit manages bidirectional data streams; The encrypted communication management unit employs end-to-end encryption and homomorphic encryption using national cryptographic algorithms. The energy consumption optimization control unit automatically adjusts the lighting and air conditioning in public areas based on pedestrian flow prediction.

2. The cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The multi-protocol adaptation unit achieves unified access and data parsing for heterogeneous terminals through a protocol conversion middleware. The protocol conversion middleware adopts data standardization output based on JSON Schema to eliminate data format differences between heterogeneous devices.

3. The cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The BeiDou / GPS dual-mode positioning unit is configured to automatically switch to BeiDou short message mode in the event of network interruption or extreme conditions, and send an encrypted alarm message containing geographic coordinates, timestamps and anomaly codes through the BeiDou RDSS service; the encrypted alarm message is encrypted using the national cryptographic algorithm, and the transmission rate is limited to a preset range to ensure channel availability.

4. The cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The resident outdoor equipment linkage unit securely accesses residents' own outdoor equipment data through the OAuth2.0 protocol, establishes an abstract model library of equipment capabilities, implements JSON Schema-based format standardization for heterogeneous equipment data, and sets a data traffic threshold to limit the maximum bandwidth usage of a single device to no more than a preset proportion of the public channel capacity.

5. A cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The lightweight AI inference engine unit is based on the TensorFlow Lite or PyTorch Mobile framework, integrates a model quantization and compression module, uses fixed-point quantization technology to compress the model delivered from the cloud, and improves the inference speed through layer fusion optimization technology. It supports real-time analysis of face recognition, abnormal behavior detection, and high-altitude object throwing recognition, and realizes real-time video stream analysis on edge computing devices.

6. A cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The disaster recovery backup cluster unit consists of at least two edge servers forming a primary-backup architecture. It adopts an active-passive redundancy architecture based on heartbeat detection. When a preset number of heartbeat packets are lost consecutively, it automatically switches to the backup server. The service interruption time during the switching process is controlled within a preset threshold. It also uses memory mirroring synchronization technology to achieve hot migration of service status, and the data synchronization delay is controlled within the transaction cycle.

7. A cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The pedestrian flow prediction and resource scheduling unit analyzes historical pedestrian flow data and real-time video stream information based on a deep learning LSTM model. The input dimensions include timestamp, area ID, historical pedestrian flow, and environmental factors. The output is a predicted value of pedestrian flow for the future time period. The unit also dynamically adjusts the allocation of GPU resources based on the prediction results, allocating more AI computing power resources to areas with high pedestrian flow.

8. A cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The multi-source data fusion and analysis unit generates a community health index using a weighted algorithm, and the index formula is as follows: ； Eenv is the environmental score, Ssec is the security coefficient, Penergy is the energy efficiency score, and the weighting coefficients α, β, and γ are dynamically calculated using the entropy weighting method or dynamically adjusted according to the community type. The index is updated every 30 minutes and a graded early warning signal is generated.

9. A cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The blockchain-based evidence storage service unit automatically executes community affairs rules based on smart contracts, enabling automatic transfer of property fees, automatic release of maintenance funds, and automatic effectiveness of community voting results.

10. A cloud-edge collaborative data service system for smart communities according to claim 1, characterized in that, The encrypted communication management unit implements a layered encryption strategy, wherein: Video stream data transmission is encrypted using national cryptographic algorithms; User identity information is processed using homomorphic encryption to ensure that the original data is not decrypted during the cloud analysis process; Structured data uses homomorphic encryption algorithms, supporting aggregation operations in ciphertext state; Key management employs a threshold signature scheme to achieve distributed key escrow; The encrypted communication management unit also uses an identifier cryptography algorithm to establish a device identity authentication system, implements segmented encrypted transmission of video stream data, and enables hierarchical access to data according to permissions at the edge through a key derivation mechanism.

Images