Distributed monitoring network architecture deployment method

Abstract

The invention relates to a distributed monitoring network architecture deployment method. The method comprises the following steps: acquiring multi-source node planning data associated with a target area, and determining a node planning position in the target area based on the multi-source node planning data and a preset node constraint condition; determining distributed nodes based on the node planning positions; deploying a data reasoning model in the edge node, and determining a check breakpoint synchronization scheme of the edge node based on a data acquisition scene; and encrypting monitoring data by using the spatio-temporal information involved in the edge node and the sensing node in the data transmission process. By adopting the method, the strict requirements of various key scenes on various different types of monitoring data can be met.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method for deploying a distributed monitoring network architecture. Background Technology

[0002] With the rapid development of IoT and edge computing technologies, distributed monitoring networks have become the core architecture for acquiring data from the physical world. Through the collaborative operation of the perception layer, network layer, platform layer, and application layer, they achieve real-time monitoring of multiple regions and indicators. In critical power scenarios such as transmission line operation and maintenance, substation operation monitoring, distribution network dispatch management, new energy power generation monitoring, and industrial power supply assurance, the real-time performance and reliability of data transmission directly affect the application value of the monitoring network. For example, real-time transmission of industrial equipment vibration data needs to support millisecond-level fault alarms; transmission line monitoring data in remote areas (such as tower status and line icing data) needs to be fully preserved even in network outage environments; and urban distribution network monitoring data (such as ring main unit operating parameters and cable temperature data) needs to withstand transmission interference caused by complex electromagnetic environments and building obstructions. Therefore, low-latency transmission and no data loss in network outage environments have become the core requirements for distributed monitoring network data transmission.

[0003] However, existing distributed monitoring network data transmission technologies still have significant shortcomings in terms of low latency assurance, prevention of data loss during network outages, and adaptability to multiple scenarios, making it difficult to meet the stringent requirements of various key scenarios for different types of monitoring data. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, and storage medium for deploying a distributed monitoring network architecture to address the aforementioned technical problems.

[0005] In a first aspect, this disclosure provides a method for deploying a distributed monitoring network architecture, the method comprising: Obtain multi-source node planning data associated with the target area, and determine the node planning positions in the target area based on the multi-source node planning data and pre-set node constraints. Based on the planned locations of the nodes, distributed nodes are determined. The distributed nodes include edge nodes and sensing nodes. The sensing nodes are used to collect monitoring data and transmit it to the edge nodes. The edge nodes are used for inference and data caching. A data inference model is deployed in the edge node, and a verification-based breakpoint synchronization scheme for the edge node is determined based on the data acquisition scenario. The data acquisition scenario includes high-frequency data acquisition scenario and low-bandwidth data acquisition scenario. The data inference model is used to verify the data acquired by the sensing node. The monitoring data is encrypted using the spatiotemporal information involved in the data transmission process of the edge nodes and the sensing nodes.

[0006] Secondly, this disclosure also provides a distributed monitoring network architecture deployment device, the device comprising: The location determination module is used to acquire multi-source node planning data associated with the target area, and determine the node planning location in the target area based on the multi-source node planning data and pre-set node constraints. The node planning module is used to determine distributed nodes based on the planned node locations. The distributed nodes include edge nodes and sensing nodes. The sensing nodes are used to collect monitoring data and transmit it to the edge nodes. The edge nodes are used for inference and data caching. The breakpoint synchronization module is used to deploy a data inference model in the edge node and determine the verification-based breakpoint synchronization scheme of the edge node based on the data acquisition scenario. The data acquisition scenario includes high-frequency data acquisition scenario and low-bandwidth data acquisition scenario. The data inference model is used to verify the data acquired by the sensing node. The encryption module is used to encrypt the monitoring data using the spatiotemporal information involved in the data transmission process between the edge node and the sensing node.

[0007] Thirdly, this disclosure also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in any of the above method embodiments.

[0008] Fourthly, this disclosure also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps in any of the above method embodiments.

[0009] Fifthly, this disclosure also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0010] In the above embodiments, the node planning locations are determined based on multi-source node planning data and preset constraints, avoiding the subjectivity and blindness of traditional manual site selection. By combining core factors such as terrain, cost, and coverage, optimal node layout can be achieved in different scenarios (dense urban areas, complex terrain areas, remote areas, etc.), satisfying the hard requirements of monitoring coverage while controlling node deployment and maintenance costs, and adapting to the dynamic needs of temporary and sudden monitoring tasks, thus improving the feasibility of node planning. A distributed node architecture of sensing nodes and edge nodes is adopted to achieve functional layering and collaboration: sensing nodes focus on front-end data collection, ensuring full-area monitoring coverage of the target area; edge nodes undertake local data inference and caching responsibilities, pushing data processing tasks to the edge, eliminating the need to upload all raw data to the cloud, significantly reducing network transmission pressure and cloud computing power consumption, while improving the real-time performance of data processing. The data inference model deployed at edge nodes can locally verify the validity of data collected by sensing nodes and identify anomalies, directly filtering invalid, tampered, or abnormal data. This ensures the quality of uploaded data from the source and prevents low-value data from consuming transmission and storage resources. A verification-based breakpoint synchronization scheme, matched to data collection scenarios, specifically addresses transmission pain points in different scenarios: it can efficiently handle massive amounts of data in high-frequency scenarios and reduce data transmission volume in low-bandwidth scenarios. Simultaneously, the breakpoint resumption mechanism ensures that data synchronization does not require full retransmission after interruption, improving synchronization efficiency and reliability and ensuring the complete retention of data in complex network environments. Data encryption is achieved using spatiotemporal information during transmission between edge nodes and sensing nodes. Unlike traditional key encryption methods, it eliminates the need for an additional key management system, offering advantages such as lightweight design and ease of deployment. By binding spatiotemporal fingerprints and constructing transmission trajectory chains, triple verification of data source legitimacy, transmission path security, and content integrity can be achieved, effectively preventing risks such as data tampering and unauthorized node access, and improving the credibility of monitoring data throughout the entire collection and transmission process. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the specific embodiments of this disclosure or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0012] Figure 1 This is a schematic diagram of the application environment for a distributed monitoring network architecture deployment method in one embodiment; Figure 2 This is a flowchart illustrating a distributed monitoring network architecture deployment method in one embodiment; Figure 3This is a schematic block diagram of a distributed monitoring network architecture deployment device in one embodiment; Figure 4 This is a schematic diagram of the internal structure of a computer device in one embodiment. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this disclosure.

[0014] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings herein are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0015] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0016] As described in the background section, existing distributed monitoring networks suffer from the following main problems in data transmission: 1. Insufficient latency assurance. Traditional distributed monitoring networks generally adopt a centralized transmission architecture of "sensing layer-cloud". All data collected by the sensing layer needs to be uploaded to the remote cloud for processing. The round-trip data transmission and cloud computing power scheduling lead to a significant increase in transmission latency, making it difficult to meet the real-time requirements of equipment fault alarms and pollutant exceedance warnings (for example, when the temperature of industrial equipment exceeds the standard, an immediate shutdown command needs to be triggered, and the transmission latency of the traditional architecture may lead to the expansion of the fault). At the same time, the selection of existing network layer links lacks scenario-based adaptability. For example, high-bandwidth but high-power transmission links are still used in complex terrain in the field, or long-distance low-bandwidth links are used for nodes in the park, which further aggravates the transmission latency. In addition, different sensing nodes use heterogeneous communication protocols (such as LoRa, NB-IoT, Modbus, etc.). The existing protocol conversion mechanism is complex and the data parsing time is long, which indirectly increases the transmission latency.

[0017] 2. High risk of data loss during network outages. Distributed monitoring networks are often deployed in scenarios such as areas without network coverage in the wild, urban areas with signal interference, and complex electromagnetic environments in industrial plants, where network outages are frequent occurrences. Current technologies have limited caching capabilities for sensing nodes and edge nodes, and lack efficient breakpoint resumption mechanisms: most solutions use fixed-time-granularity data storage, which easily leads to "empty data blocks" (no valid monitoring data) or "extremely large data blocks" (sudden high-frequency data collection), resulting in wasted caching resources; after network recovery, all unsynchronized data needs to be retransmitted, consuming significant bandwidth and potentially leading to further data loss due to transmission interruptions. Furthermore, existing redundant node configurations often use a "fixed N+1" model, maintaining a fixed redundancy ratio regardless of network status. This means that backup resources cannot be quickly activated during network outages, leaving core monitoring data at risk of loss.

[0018] 3. Insufficient transmission reliability and adaptability. Existing data transmission schemes lack tiered transmission strategies for different scenarios. Emergency data (such as equipment failure data) and routine data (such as daily statistics) share the same transmission channel, causing emergency data transmission to be congested and further reducing real-time performance. At the same time, end-to-end encryption schemes often employ complex key authentication and re-encryption mechanisms, which have poor adaptability and are difficult to meet the transmission needs of low-computing-power nodes in the perception layer (such as MCU-level sensors). Moreover, the encryption process consumes additional computing power and transmission bandwidth, indirectly affecting the achievement of low-latency goals. In addition, traditional network topologies are mostly static structures, which cannot quickly self-heal when links fail, leading to interruptions in data transmission paths and exacerbating the risk of data loss during network outages.

[0019] 4. Existing breakpoint resumption and synchronization solutions are inefficient. Most existing breakpoint resumption solutions are based on fixed-time block or full data transmission, resulting in high data redundancy and low synchronization efficiency. While some solutions employ simple differential transmission, they lack effective verification mechanisms, making them prone to data transmission errors. Furthermore, they fail to consider the continuous nature of monitoring data, leading to persistently high bandwidth consumption. For ultra-large-scale distributed monitoring networks, existing centralized data synchronization models still suffer from high planning latency and large bandwidth consumption, failing to meet the transmission needs of cross-regional and large-scale node operations.

[0020] Therefore, in order to at least partially or completely solve the above-mentioned technical problems, embodiments of this disclosure provide a distributed monitoring network architecture deployment method, which can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with cloud node 104, edge node 106, and sensing node 108 via a network. Terminal 102 can also connect to server 110 via the network. Terminal 102 obtains multi-source node planning data associated with the target area from server 110. Based on the multi-source node planning data and pre-set node constraints, it determines the planned node positions in the target area. Based on the planned node positions, terminal 102 determines distributed nodes, including edge nodes and sensing nodes. The sensing node 108 collects monitoring data and transmits it to the edge node, while the edge node 106 is used for inference and data caching. Terminal 102 deploys a data inference model in the edge node 106 and determines a verification-based breakpoint synchronization scheme for the edge node 106 based on the data acquisition scenario, which includes high-frequency data acquisition scenarios and low-bandwidth data acquisition scenarios. The data inference model is used to verify the data collected by the sensing node 108. Terminal 102 encrypts the monitoring data using the spatiotemporal information involved in the data transmission process between the edge node 106 and the sensing node 108. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, etc. Server 110 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0021] In one embodiment, such as Figure 2 As shown, a distributed monitoring network architecture deployment method is provided, which can be applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps: S202, acquire multi-source node planning data associated with the target area, and determine the node planning positions in the target area based on the multi-source node planning data and pre-set node constraints.

[0022] The target area typically refers to the physical space where a distributed monitoring network needs to be deployed. This includes four typical scenarios: densely populated urban areas, complex terrain areas (mountainous and forested areas), temporary monitoring areas (sudden pollution areas), and remote areas difficult to survey. It serves as the geographical carrier for node planning. Multi-source node planning data typically refers to a structured set of data strongly correlated with the target area, collected from different dimensions to achieve scientific node planning. Examples include: topographic data: geographical features such as elevation, slope, distribution of obstacles (rivers, buildings), and the proportion of signal obstruction areas in the target area; monitoring task requirement data: task-related data such as monitoring indicators (e.g., pollutant concentration, equipment vibration values), monitoring frequency, key monitoring point locations, and monitoring coverage targets; node deployment cost data: economic data such as individual costs of equipment procurement, construction and installation, operation and maintenance, and the regional total cost budget and cost level classification; and auxiliary support data: feasibility data such as communication signal strength distribution, available power supply points (mains / solar), accessibility of manual surveying, and surveying efficiency. Node constraints refer to the hard rules and optimization objectives that must be met during the node planning process, serving as the core basis for selecting and determining node locations. For example, it can include four categories: cost constraints: the upper limit of single-node cost for node deployment and the threshold of total regional cost to avoid cost overruns; coverage constraints: the minimum monitoring coverage threshold for the target area and the requirement of 100% coverage of key monitoring points to eliminate monitoring blind spots; dynamic adaptation constraints: the response latency of node deployment and the frequency of location adjustments in temporary monitoring scenarios to adapt to the target's movement needs; and efficiency constraints: the survey efficiency threshold (e.g., survey time ≤ 30 seconds / square kilometer) and the rate of manual survey substitution to improve the efficiency of planning implementation. Node planning locations typically refer to the specific deployment coordinates (latitude and longitude / grid positions) of sensing nodes (responsible for data collection) and edge nodes (responsible for data inference, caching, and forwarding) within the target area, determined after integrating multi-source data and constraints. This is the direct basis for the physical deployment of the distributed monitoring network. Additionally, in some other exemplary embodiments, sensing nodes can also be mobile devices, such as robots or drones.

[0023] Specifically, the process begins by acquiring multi-source data (topography, monitoring tasks, costs, communication and power supply, etc.) of the target area. This data is then cleaned and normalized, and undeployable areas (such as areas with steep slopes or poor geographical locations), signal blind spots, and available power supply points are marked. The data is then transformed into a parameter set recognizable by the algorithm. Based on the preprocessed data, constraints are converted into quantitative thresholds (e.g., cost ≤ 500,000 RMB, coverage ≥ 95%). Entropy weighting and analytic hierarchy process (AHP) are used to determine the weighting coefficients for cost, coverage, dynamic adaptation, and efficiency, clarifying planning priorities (e.g., prioritizing cost in densely populated urban areas). According to terrain and task requirements matching algorithms (e.g., multi-objective ant colony algorithm, hierarchical site selection, follower-based site selection), constraints and weights are embedded into the algorithm, iteratively generating initial node locations. A triple check is performed: whether deployment is feasible, whether the cost is compliant, and whether the coverage is satisfied. If not, parameters are adjusted and recalculated until all constraints are met, ultimately determining the planned node locations.

[0024] S204. Based on the planned location of the nodes, determine the distributed nodes, which include edge nodes and sensing nodes. The sensing nodes are used to collect monitoring data and transmit it to the edge nodes, and the edge nodes are used for inference and data caching.

[0025] Distributed nodes typically refer to the core hardware node set forming a distributed monitoring network, deployed at planned locations within the target area. They consist of two types of functional nodes: sensing nodes and edge nodes. Nodes can interact with each other via a Mesh self-organizing network protocol, operating without a central node and exhibiting high fault tolerance. Sensing nodes are typically terminal nodes deployed at the monitoring front end, with data acquisition as their core function. They are usually equipped with various sensors (such as temperature, humidity, pollutant, and vibration sensors) and are deployed at a higher density than edge nodes. They are responsible for collecting monitoring data from the target area and transmitting it to the edge nodes. Edge nodes are typically deployed at planned locations with high computing power requirements (such as high ground or core monitoring areas), with local data processing as their core function. They are deployed at a lower density than sensing nodes, and their core responsibilities include: receiving monitoring data transmitted from sensing nodes and verifying its validity; running data inference models to verify data validity; caching high-frequency / critical data to reduce cloud transmission pressure; and performing operations such as data sharding and trajectory fragment generation to support data disaster recovery storage.

[0026] Specifically, based on the scene characteristics and functional positioning of the planned node locations, the type of node to be deployed at each location is determined: if the planned location is a front-end point of a monitoring sub-area (such as a city street or a mountain monitoring point) and requires high-frequency data collection, then sensor nodes are deployed. These locations are typically numerous and widely distributed, enabling full data coverage of the target area. If the planned location is a core point with a wide field of view and good communication conditions (such as a city high point or a mountain ridge) and needs to undertake data aggregation and processing tasks, then edge nodes are deployed. These locations need to cover the transmission range of surrounding sensor nodes to ensure efficient reception of data from them. Based on the terrain and communication signal strength data in the multi-source node planning data, it is ensured that the communication links between edge nodes and surrounding sensor nodes are unobstructed, and the transmission latency is ≤ a preset threshold. According to the matching results, sensor node and edge node hardware are deployed at the corresponding planned node locations: sensor nodes are equipped with sensors and calibrated to ensure that the accuracy of the collected data meets the requirements of the monitoring task; edge nodes are configured with computing resources (such as embedded chips) and storage resources to meet the performance requirements of data inference and caching. Initialize all nodes by writing their unique identifiers and physical coordinates of their deployment locations. Mesh self-organizing protocols can also be pre-installed to ensure nodes automatically identify neighboring nodes upon startup. Core functions can be configured for sensing nodes, such as enabling sensor data acquisition and setting the acquisition frequency (e.g., 1 time / second for high-frequency scenarios, 1 time / minute for regular scenarios). Core functions can also be configured for edge nodes, including deploying data inference models matching the monitoring tasks (e.g., power anomaly identification models, equipment fault diagnosis models) to verify the validity of sensing node data; and configuring data caching strategies and setting caching thresholds (e.g., caching high-frequency data from the past 24 hours).

[0027] In addition, after all nodes are deployed, the Mesh cluster's autonomous networking process can be triggered: nodes discover neighboring nodes and establish links through a preset self-organizing network protocol, forming a hierarchical data transmission network of "sensing nodes → edge nodes". Test commands are issued to the sensing nodes to collect test data and transmit it to the edge nodes; it is then checked whether the edge nodes can successfully receive the data, complete inference verification, and cache; if the verification passes, the distributed node deployment is considered complete; if it fails, the node positions or functional configurations are adjusted, and verification is repeated until the target is met.

[0028] S206, Deploy a data inference model in the edge node and determine a verification-based breakpoint synchronization scheme for the edge node based on the data acquisition scenario. The data acquisition scenario includes a high-frequency data acquisition scenario and a low-bandwidth data acquisition scenario. The data inference model is used to verify the data acquired by the sensing node.

[0029] Among these, the data inference model is typically an algorithm model customized to meet the needs of the monitoring task, and its core function is to identify anomalies in the monitoring data collected by the sensing nodes. The verification-based breakpoint synchronization scheme is a breakpoint resumption synchronization mechanism with data integrity verification functionality. Its core characteristic is that after data synchronization is interrupted, it does not need to retransmit all data; only the unsynchronized parts are resumed, while verification methods ensure the integrity of the synchronized data. High-frequency data acquisition scenarios are monitoring scenarios where sensing nodes collect data frequently and generate large amounts of data, such as real-time tracking of sudden power outages and high-frequency vibration monitoring of power equipment. Low-bandwidth data acquisition scenarios are monitoring scenarios where the target area has limited network bandwidth and weak data transmission capabilities, such as remote mountainous areas and enclosed environments where power data transmission is restricted.

[0030] Specifically, based on the specific needs of the monitoring task, a suitable data inference model is selected and deployed to edge nodes. This model is then used to perform local validity verification of the monitoring data uploaded by the sensing nodes. Subsequently, the data acquisition scenario type of the target area (high frequency or low bandwidth, etc.) is identified, and a corresponding verification-based breakpoint synchronization scheme is matched. Finally, data quality is ensured through model verification, and efficient and reliable data synchronization between edge nodes and cloud nodes is achieved through a scenario-based synchronization scheme, adapting to the monitoring needs of different scenarios. In some exemplary embodiments, the data inference model can be a Random Forest model, a Logistic Regression model, a Support Vector Machine (SVM) model, or a Tiny-LSTM / GRU model, etc. Regardless of the model chosen, a four-layer lightweight architecture can be used to ensure seamless integration with the caching and synchronization modules of edge nodes, as shown in the following architecture: 1. Data Input Layer: Core functions include receiving data packets transmitted from sensing nodes, parsing data formats and fingerprint fields, supporting multi-source heterogeneous data input (such as numerical, waveform, and image monitoring data), and having a built-in format conversion interface, eliminating the need for additional module adaptation.

[0031] 2. Preprocessing layer (lightweight edge version) core functions: Perform minimal data cleaning and feature extraction to avoid excessive computing power. Specifically, it includes: using moving average filtering to remove sensor drift noise instead of complex wavelet transform; extracting core features (such as mean / variance of numerical data, peak / valley of time series data) and discarding redundant features; mapping data to the [0, 1] interval to match model input requirements; and data validation.

[0032] 3. Inference Validation Layer (Core Layer): This layer is divided into a validity validation sublayer and an anomaly identification sublayer. It uses dual-link parallel inference to ensure data quality. The validity validation sublayer determines whether the data is within a reasonable range (such as the normal range of sensor range and monitoring indicators) based on preset threshold rules and model-assisted judgment. Anomaly identification sublayer: Calls the selected machine learning / deep learning model to infer the preprocessed features and identify abnormal data that deviates from the normal pattern; supports incremental inference (inference only on newly added data segments) instead of recalculating the entire data, adapting to the low latency requirements of high-frequency acquisition scenarios.

[0033] 4. Result Output Layer: Core function: Outputs standardized inference result instructions, which are fed back to the caching and synchronization modules of the edge nodes. The instructions are divided into three categories: Trusted data: verified and marked as synchronizable data, pushed to the cache queue; Untrusted data: invalid / abnormal / tampered data, marked as discarded, and an anomaly log is generated; Suspected data: data on the boundary between normal and abnormal, marked as pending review, temporarily stored locally, and judged again later by combining multiple batches of data; The output instruction format is consistent with the breakpoint synchronization scheme interface of the edge nodes, and no additional format conversion is required.

[0034] Furthermore, the training methods for data inference models vary depending on the scenario, and may include: 1. Each edge node independently trains its own initial model based on its local historical monitoring data (e.g., the equipment fault model for the northern area of ​​node A, and the equipment fault model for the southern area of ​​node B), with completely heterogeneous model structures and parameters. Edge nodes form model symbiotic groups based on the correlation of monitoring areas (e.g., adjacent areas, similar equipment); nodes within the group send each other lightweight feature vectors (not complete parameters, only extracting the "key features for fault identification"), rather than parameter differences; after receiving the feature vectors, nodes use an adversarial game algorithm to retain the features in their own models that are adapted to local data, and integrate the common features of other nodes' models to generate a new generation of local models. In this process, there is no cloud involvement, and edge nodes serve as training data and optimization sources for each other. 2. When the edge node runs the model, it records all deviation cases where the inference results do not match the actual operating conditions (e.g., the model determines that the equipment is not faulty, but the actual equipment stops). Each case only stores "key features + deviation type + operating condition label" (no original monitoring data is required), and the edge node uses deviation cases for training. Training can also be performed using the positive and negative samples recorded in each edge node.

[0035] S208, the monitoring data is encrypted using the spatiotemporal information involved in the data transmission process of the edge node and the sensing node.

[0036] Spatiotemporal information typically refers to spatiotemporal characteristic data strongly correlated with nodes and data transmission, including the node's own spatiotemporal attributes (unique device identifier, physical deployment location, data acquisition / forwarding timestamps) and the spatiotemporal characteristics of the transmission path (the order of data flow between nodes, the positional correlation between nodes). Encryption usually refers to encryption based on the trustworthiness verification of spatiotemporal information. By binding spatiotemporal information and constructing a transmission trajectory chain, it achieves triple protection of data source legitimacy, transmission path security, and content integrity, rather than using keys to encrypt and decrypt data. Monitoring data can typically be power-related data, such as transmission line monitoring data, distribution network monitoring data, power plant monitoring data, etc.

[0037] Specifically, after collecting monitoring data, the sensing node generates its own spatiotemporal information and binds it to the data. When the data is transmitted between the sensing node and the edge node, each time it passes through a forwarding node, that node adds its own spatiotemporal information to form a spatiotemporal trajectory chain containing the entire transmission path. After receiving the data, the edge node verifies the consistency of the spatiotemporal information and the continuity of the trajectory chain to confirm that the data has not been tampered with and that the transmission path is legitimate, thereby completing the encryption protection of monitoring data based on spatiotemporal information.

[0038] In the aforementioned distributed monitoring network architecture deployment method, the node planning locations are determined based on multi-source node planning data and preset constraints, avoiding the subjectivity and blindness of traditional manual site selection. By combining core factors such as terrain, cost, and coverage, optimal node layout can be achieved in different scenarios (dense urban areas, complex terrain areas, remote areas, etc.), satisfying the hard requirements of monitoring coverage while controlling node deployment and maintenance costs, and adapting to the dynamic needs of temporary and sudden monitoring tasks, thus improving the feasibility of node planning. A distributed node architecture of sensing nodes and edge nodes is adopted to achieve functional layering and collaboration: sensing nodes focus on front-end data collection, ensuring full-area monitoring coverage of the target area; edge nodes undertake local data inference and caching responsibilities, decentralizing data processing tasks to the edge, eliminating the need to upload all raw data to the cloud, significantly reducing network transmission pressure and cloud computing power consumption, while improving the real-time performance of data processing. The data inference model deployed at edge nodes can locally verify the validity of data collected by sensing nodes and identify anomalies, directly filtering invalid, tampered, or abnormal data. This ensures the quality of uploaded data from the source and prevents low-value data from consuming transmission and storage resources. A verification-based breakpoint synchronization scheme, matched to data collection scenarios, specifically addresses transmission pain points in different scenarios: it can efficiently handle massive amounts of data in high-frequency scenarios and reduce data transmission volume in low-bandwidth scenarios. Simultaneously, the breakpoint resumption mechanism ensures that data synchronization does not require full retransmission after interruption, improving synchronization efficiency and reliability and ensuring the complete retention of data in complex network environments. Data encryption is achieved using spatiotemporal information during transmission between edge nodes and sensing nodes. Unlike traditional key encryption methods, it eliminates the need for an additional key management system, offering advantages such as lightweight design and ease of deployment. By binding spatiotemporal fingerprints and constructing transmission trajectory chains, triple verification of data source legitimacy, transmission path security, and content integrity can be achieved, effectively preventing risks such as data tampering and unauthorized node access, and improving the credibility of monitoring data throughout the entire collection and transmission process.

[0039] In one embodiment, the method for determining the verification-based breakpoint synchronization scheme for the edge nodes based on the data acquisition scenario includes: S302, in response to the data acquisition scenario being a data acquisition scenario other than a high-frequency data acquisition scenario and / or a low-bandwidth data acquisition scenario, the edge node calculates a content fingerprint based on the received monitoring data, and merges the monitoring data based on the content fingerprint and a preset similarity threshold to generate at least one first data block, wherein the content fingerprint is used to uniquely characterize the content features of the corresponding monitoring data.

[0040] Content fingerprints refer to the unique string identifiers generated by edge nodes using hash algorithms (such as SHA-256 and MD5) on monitoring data. Their core characteristic is "unique content correspondence"—monitoring data with the same content generates completely identical content fingerprints, while monitoring data with different content will inevitably have different fingerprints, allowing for rapid determination of data duplication and similarity. The preset similarity threshold is typically a pre-written threshold (usually expressed as a percentage or hash value matching degree, such as 90%) used to determine whether two sets of monitoring data are similar, and is the core criterion for edge nodes to perform data merging operations.

[0041] The first data block refers to the standardized data packet formed by the edge node merging multiple similar monitoring data. It contains meta-information such as the merged core monitoring data, unified content fingerprint, data start and end timestamps, and data block number. It is the basic data unit for subsequent synchronization with cloud nodes.

[0042] Specifically, edge nodes identify the type of the current data acquisition scenario in real time. By retrieving monitoring task configuration parameters (such as acquisition frequency and regional bandwidth data), they determine that the current scenario does not belong to high-frequency data acquisition scenarios or low-bandwidth data acquisition scenarios, triggering content fingerprint calculation and data merging processes. This process can only be executed on trusted data verified by the data inference model; invalid or abnormal data is directly filtered and discarded. Edge nodes receive multiple sets of trusted monitoring data transmitted by sensing nodes and independently calculate the content fingerprint for each set of data according to a preset hash algorithm. For example, for three consecutive collections of "PM2.5 concentration 25μg / m³", identical content fingerprints are generated after calculation; different content fingerprints are generated for data with concentrations of 25μg / m³ and 30μg / m³. Edge nodes retrieve a preset similarity threshold (such as 90%) and compare the content fingerprints of all data pairwise to calculate the fingerprint matching degree: if the fingerprint matching degree of two sets of data is ≥ the preset threshold, they are determined to be similar data (such as continuously collected environmental monitoring data with small fluctuations); if the matching degree is < the preset threshold, they are determined to be differentiated data (such as abnormal data with sudden changes in monitoring indicators). Edge nodes merge data deemed similar, removing redundant and duplicate data fields while retaining valid values ​​for core monitoring indicators (such as mean and extreme values). They assign a unified content fingerprint to the merged data (based on the baseline fingerprint of similar data in the same group), supplement metadata (such as the start and end times of data collection and the sensing node number of the data source), and encapsulate it into one or more first data blocks. Differentiated data are encapsulated separately into independent first data blocks to ensure that key data is not merged or omitted.

[0043] S304, the edge node assigns a unique identifier to each generated first data block, and generates a bidirectional index table based on the unique identifier, the content fingerprint of the first data block, the data start and end time, the data block size, the synchronization status and the check hash value. The bidirectional index table is stored in the edge node and the cloud node of the distributed monitoring network, respectively.

[0044] The data start and end times refer to the collection time range corresponding to all the original monitoring data contained in the first data block (e.g., 2026-01-01 08:00:00 to 2026-01-01 08:10:00), which is key metadata for tracing the data collection period. The data block size refers to the storage volume of a single first data block (in KB / MB), used to plan the transmission order and determine bandwidth usage during synchronization between edge nodes and cloud nodes. The synchronization status indicates the synchronization progress of the first data block between edge nodes and cloud nodes. Common statuses include unsynchronized (data block is only stored on the edge node), synchronized (data block is being transmitted), and synchronized (data block is stored on both ends), serving as the core status basis for the breakpoint synchronization scheme. The verification hash value refers to the hash value calculated for the entire first data block (including monitoring data and metadata), distinct from the content fingerprint that only represents the original data. Its function is to verify whether the first data block is complete and untampered during transmission and storage. A bidirectional index table is a structured lookup table built by edge nodes that stores the core metadata of all first data blocks. It supports bidirectional queries and verification between edge nodes and cloud nodes. Edge nodes can query the cloud synchronization status through the index table, and cloud nodes can verify the integrity of data blocks through the index table. It is the core carrier for implementing verification-based breakpoint synchronization.

[0045] Specifically, after the edge nodes complete the merging and generation of the first data block, they assign a globally unique identification code to each data block (such as a 64-bit sequence number generated based on the snowflake algorithm or other generated codes; the generation method of the identification code is not restricted in this disclosure). This ensures that each data block has a unique identity in the distributed monitoring network, avoiding confusion between different data blocks. For each first data block, the edge nodes extract or calculate the required metadata one by one: extract the generated content fingerprint, the start and end times of the merged data, and the storage size of the data block; calculate the verification hash value for the entire data block (including monitoring data and the above metadata); and initialize the synchronization status of the data block to "unsynchronized" (by default, the data block is only stored locally on the edge node). The edge nodes fill in the unique identifier, content fingerprint, start and end times, data block size, synchronization status, and verification hash value of each first data block into a table according to a preset field format, forming a structured bidirectional index table. The edge node stores a copy of the completed bidirectional index table locally as a benchmark for subsequent data synchronization and verification. At the same time, the edge node pushes the complete contents of the bidirectional index table to the cloud node of the distributed monitoring network. After receiving it, the cloud node synchronously stores an identical index table, so that the two index tables remain initially consistent, providing a basis for subsequent data block synchronization, breakpoint resume, and integrity verification.

[0046] S306, after the edge node generates the first data block, it pushes the new entry of the bidirectional index table to the cloud node; the cloud node compares the local bidirectional index table, and if there is no entry with the same unique identifier, it adds the entry corresponding to the unique identifier and marks the synchronization status as synchronizing. The new entry matches the generated first data block; Among them, a newly added table entry refers to a structured record added to the bidirectional index table after a new data block is generated at the edge node. This record corresponds one-to-one with the first data block and contains a complete set of metadata, including the unique identifier of the data block, content fingerprint, and data start and end times. It is distinct from the existing historical data block records in the index table. The local bidirectional index table refers to a copy of the bidirectional index table stored by the cloud node itself. It serves as the benchmark for data synchronization verification and status tracking between the cloud and the edge node, and its table entry structure is completely consistent with the index table stored at the edge node.

[0047] Specifically, after the edge node completes the generation of the first data block and the entry of the new bidirectional index table, it only extracts the data of the newly added entry (rather than pushing the entire index table) and pushes it to the cloud node through a lightweight communication link. This method can significantly reduce the amount of data transmitted and adapt to the bandwidth requirements of conventional scenarios. After receiving the new entry pushed by the edge node, the cloud node immediately retrieves the locally stored bidirectional index table and compares it with the unique identifier as the core key field: it traverses the unique identifiers of all entries in the local index table to determine if there is a unique identifier that is the same as the new entry; if there is a unique identifier that is the same, the data block is determined to be duplicate data, and the new entry is discarded without performing subsequent operations; if there is no unique identifier that is the same, the data block is determined to be new data, and the next step of adding an entry is performed. The cloud node enters the new entry into the local bidirectional index table, completing the synchronous update of the index table; at the same time, it changes the synchronization status field in the entry from the initial "not synchronized" to "synchronizing", thereby marking that the metadata of the data block has been synchronized, and the actual data block transmission process will soon begin. After the cloud node completes the addition of table entries and status marking, it can send a confirmation message to the edge node that "index table entry synchronization successful, status updated to synchronization in progress", so that the edge node knows when to start the subsequent data block transmission.

[0048] S308, after receiving the first data block, the cloud node performs integrity verification using the content fingerprint and verification hash value. In response to the integrity verification being passed, it updates the synchronization status to synchronized and sends the status change information back to the edge node. The edge node receives the status change information and synchronously updates the synchronization status of the corresponding entry in its local bidirectional index table.

[0049] Integrity verification refers to the process by which cloud nodes verify that the received first data block is "unaltered and without overall loss" by comparing content fingerprints and check hash values. This is a core step in ensuring the trustworthiness of synchronized data. State change information refers to the standardized notification information sent back to the edge nodes by the cloud nodes after completing the synchronization state update. This information includes the unique identifier of the target data block and the latest synchronization state, and is used to trigger updates to the local index tables of the edge nodes.

[0050] Specifically, after marking a new index entry as being synchronized, the cloud node sends a data block transmission command to the edge node. Upon receiving the command, the edge node transmits the first data block with the corresponding unique identifier to the cloud node via the network link. The cloud node recalculates the content fingerprint of the monitoring data within the data block and compares it with the content fingerprint of the corresponding data block in the bidirectional index table. If they match, the data content is determined to be unaltered; if they do not match, the data is determined to have been tampered with, the data block is discarded, and the synchronization status is marked as synchronization failed. The cloud node recalculates the check hash value of the complete first data block (including monitoring data and metadata) and compares it with the check hash value in the index table. If they match, the data block is determined to have no missing or damaged data during transmission; if they do not match, the data block is determined to be incomplete, and a data block retransmission command is sent to the edge node. After the integrity verification is passed, the cloud node immediately updates the synchronization status of the corresponding data block in the bidirectional index table from "synchronized" to "synchronized," marking the data block synchronization process as complete. The cloud node generates status change information containing the unique identifier of the target data block and the latest synchronization status (synchronized), and sends it back to the edge node. After the edge node receives the status change information, it uses the unique identifier as the key field to find the corresponding data block table entry in the local bidirectional index table and updates its synchronization status to synchronized. This completes the full-process synchronization loop for the first data block.

[0051] S310, in response to the interruption of data synchronization between the edge node and the cloud node in the distributed monitoring network, the edge node caches the data in the bidirectional index table corresponding to the newly added first data block locally and changes the synchronization status to unsynchronized.

[0052] Data synchronization interruption refers to a failure in the data transmission link or an abnormal communication connection between the edge node and the cloud node, resulting in the inability to transmit the first data block or index entry normally. Common causes include network fluctuations, bandwidth exhaustion, node hardware failure, and transmission timeout.

[0053] Specifically, edge nodes monitor the synchronization link status with cloud nodes in real time through a preset heartbeat detection mechanism or transmission status feedback mechanism. If an edge node sends the first data block to the cloud but does not receive a receipt confirmation from the cloud within a preset timeout period, or if the link communication module directly reports abnormal signals such as connection disconnection or transmission failure, it is determined that data synchronization is interrupted, and the local caching and status modification process is immediately triggered. Based on the transmission log, the edge node quickly locates the newly added first data block that has not yet been uploaded when synchronization is interrupted, and simultaneously extracts the metadata records (i.e., "data in the bidirectional index table") that match these data blocks from the local bidirectional index table. The edge node synchronously writes the located newly added first data block body and the corresponding index table metadata into the local cache module, and establishes an association mapping according to the unique identifier of the data block to ensure that the cached data is traceable and not confused. The caching strategy follows the first-in-first-out principle, prioritizing the retention of the most recently generated data block to avoid cache overflow. In the local bidirectional index table, the edge node modifies the synchronization status field corresponding to these data blocks that have not been uploaded from the current synchronization to "not synchronized," thus completing the unified marking of the status. The modified status is synchronously written into the cached index data to ensure that the status information is consistent when synchronization is restored.

[0054] S312, in response to the recovery of data synchronization between the edge node and the cloud node in the distributed monitoring network from interruption, based on the synchronization status, the unsynchronized first data block is synchronized.

[0055] Specifically, edge nodes monitor link status in real time through a heartbeat detection mechanism. When the communication link with the cloud node is detected to have returned to normal and meets the preset communication stability threshold (e.g., three consecutive normal heartbeat responses), it is determined that synchronization has recovered from the interruption, and the breakpoint resume transmission process is immediately triggered. The edge node retrieves the bidirectional index table stored locally, using the synchronization status = unsynchronized as the filtering condition, to filter out all the first data blocks cached during the interruption; at the same time, it extracts the index table entries corresponding to these data blocks to ensure that the data blocks correspond one-to-one with the metadata, avoiding omissions or mistransmissions. Following the principle of "index table entries first," the edge node pushes the index table entries corresponding to the filtered unsynchronized data blocks to the cloud node one by one; after receiving the index table entries, the cloud node compares them with the local index table using the unique identifier as the key field. After confirming that there are no duplicate entries, it marks the synchronization status of these entries as synchronized and sends a "data blocks can be received" instruction to the edge node; after receiving the feedback instruction, the edge node transmits the locally cached unsynchronized first data blocks to the cloud node in the order of the index table entries. After receiving the first data block, the cloud node performs a dual integrity verification of content fingerprint and verification hash value. After the verification is successful, the synchronization status of the corresponding entry in the bidirectional index table is updated to synchronized, and the status change information is sent back to the edge node. After receiving the information, the edge node synchronously updates the status of its local bidirectional index table, thus completing the closed loop of the entire process of resuming interrupted transmission.

[0056] In this embodiment, similar monitoring data are filtered and merged based on content fingerprints and similarity thresholds to generate the first data block. This process eliminates redundant and duplicate data while retaining core and valid information, significantly reducing the amount of data to be synchronized subsequently. This lowers the storage pressure and network transmission overhead for edge and cloud nodes, adapting to the high-efficiency data processing needs of typical scenarios. The bidirectional index table of dual-end storage integrates key metadata such as unique identifiers and content fingerprints for data blocks. Unique identifier comparison avoids duplicate synchronization and provides a unified benchmark for data transmission, verification, and status tracking, making the synchronization process traceable and controllable, reducing invalid transmission and verification costs, and improving overall synchronization efficiency. When synchronization is interrupted, unsynchronized data and index information are cached locally and the status is marked. After recovery, the unsynchronized data blocks are accurately retransmitted based on the synchronization status, eliminating the need for full retransmission. This avoids data loss, saves bandwidth and time costs, and improves the reliability of data synchronization in complex environments such as network fluctuations and unstable links.

[0057] In one embodiment, the method for determining the verification-based breakpoint synchronization scheme for the edge nodes based on the data acquisition scenario includes: S402, in response to the data acquisition scenario being a high-frequency data acquisition scenario and / or a low-bandwidth data acquisition scenario, the edge node determines an initial reference value based on the received monitoring data, and calculates the difference based on the initial reference value and each monitoring data, generating an incremental difference sequence composed of the initial reference value and the difference.

[0058] The initial baseline value typically refers to a benchmark reference value determined by the edge node from the received reliable monitoring data (verified by the data inference model). It serves as the basis for calculating the differences in subsequent data. It is usually selected as the first valid data point or the average of the first N data points to ensure data continuity and correlation. The incremental difference sequence refers to a compressed data sequence composed of the initial baseline value plus the differences between each monitoring data point and the baseline value. Its core characteristic is replacing the original full data with the baseline value plus the incremental differences, significantly reducing data volume and adapting to the transmission and storage needs of high-frequency / low-bandwidth scenarios.

[0059] Specifically, the edge node first identifies the current data acquisition scenario. If it is determined to be a high-frequency data acquisition scenario and / or a low-bandwidth data acquisition scenario, this process is triggered. At the same time, only reliable data verified by the data inference model (filtering out invalid and abnormal data) is selected as the data source for subsequent processing. The edge node extracts a benchmark from the filtered reliable monitoring data. For example, the first valid monitoring data can be selected as the initial benchmark value (adapting to scenarios with strong data continuity, such as temperature and humidity monitoring). If there are small initial fluctuations in the data, the average of the first 3-5 valid data can be calculated as the initial benchmark value to improve the rationality of the difference calculation.

[0060] S404, the edge node divides the incremental difference sequence into several consecutive sliding window data segments based on a preset sliding window length, and calculates a check hash value for each sliding window data segment; wherein, the window reference value of the first sliding window data segment is the initial reference value; in two adjacent sliding window data segments, the window reference value of the latter sliding window data segment is the last value of the monitoring data corresponding to the former sliding window data segment. The edge node calculates the difference (difference = current monitoring data - initial reference value) between each subsequent monitoring data and the initial reference value according to the data acquisition time sequence, obtaining a series of incremental differences (which can be positive, negative, or zero). The edge node uses the "initial reference value" as the first item of the sequence, and the subsequently calculated "incremental differences" are concatenated sequentially according to time sequence to form a complete incremental difference sequence, completing the conversion of the original full data into a compressed sequence. For example, if the original monitoring data is [25, 26, 25.5, 27, 26.8], and the first data point 25 is taken as the initial baseline value, the difference is calculated to obtain [0, 1, 0.5, 2, 1.8]. The final generated incremental difference sequence is [25, 0, 1, 0.5, 2, 1.8]. Only this sequence needs to be transmitted to replace the original full data, thus achieving data compression.

[0061] S406, based on the window baseline value, window start and end timestamps and check hash value, encapsulate to obtain a window data packet.

[0062] S408, the edge node transmits the window data packet to the cloud node of the distributed monitoring network; after receiving the window data packet, the cloud node reverses the reconstruction of the monitoring data based on the window reference value and the incremental differential sequence, and performs integrity verification of the window data segment through the verification hash value.

[0063] The window start and end timestamps refer to the monitoring data collection time range corresponding to each sliding window data segment. They consist of the collection time of the first and last data entries within the window, used to mark the time attribution of data and achieve time-series tracing and splicing of monitoring data. A window data packet is a standardized transmission unit formed by edge nodes encapsulating the window baseline value, window start and end timestamps, check hash value, and incremental differential sequence corresponding to a single sliding window data segment according to a preset format. It has a compact structure and is suitable for narrowband transmission requirements in high-frequency / low-bandwidth scenarios. Reverse restoration refers to the process by which cloud nodes, based on the window baseline value and incremental differential sequence in the window data packet, calculate and recover the original monitoring data. It is the inverse operation of incremental differential compression, and its core purpose is to restore the original form of the data to support subsequent analysis. Integrity verification refers to the process by which cloud nodes verify whether the received window data packet is complete and has not been tampered with by comparing the check hash value. It is a core step in ensuring the reliability of data synchronization in high-frequency / low-bandwidth scenarios.

[0064] Specifically, after the edge nodes complete the division of the sliding window data segments, they perform encapsulation operations for each window: extracting the window baseline value corresponding to the window (the first window uses the initial baseline value, and subsequent windows use the last monitoring value of the previous window); extracting the collection time of all data within the window and generating window start and end timestamps; calculating the check hash value of the complete data segment of the window (window baseline value + incremental differential sequence + timestamp); and integrating and encapsulating the above three elements with the incremental differential sequence according to the preset data packet format to generate a window data packet with a unified structure, ensuring that the data packet can be parsed by the cloud nodes. The edge nodes then transmit the encapsulated window data packets sequentially to the cloud nodes of the distributed monitoring network; for low-bandwidth scenarios, the data packets can be further compressed to reduce transmission bandwidth usage. After receiving the window data packet, the cloud node first parses out the window baseline value, incremental difference sequence, and window start and end timestamps within the packet, and then performs reverse reconstruction calculation: for each incremental difference value within the window, the calculation is performed using the formula Original monitoring data = Window baseline value + Incremental difference value; the calculation results are arranged in the order of the incremental difference sequence, and the data time sequence is marked by the window start and end timestamps to reconstruct the original monitoring data corresponding to the sliding window; for multiple consecutive window data packets, the cloud node splices the reconstructed data of each window in the order of the timestamps to form a complete time-series monitoring data chain.

[0065] After the cloud node completes data restoration, it initiates an integrity verification process: recalculates the check hash value for the parsed window data segment (window baseline value + incremental differential sequence + timestamp); compares the recalculated hash value with the check hash value carried in the window data packet; if the two are completely consistent, the window data packet is determined to be complete and untampered, and the restored data is valid; if the two are inconsistent, the data packet is determined to be damaged or tampered with during transmission, and the cloud node sends a data packet retransmission request to the edge node to ensure the accuracy of data synchronization.

[0066] S410, in response to an interruption in data synchronization between the edge node and the cloud node in the distributed monitoring network, the edge node buffers untransmitted window data packets. S412, in response to the recovery of data synchronization between the edge node and the cloud node in the distributed monitoring network from the interruption, the edge node sequentially sends the cached window data packets to the cloud node, and the cloud node receives and splices the window data packets based on the window start and end timestamps.

[0067] Among them, splicing window data packets refers to the operation of cloud nodes integrating multiple received window data packets into a continuous data chain according to the order of the window start and end timestamps, with the aim of restoring the complete and orderly time sequence of monitoring data in the target area.

[0068] Specifically, edge nodes monitor the link status with cloud nodes in real time through heartbeat detection or transmission status feedback mechanisms. When a link interruption is detected (such as transmission timeout or connection loss), the window data packet sending process is immediately paused, and the encapsulated but untransmitted window data packets are quickly located. The edge node stores the located untransmitted window data packets in the local cache module in the order of their generation time and establishes a cache index table (recording the data packet generation time, window start and end timestamps, and storage location) to avoid data packet confusion or loss; at the same time, a cache validity period is set to prevent long-term caching from consuming too many node resources. The edge node continuously monitors the link status, and when it detects that the link has been re-established and the communication interactions are normal for a preset number of consecutive times, it is determined that the synchronization has recovered from the interruption, and the breakpoint resume process is immediately triggered. The edge node sends the cached window data packets one by one and in an orderly manner to the cloud node according to the time order of the cache index table; during the sending process, a low-bandwidth adaptation strategy (such as packet transmission and rate adaptation) is adopted to avoid link congestion caused by batch sending. After receiving the window data packets, the cloud node first verifies their integrity by checking the hash value to ensure that the data packets have not been tampered with or damaged during caching and transmission. Once the verification is successful, the start and end timestamps of each window are extracted, and all data packets are sorted according to their timestamps. Based on the sorting result, multiple window data packets are concatenated into a continuous data chain, laying the foundation for subsequent reverse reconstruction of the complete original monitoring data. The reverse reconstruction process can be found in the steps described above.

[0069] In this embodiment, a differential sequence is generated by combining an initial baseline value with incremental differences. A small amount of difference data replaces the entire original data. Combined with sliding window segmentation, this significantly reduces data transmission and storage volume. It addresses the transmission pressure of massive data in high-frequency scenarios while adapting to the narrowband transmission limitations of low-bandwidth scenarios, reducing network congestion risks. The sliding window uses the last data of the previous window as the baseline value for the subsequent window, along with window start and end timestamps, to ensure the temporal correlation of segmented data. The cloud uses the baseline value to reverse-engineer the original data, splicing them together to form a complete time-series chain, avoiding data breaks or errors and ensuring the usability of monitoring data analysis. Each sliding window data segment has a verification hash value calculated. After receiving the data, the cloud verifies its integrity through hash comparison, effectively preventing tampering, loss, or damage during transmission and providing a reliable data foundation for subsequent data analysis. When synchronization is interrupted, edge nodes cache untransmitted window data packets. Upon recovery, they are sent in sequence, and the cloud accurately splices them based on timestamps, eliminating the need for full retransmission. This avoids data loss, saves bandwidth and time costs, adapts to potential link fluctuations in high-frequency / low-bandwidth scenarios, and improves the anti-interference capability of the synchronization process.

[0070] In one embodiment, encrypting the monitoring data using the spatiotemporal information involved in the data transmission process between the edge node and the sensing node includes: S502, when the sensing node collects monitoring data, it generates a spatiotemporal fingerprint containing its own unique device identifier, deployment physical location, data collection time and data content characteristics, and binds the spatiotemporal fingerprint to the collected monitoring data.

[0071] S504, the bound data is transmitted to the edge node. The edge node receives the bound data, verifies the bound data, and in response to the verification passing, the edge node synchronizes the bound data to the cloud node in the distributed monitoring network.

[0072] Spatiotemporal fingerprints typically refer to multi-dimensional unique identifiers generated when sensing nodes collect monitoring data. These identifiers include four core types of information: unique device identifier, physical deployment location, data collection time, and data content characteristics. They serve as the core basis for data traceability, tamper prevention, and legality verification, possessing the unique characteristics of one fingerprint per device, one fingerprint per time, and one fingerprint per data point. The unique device identifier is a pre-assigned unique code (such as a UUID or hardware serial number) to each sensing node, used to uniquely distinguish different sensing nodes in a distributed monitoring network and avoid confusion about data sources. The physical deployment location refers to the planned deployment coordinates of the sensing node within the target area (such as latitude and longitude, or grid location), which is key information in the spatiotemporal fingerprint marking the geographical source of data collection. The data collection time refers to the precise timestamp of the monitoring data collected by the sensing node, used to mark the time attribute of the data, supporting subsequent analysis and splicing of time-series data. Data content characteristics refer to the core feature values ​​calculated from the original monitoring data, used to characterize the uniqueness of the data content and are a key basis for verifying whether the data has been tampered with. The bound data refers to the standardized data packet formed by integrating and encapsulating spatiotemporal fingerprints with the original monitoring data. It is the basic unit for transmission and verification between sensing nodes and edge nodes.

[0073] Specifically, sensing nodes collect monitoring data (such as power output and vibration values ​​of power equipment) from the target area at a preset frequency; simultaneously, they automatically extract their own unique device identifier (such as code, serial number, etc.), deployment physical location, record the precise timestamp of data collection, and calculate content feature values ​​for the collected raw data; integrating the above four types of information, a spatiotemporal fingerprint corresponding one-to-one with the monitoring data is generated. The sensing nodes bind and encapsulate the spatiotemporal fingerprint with the raw monitoring data according to a preset data packet format, forming an integrated data packet of monitoring data + spatiotemporal fingerprint. The sensing nodes transmit the bound data packet to the edge nodes. After receiving the bound data packet, the edge nodes initiate multi-dimensional verification of the spatiotemporal fingerprint: 1. Verify whether the device's unique identifier is a legitimate node identifier registered within the network; 2. Verify whether the deployment physical location is within the preset node planning location range and whether there is location drift; 3. Verify whether the data collection timestamp is reasonable (e.g., not exceeding the current time and not conflicting with historical data times); 4. Verify whether the data content feature values ​​are consistent with the calculation results of the raw monitoring data, confirming that the data has not been tampered with. If all the above verifications pass, the edge node determines that the data is trustworthy and includes it in the synchronization queue. According to the corresponding scenario synchronization scheme (such as index table synchronization in normal scenarios and window data packet synchronization in high-frequency / low-bandwidth scenarios), the bound data packet is synchronized to the cloud node of the distributed monitoring network. If the verification fails, the data packet is directly discarded and an exception log is generated and fed back to the perception node.

[0074] In this embodiment, spatiotemporal fingerprinting integrates the device's unique identifier, physical location, collection time, and data content characteristics, deeply binding them with the monitoring data to form a unique "data-identity-spatiotemporal" correspondence. Any data tampering or forgery will cause fingerprint verification to fail, blocking ineffective and false data from flowing into the network at the source and ensuring the authenticity of the monitoring data. The device identifier, deployment location, collection time, and other information in the spatiotemporal fingerprint can accurately locate the data collection node, geographical source, and temporal attribution, achieving full-link data traceability. If data anomalies occur subsequently, they can be quickly traced back to the specific sensing node and collection scenario, facilitating problem investigation and responsibility determination. After receiving data, the edge node first performs multi-dimensional verification (legality, integrity, spatiotemporal consistency) to filter out problematic data such as illegal node access, location drift, and data tampering in advance, avoiding low-value data from consuming transmission bandwidth and cloud storage and computing resources, significantly improving the overall data processing efficiency of the network.

[0075] In one embodiment, the sensing nodes and the edge nodes form a mesh cluster. Each node within the mesh cluster autonomously discovers and establishes a neighboring network through a pre-defined self-organizing network protocol. The bound data is the data packet to be transmitted. Here, a mesh cluster refers to a wireless mesh network cluster composed of sensing nodes and edge nodes. Unlike traditional star or bus network structures, each node within the cluster can act as a data transmission relay station, possessing core characteristics of multi-path, self-organization, and self-healing, adapting to the needs of distributed monitoring networks with dispersed node deployments. A self-organizing network protocol refers to a distributed network communication protocol (such as lightweight protocols like ZigBee, Bluetooth Mesh, and LoRaWAN) pre-embedded in the nodes, allowing nodes to autonomously complete neighbor identification, link establishment, and data routing without relying on a central controller. Autonomous discovery refers to the process by which nodes within the mesh cluster, after powering on, automatically scan for other nodes within their signal coverage area using a self-organizing network protocol, identifying and recording neighboring node information (such as node identifier and signal strength). Neighboring network formation refers to the process by which, after autonomous discovery, adjacent nodes establish bidirectional communication links, thereby forming a mesh network topology with interconnected nodes and multi-path communication.

[0076] The edge node receives the bound data and verifies the bound data, including: S602, the data packet to be transmitted is forwarded in the Mesh cluster by adjacency route. Each time it passes through a forwarding node, the forwarding node verifies the spatiotemporal fingerprint in the data packet to be transmitted. S604, in response to successful verification, the forwarding node adds its own spatiotemporal fingerprint to the fingerprint sequence of the data packet to be transmitted, forming a relay-style trajectory chain. The forwarding node includes sensing nodes and edge nodes, and the fingerprint sequence contains the spatiotemporal fingerprints of multiple forwarding nodes.

[0077] In this context, adjacency routing forwarding typically refers to nodes within a mesh cluster forwarding data packets only to adjacent nodes within their signal coverage area (rather than relying on a central routing device) based on a self-organizing network protocol. This multi-hop adjacency forwarding enables data transmission from the source node to the target edge node. Forwarding nodes are nodes within the mesh cluster that participate in data packet relay transmission, including sensing nodes and edge nodes. Both types of nodes can receive data packets, perform verification, and forward them, undertaking the dual responsibilities of data transmission relay and security verification. A fingerprint sequence is a set of spatiotemporal fingerprints stored in the data packets to be transmitted, arranged in forwarding order. It includes the original spatiotemporal fingerprint of the source sensing node and the spatiotemporal fingerprints of each forwarding node, serving as the core carrier recording the entire data transmission path. A relay-style trajectory chain refers to a fingerprint sequence formed by each forwarding node sequentially adding its own spatiotemporal fingerprint, linked in the transmission order, resembling a relay baton, allowing for complete tracing of the data packet's transmission path within the mesh cluster.

[0078] Specifically, after the sensing node generates a data packet bound to the original spatiotemporal fingerprint, it selects the neighboring node with the best signal strength as the first forwarding node based on the adjacency routing algorithm of the Mesh self-organizing network protocol, and initiates the data forwarding process. Upon receiving the data packet, the first forwarding node (which can be a sensing node or an edge node) immediately initiates spatiotemporal fingerprint verification: 1. Verify whether the unique device identifier in the original spatiotemporal fingerprint belongs to a legitimate node within the Mesh cluster; 2. Verify whether the data content feature values ​​are consistent with the calculated results of the monitoring data in the data packet, confirming that the data has not been tampered with; 3. Verify whether the collection timestamp and location information are logically consistent (e.g., the time is not ahead of schedule, and the location is not outside the cluster deployment range). If all verifications pass, the forwarding node generates its own spatiotemporal fingerprint (including its own device identifier, current location, forwarding time, and data packet forwarding characteristics) and adds this fingerprint to the end of the fingerprint sequence of the data packet; if the verification fails, the forwarding node directly discards the data packet and reports an exception log indicating that the data is invalid to the source node. After the forwarding node completes fingerprint addition, it continues to select the next optimal neighbor node using the adjacency routing algorithm and forwards the data packet carrying the updated fingerprint sequence. Each subsequent forwarding node repeats the "receive-verify-add fingerprint-forward" process, allowing the fingerprint sequence of the data packet to continuously extend along the forwarding path, ultimately forming a complete relay-style trajectory chain. When the data packet arrives at the final edge node after multiple hops, the edge node can use the fingerprint sequence to fully trace the source node information of the data packet and the transmission path of all forwarding nodes, providing a reliable end-to-end basis for subsequent data verification and synchronization. The source node refers to the initial node that generates and initiates the transmission of the data packet.

[0079] S606, determine the target receiving node of the data packet to be transmitted. After receiving the data packet to be transmitted carrying the relay trajectory chain, the target receiving node performs node peer verification based on the relay trajectory chain. The node peer verification includes comparing the timestamp continuity of the spatiotemporal fingerprints of adjacent nodes and the physical correlation between the node deployment location and the transmission path. The target receiving node includes edge nodes and sensing nodes.

[0080] Among them, node peer verification refers to the full-path consistency verification performed by the target receiving node on the relay-style trajectory chain. Its core is to verify the legality, continuity, and authenticity of the data packet transmission path by comparing the spatiotemporal fingerprint information of adjacent nodes, which differs from the single-node fingerprint verification of forwarding nodes. Timestamp continuity is one of the core dimensions of node peer verification. It requires that the timestamps of the spatiotemporal fingerprints of adjacent nodes in the relay-style trajectory chain increase sequentially according to the forwarding order, and that the time difference is within the reasonable forwarding latency range preset by the Mesh cluster (e.g., milliseconds), without any time reversal, gaps, or other anomalies. The physical correlation between node deployment location and transmission path is another core dimension of node peer verification. It requires that the physical deployment locations of adjacent nodes in the trajectory chain conform to the adjacency networking rules of the Mesh cluster, that is, the location distance between adjacent nodes is within the signal coverage area, the transmission path conforms to the physical logic of adjacency forwarding, and there are no anomalies such as cross-regional skip forwarding.

[0081] Specifically, when the source node generates a data packet to be transmitted, it determines the target receiving node based on the routing algorithm of the Mesh cluster and the task requirements (such as synchronization to the cloud, data sharing between nodes), and writes the target receiving node identifier into the data packet header. During the forwarding process, the forwarding node selects the optimal adjacent path to transmit to the target receiving node based on this identifier. The target receiving node (edge ​​node or sensing node) receives the data packet to be transmitted carrying the complete relay trajectory chain, extracts the target identifier from the data packet header to confirm the match, and then initiates the node peer verification process. The target receiving node traverses the spatiotemporal fingerprint in the relay trajectory chain according to the forwarding order, and compares the timestamps of adjacent nodes one by one: verifying whether the timestamps increase sequentially in the order of "source node → first forwarding node → second forwarding node → ... → target receiving node"; verifying whether the time difference between adjacent nodes is within the preset reasonable forwarding delay; if anomalies such as timestamp reversal or excessive time difference occur, the verification is deemed to have failed, the data packet is directly discarded and the abnormal trajectory is recorded. The target receiving node retrieves the deployment locations of nodes in the Mesh cluster and, combined with the physical location information of each node in the trajectory chain, verifies the positional relationships of adjacent nodes: it checks whether the deployment locations of adjacent nodes are within each other's signal coverage range and whether they are registered adjacent nodes within the cluster; it checks whether the entire transmission path conforms to the physical logic of adjacency multi-hop, without skipping forwards across non-adjacent nodes; if a location mismatch or path abnormality occurs, the verification fails, the data packet is discarded, and a report is submitted. If both the timestamp continuity and location physical correlation verifications pass, the target receiving node determines that the data packet's transmission path is trustworthy and the data has not been tampered with: if the target receiving node is an edge node, the data packet is added to the cloud synchronization queue; if the target receiving node is a sensing node, it is directly used for local collaborative monitoring and analysis. Local collaborative monitoring and analysis refers to a localized monitoring mode that does not rely on remote computing and scheduling of cloud nodes, but rather on sensing nodes and edge nodes within the Mesh cluster to achieve data sharing and joint analysis through adjacency networking. The core is to utilize the computing power and data resources of the nodes within the cluster to complete a real-time, collaborative status assessment of the target area.

[0082] S608, if the node peer verification passes, the target receiving node receives and caches the monitoring data in the data packet to be transmitted; if the verification fails, it is determined that the data packet to be transmitted has a transmission abnormality or node illegal access risk, the target receiving node intercepts the data packet to be transmitted, and broadcasts the abnormal information to the cluster.

[0083] Among these, the risk of unauthorized node access refers to the presence of spatiotemporal fingerprints of nodes not registered in the Mesh cluster in the trajectory chain, or the drift of legitimate nodes to non-deployment areas to participate in forwarding, posing a security risk of unauthorized nodes infiltrating the cluster to steal and tamper with data. The risk of broadcasting abnormal information within the cluster refers to the operation by which the target receiving node sends standardized abnormal notifications to all sensing nodes and edge nodes within the cluster via the Mesh self-organizing network protocol. The notification content includes core information such as the unique identifier of the abnormal data packet, the fingerprint of the unauthorized node, and the type of transmission abnormality.

[0084] Specifically, after the target receiving node completes the two core verifications of the force-type trajectory chain, it generates a clear verification result: if the timestamp increases sequentially according to the forwarding order and the delay is within a reasonable range, and the positions of adjacent nodes are all within each other's signal coverage range and the transmission path conforms to the adjacency multi-hop logic, it is determined that the verification is passed; If any verification fails, the verification is deemed unsuccessful, and the exception handling process is immediately triggered. After the target receiving node confirms the trustworthiness of the data packet, it performs the following operations: extracts the monitoring data and complete trajectory chain from the data packet to be transmitted and stores it in the local cache module; performs subsequent processing according to the node type: if it is an edge node: adds the cached data to the cloud synchronization queue and completes cloud synchronization according to the corresponding scenario synchronization scheme (such as window data packet synchronization, bidirectional index table synchronization); if it is a sensing node: uses the cached data for local collaborative monitoring and analysis (such as regional data complementary verification, device linkage fault diagnosis). Once the target receiving node determines that a data packet poses a transmission anomaly or unauthorized access risk, it activates a joint defense mechanism: directly discarding the data packet to be transmitted without any storage or forwarding operations, thus blocking the flow path of the abnormal data; generating anomaly information: compiling key characteristics of the abnormal data packet (such as source node identifier, abnormal trajectory fragments, and unauthorized node fingerprints), and generating a standardized anomaly notification according to a preset format; broadcasting the anomaly information to all nodes in the Mesh cluster via the self-organizing network protocol to ensure that nodes within the cluster are synchronously aware of the risk; after receiving the anomaly information, other nodes automatically update their local protection policies, such as blacklisting unauthorized node identifiers, avoiding the forwarding path of abnormal data packets, and recording anomaly logs to provide a basis for subsequent cluster security audits.

[0085] In this embodiment, each data packet undergoes spatiotemporal fingerprint verification upon passing through a forwarding node. This directly filters out tampered, forged, or illegally sourced data during transmission, preventing the spread of abnormal data across multiple hops within the cluster, blocking risks midway through the transmission link, and improving the purity of data transmission. Each forwarding node sequentially adds its own spatiotemporal fingerprint to form a trajectory chain, completely recording the data packet's flow path from the source node to the target node. Subsequently, each forwarding node can be accurately located through the trajectory chain. If data anomalies occur, the problematic node and transmission link can be quickly traced, providing a clear basis for fault diagnosis and responsibility determination. The target receiving node, by verifying the continuity of timestamps and the physical correlation of locations, can effectively identify anomalies such as time reversal and cross-regional hop forwarding, ensuring that the data packet's transmission path fully complies with the Mesh cluster adjacency networking rules, further verifying that the data has not been tampered with or interfered with by illegal nodes after multiple hop forwarding.

[0086] In one embodiment, multiple edge nodes constitute an edge node cluster. The process of transmitting the bound data to the edge nodes, the edge nodes receiving the bound data, verifying the bound data, and, upon successful verification, synchronizing the bound data to cloud nodes in the distributed monitoring network, includes: S702, the edge node cluster receives the monitoring data bound to the spatiotemporal fingerprint, performs fragmentation processing on the monitoring data to generate several data fragments, and assigns a unique fragment identifier, fragment index and fragment verification hash value to each data fragment.

[0087] In this context, an edge node cluster refers to a distributed processing cluster composed of multiple functionally independent and interconnected edge nodes. Unlike the independent operation of a single edge node, nodes within the cluster can collaboratively complete tasks such as data reception, fragmentation processing, and storage backup. This provides stronger computing power scalability and fault tolerance, adapting to the parallel processing needs of large-scale monitoring data. Fragmentation processing refers to the operation where the edge node cluster divides the received complete monitoring data into several independent small data units according to preset rules (such as data volume thresholds, time segment lengths, and content logical blocks). Data fragmentation refers to the independent small data units generated after fragmentation processing of complete monitoring data. Each fragment is a part of the original data, and there is an ordered relationship between fragments, which can be restored to complete data through fragment indexes. A unique fragment identifier is a globally unique code assigned to each data fragment by the edge node cluster (such as a combination of fragment generation time, cluster node identifier, and random sequence code). This code is used to uniquely mark fragments within the cluster and during subsequent synchronization processes, avoiding fragment confusion or duplicate processing. A shard index is a parameter used to mark the location of a data shard within the original complete data. It includes core information such as shard number, shard start offset, and shard length, and is crucial for subsequently restoring multiple data shards into complete monitoring data. A shard verification hash value is a unique hash string calculated by the edge node cluster for the complete content of a single data shard. It is used to verify whether the data shard is complete and untampered during transmission and storage, and is a core verification indicator to ensure the trustworthiness of sharded data.

[0088] Specifically, the edge node cluster can shard data based on data type and application requirements, choosing from preset sharding rules. For example, it can shard by volume, suitable for large-capacity continuous monitoring data, setting a fixed shard size threshold (e.g., 128KB per shard) to divide the complete data into several equal-sized shards; if the last shard is less than the threshold, it is divided according to its actual length. It can also shard by time series, suitable for time-series monitoring data, splitting it at fixed time intervals (e.g., every 5 minutes as a shard), so that each shard corresponds to a continuous monitoring period, facilitating subsequent time-series analysis. Alternatively, it can shard by content logic, suitable for structured monitoring data, splitting it according to the data's logical content blocks (e.g., a complete set of equipment operating parameters), ensuring that each shard has independent content meaning. An edge node in the edge node cluster can split the complete monitoring data into several independent data shards according to the selected rules, while recording the location information of each shard in the original data, providing a basis for subsequent shard index allocation. This edge node can also assign a unique fragment identifier to each data fragment, with encoding rules ensuring no duplication within the cluster, and includes traceability information such as the fragment's owning node and generation time. Based on the fragment's position in the original data, a fragment index is assigned to each fragment, specifying the fragment sequence number, starting offset, fragment length, and total length of the original data, ensuring the ordered nature of fragment reconstruction. This edge node can also calculate a hash value for the complete content of each data fragment (excluding metadata), generating a fragment verification hash value. Subsequently, the unique fragment identifier, fragment index, and fragment verification hash value are bound to the data fragment itself, forming a standardized fragment data package containing fragmented data and metadata, completing the entire fragmentation process.

[0089] S704, the edge node cluster binds each data shard with the corresponding shard spatiotemporal fingerprint fragment to form a distributed trajectory fragment. The distributed trajectory fragment includes: a unique shard identifier, a shard index, shard data, a shard spatiotemporal fingerprint fragment, a shard verification hash value, and a complete trajectory chain association identifier.

[0090] The sharded spatiotemporal fingerprint fragment refers to a lightweight spatiotemporal identifier fragment generated by the edge node cluster for a single data shard, which is a split and mapping of the original complete spatiotemporal fingerprint. It contains three core pieces of information: first, the unique identifier and physical deployment location of the edge node device that generated the shard; second, the timestamp of the sharding process; and third, the feature values ​​of the shard's data content. The distributed trajectory fragment refers to a standardized distributed storage unit formed by binding data shards with sharded spatiotemporal fingerprint fragments and various metadata, serving as an extension of the relay trajectory chain at the sharding level. It carries all the key information from the generation to the flow of the data shard and can be distributed and transmitted within the edge node cluster, adapting to the sharding management needs of large-scale data. The complete trajectory chain association identifier refers to a globally unique association code used to associate data shards with the original complete monitoring data packet, and its value is completely consistent with the unique identifier of the relay trajectory chain of the original data packet. Through this identifier, the original data packet corresponding to a certain data shard can be quickly located, ensuring the consistency of traceability between the shard and the original data.

[0091] Specifically, for each data shard in the edge node cluster, a specific edge node generates a corresponding spatiotemporal fingerprint fragment: extracting its own unique device identifier and physical deployment location; recording the generation timestamp of the shard; calculating feature values ​​for the shard data content; and integrating these three types of information to form a spatiotemporal fingerprint fragment that corresponds one-to-one with the data shard. The relay-style trajectory chain of the original monitoring data packet is retrieved, and its globally unique identifier is extracted and directly used as the complete trajectory chain association identifier, assigned to the current data shard to ensure the shard's traceability association with the original data packet. The relay-style trajectory chain is a fully traceable sequence formed by connecting nodes in the forwarding order, including the nodes traversed during data transmission. The data shard is used as the core carrier, and the following metadata is bound sequentially: the shard's unique identifier, shard index, spatiotemporal fingerprint fragment, and the determined complete trajectory chain association identifier. Integration and encapsulation are completed according to a preset data packet format to form structured distributed trajectory fragments.

[0092] S706, the edge node cluster distributes the distributed trajectory fragments to different edge nodes within the edge node cluster; Each of the distributed trajectory fragments is backed up to at least two non-adjacent edge nodes.

[0093] Specifically, the sharding processing nodes of the edge node cluster send the current distributed trajectory fragments to the selected primary storage node and two backup storage nodes through the internal communication link of the cluster. After receiving the fragments, each target node first verifies the integrity by checking the sharding check hash value (comparing whether the hash value calculated from the received data is consistent with the hash value carried by the fragment) to avoid data corruption during transmission. After the verification is successful, the fragment is associated with the metadata and stored, and a storage confirmation feedback is generated (including fragment identifier, storage node ID, storage partition, and storage time).

[0094] S708, when a cloud node in the distributed monitoring network initiates a data acquisition request, it retrieves all the corresponding distributed trajectory fragments from the edge node cluster based on the complete trajectory chain association identifier. Through fragment index sorting, fragment verification hash value verification, and fragment spatiotemporal fingerprint segment continuity comparison, the legality verification of the trajectory fragments and the reconstruction of the complete spatiotemporal trajectory chain are completed to obtain the monitoring data.

[0095] Among them, a data acquisition request refers to an instruction initiated by a cloud node to an edge node cluster to retrieve specified monitoring data.

[0096] Specifically, cloud nodes generate data acquisition requests based on business needs (such as monitoring data analysis for a specific time period or region). These requests include a complete trajectory chain association identifier corresponding to the target monitoring data and are sent to a specific edge node in the edge node cluster, referred to as the control node. Upon receiving the request, the control node parses the complete trajectory chain association identifier. Based on this identifier, it traverses the storage ledgers of all edge nodes within the cluster, retrieving all distributed trajectory fragments bound to that identifier. Following a load balancing strategy, the control node instructs the edge nodes storing these fragments to transmit them in batches to the cloud node. After receiving all the target distributed trajectory fragments, the cloud node extracts the fragment index information from each fragment and sorts the fragments according to the rules of ascending fragment number and ascending starting offset, restoring the disordered fragments to the original fragment order of the monitoring data, forming a preliminary fragment sequence. The cloud node performs a three-layer verification on the sorted fragment sequence. If any step fails, the data is deemed invalid, and the process terminates: the hash value of each fragment's fragment data is recalculated and compared with the fragment verification hash value carried by the fragment. If all matches, the fragment data has not been tampered with; if there is a mismatch, the fragment is marked as abnormal and removed. The spatiotemporal fingerprint segments of adjacent fragments are compared in sorted order. Two points are verified: first, the processing timestamps of the fragments increase sequentially, and the time difference is within the normal sharding processing latency range of the edge cluster; second, the edge node identifiers of the fragments conform to the task allocation logic of the cluster sharding processing, and no abnormal nodes are involved. A successful comparison indicates that the fragment sequence has not been tampered with or replaced. The complete trajectory chain association identifiers of all fragments are verified to be completely consistent to avoid the mixing of fragments from other data. All fragment sequences that pass the verification are spliced ​​together according to the starting offset and length of the sharding index to restore the complete original monitoring data.

[0097] In this embodiment, the edge node cluster distributes the track fragments across different nodes, with each fragment backed up to at least two non-adjacent nodes, avoiding the single point of failure risk of traditional centralized storage. Even if some edge nodes go offline or their storage is damaged, complete fragments can still be retrieved from other backup nodes, ensuring that monitoring data is not lost. Simultaneously, the "non-adjacent node" backup rule avoids the cascading impact of localized failures (such as power outages or network interruptions) on multiple copies, further improving data storage stability. Cloud nodes can accurately retrieve all fragments corresponding to the target data based on the complete track chain association identifier, without traversing the massive amounts of data in the edge cluster, significantly reducing the amount of data exchanged between the cloud and the edge, and saving network bandwidth. Meanwhile, the cloud performs legality verification through rapid sorting using shard indexes, rapid hash value verification, and spatiotemporal fingerprint continuity comparison, eliminating the need for complex calculations and significantly improving the efficiency of data retrieval and restoration, adapting to the high-frequency data interaction needs of large-scale distributed monitoring networks.

[0098] In one embodiment, determining the node planning positions in the target area based on the multi-source node planning data and pre-set node constraints includes: S802, based on the multi-source node planning data, determine the weight coefficients of the node constraints, wherein the node constraints include at least cost constraints, coverage constraints, dynamic adaptation constraints, and efficiency constraints.

[0099] S804, Based on the topographic data and monitoring task requirement data in the multi-source node planning data of the target area, determine the node location method, which includes multi-target ant colony algorithm, hierarchical location method, and follower location method.

[0100] Among them, node constraints refer to the core limitations and optimization objectives that must be followed during the node planning process, and are the key principles for balancing node deployment effectiveness and resource investment. Cost constraints are the economic constraints that node planning must meet, covering equipment procurement costs, deployment and construction costs, and subsequent maintenance costs. The core objective is to control total investment while meeting monitoring needs. Coverage constraints are the monitoring range coverage requirements that node deployment must achieve, ensuring that no key monitoring points are missed and no blind spots exist within the target area, adapting to the spatial coverage requirements of monitoring tasks. Dynamic adaptation constraints are the environmental and task adaptability capabilities that node deployment must possess, ensuring that nodes can cope with dynamic scenarios such as changes in the target area's terrain and adjustments to monitoring tasks (such as adding monitoring points or increasing monitoring frequency). Efficiency constraints are the operational efficiency requirements that node deployment must meet, including data transmission latency, processing response speed, and energy consumption efficiency, ensuring the efficient operation of the monitoring network. Weighting coefficients are numerical indicators used to quantify the importance of different node constraints (e.g., cost weight 0.3, coverage weight 0.4), derived from comprehensive analysis of multi-source node planning data, determining the priority of each constraint in node planning. Node location methods are specific technical solutions used to determine the deployment locations of sensing nodes and edge nodes in a distributed monitoring network. The appropriate method must be selected based on the characteristics of the target area and monitoring requirements. Multi-target ant colony optimization (MCA) is an optimization algorithm based on ant colony foraging behavior, suitable for node location under multiple constraints. It can simultaneously consider multiple objectives such as cost, coverage, and efficiency, outputting the optimal deployment scheme. Hierarchical location methods select locations according to the hierarchical architecture of the monitoring network (e.g., sensing layer, edge layer). Differentiated location rules are formulated based on the functional requirements of nodes at different levels (sensing nodes focus on coverage, edge nodes focus on computing power and transmission), ensuring inter-level collaborative adaptation. Follow-up location methods are based on the topographic features of the target area or the distribution of monitoring targets. Node deployment locations follow key terrain markers (e.g., flat areas, unobstructed locations) or areas with concentrated monitoring targets, improving the targeting and adaptability of the deployment. Monitoring task requirements data are the core input supporting node location selection and constraint setting. This refers to the set of quantitative indicators, functional requirements, and execution rules directly related to a specific monitoring task. For example, it may include: monitoring objectives and types, coverage and accuracy, data timeliness and transmission, dynamic task adjustment, etc. The monitoring objectives and types dimension clarifies the core objects and execution mode of the monitoring task and is the foundation of all requirements. The coverage and accuracy dimension determines the spatial distribution of node deployment and equipment performance requirements, and is the direct source of coverage constraints. The data timeliness and transmission dimension determines the communication and processing capability requirements of the nodes, and is the core basis for efficiency constraints. The dynamic task adjustment dimension determines the flexible adaptability of the nodes and is the source of dynamic adaptation constraints.

[0101] Specifically, multi-source node planning data, including topography, monitoring task requirements, and resource allocation, are collected for the target area. The correlation between the constraints of each node (cost, coverage, dynamic adaptation, and efficiency) and the planning objectives is comprehensively analyzed, and the weight coefficients of each constraint are quantified to clarify the planning priority. Focusing on topographic data (such as terrain complexity and obstacle distribution) and monitoring task requirement data (such as coverage and real-time requirements) from the multi-source data, and combining the priority reflected by the weight coefficients, the node location method most suitable for the current scenario is selected from multi-target ant colony algorithm, hierarchical location method, and follower location method.

[0102] The Analytic Hierarchy Process (AHP) can be used to quantify weight coefficients, which can be adjusted based on the characteristics of the target area. For example, in remote mountainous monitoring scenarios, the complex terrain leads to high deployment costs, so the cost constraint weight can be appropriately increased. In urban core area environmental monitoring scenarios, it is necessary to ensure no monitoring blind spots, so the coverage constraint weight can be appropriately increased. Combining topographic data and monitoring task requirements data, the corresponding node location method is matched according to the priority of the weight coefficients. For areas with balanced multiple constraints (where the differences between weight coefficients are not large), the multi-target ant colony algorithm can be used. For areas with multi-level network architectures (such as the central control room of a factory with dense equipment), the hierarchical location method can be used. For areas with obvious terrain features or concentrated monitoring targets, the follow-up location method is suitable (where topographic data accounts for a large proportion of the weight coefficients). In addition, it should be noted that multiple node location methods can be combined. For example, in complex mountainous monitoring scenarios, the multi-target ant colony algorithm + hierarchical location method can be used. First, the global optimal location is determined by the ant colony algorithm, and then the location is refined layer by layer according to the perception layer and the edge layer.

[0103] In some exemplary embodiments, high-precision terrain data (such as DEM digital elevation models, satellite imagery, and obstacle distribution ledgers) of the target area can be collected; the area can be divided into fixed-resolution grid units (such as 10m×10m), and each grid can be labeled with attributes: selectable points (flat terrain, unobstructed, easy to deploy), non-selectable points (areas where deployment is impossible due to steep slopes, water bodies, buildings, etc.), and preferred points (high ground, areas with open views, suitable for deployment of sensing nodes); a terrain constraint layer can be generated to clarify the physical boundaries of node deployment. Core task indicators are extracted: monitoring type (environment / equipment / security), coverage boundaries, monitoring accuracy thresholds, data transmission latency requirements, and task adjustment frequency (dynamic adaptation requirements); task indicators are quantified: for example, "full coverage without blind spots" is converted into a constraint threshold of coverage ≥95%, and "real-time monitoring" is converted into an efficiency threshold of data transmission latency ≤500ms.

[0104] Based on the preprocessed terrain features and task requirements, the following rules are used to match the corresponding site selection methods: If the terrain is complex (mountains / obstacles) and lacks obvious linear features with multiple constraints (meeting cost, coverage, and efficiency requirements simultaneously), a multi-target ant colony algorithm can be used. If the area is large and requires hierarchical deployment; and the sensing layer collects data while the edge layer processes it, with differentiated functions at each level (sensing layer prioritizes coverage, edge layer prioritizes efficiency), a hierarchical site selection method can be used. If the terrain is linearly distributed (rivers / highways) or the monitoring targets are concentrated (equipment clusters / farmland areas) and the task objective is singular and clear (e.g., monitoring along a route, full coverage of the target area), a follow-up site selection method can be selected.

[0105] S806, Based on the determined node location method and the weight coefficients of the node constraint conditions, the initial node planning positions in the target area are calculated; S808, based on the node constraints, perform a compliance check on the initial node planning position, and determine the final node planning position in the target area after the check passes.

[0106] The initial node planning locations refer to the set of candidate node deployment locations obtained through algorithmic calculation or methodological deduction based on the selected node location method and node constraint weight coefficients. This is a preliminary plan that has not been practically verified against the constraints, and includes core information such as node coordinates, number, and hierarchical division. The compliance verification refers to the quantitative verification process of substituting the initial node planning locations into all preset node constraints to verify whether each constraint indicator requirement is met. Its core is to determine whether the preliminary plan balances economy, coverage, adaptability, and efficiency, and it is a crucial step in selecting feasible planning schemes.

[0107] Specifically, if a multi-objective ant colony algorithm is used, the target area can be rasterized, and each raster can be set as a candidate node position. The weight coefficients of the four types of constraints are substituted into the algorithm objective function to clarify the optimization priority of each constraint (e.g., coverage weight 0.35 > cost weight 0.25 > efficiency weight 0.2 > dynamic adaptation weight 0.2). The ant colony algorithm iterative calculation is started: the ant colony's "foraging" path is simulated, and the "comprehensive compliance with the constraints under the weight priority" is used as the basis for pheromone updates. The algorithm gradually converges to obtain the globally optimal set of candidate node positions, i.e., the initial node planning positions, which include the coordinates, number, and hierarchical distribution of sensing nodes and edge nodes.

[0108] If a hierarchical site selection method is adopted, the initial location of the sensing layer is calculated first: with coverage constraints as the core (thresholds are set according to weight coefficients), obstacles are avoided by combining terrain data, and points with unobstructed signals and wide monitoring fields are selected to generate the initial location of the sensing nodes; then the initial location of the edge layer is calculated: with efficiency constraints as the core (latency thresholds are set according to weight coefficients), areas close to the sensing node cluster and with sufficient computing resources are selected, while controlling the deployment cost within the constraint thresholds to generate the initial location of the edge nodes; the locations of the two layers of nodes are integrated to form a hierarchical set of initial node planning locations.

[0109] If the follow-up site selection method is adopted, candidate node locations can be generated along the target line (such as transmission line corridors, distribution network trunk lines, and substation incoming and outgoing line channels) according to a preset spacing (combined with coverage constraints, such as setting the spacing according to the coverage radius of insulator monitoring for transmission line monitoring and the signal transmission range of cable joints for distribution network monitoring). For power monitoring scenarios, weight coefficients are substituted to prioritize meeting core constraints (such as prioritizing coverage constraints in transmission line icing monitoring scenarios to ensure that there are no monitoring blind spots in the entire area of ​​towers and conductors; prioritizing dynamic adaptation constraints in urban distribution network cluster monitoring scenarios to reserve node expansion space for new ring network cabinets and charging piles), and the initial node planning locations are obtained by screening.

[0110] For the initial node planning location, quantitative verification is performed sequentially according to the weight and priority of the constraints. If the initial node planning location meets the threshold requirements of all constraints, it is directly determined as the final node planning location. For constraints that do not meet the standards (such as insufficient coverage or cost overrun), the number of nodes or coordinates of the initial location are adjusted (such as adding sensing nodes to fill blind spots or changing the deployment area of ​​edge nodes to reduce costs). A new initial solution is recalculated and generated, and the above verification process is repeated until the solution passes the verification of all constraints.

[0111] In this embodiment, initial node planning is based on quantified constraint weights and terrain-adaptive site selection methods, avoiding traditional empiricism. A multi-objective ant colony algorithm optimizes objectives such as cost and coverage, while a hierarchical site selection method matches different functional requirements, ensuring the solution fits the terrain and task, reducing adjustment costs. Cost, coverage, dynamic adaptation, and efficiency are used as verification standards to comprehensively validate the solution, avoiding overlooking any aspect. Coverage verification prevents blind spots, cost verification controls investment, and dynamic adaptation verification responds to changes, ensuring the solution meets functional standards, is economically controllable, and is usable in the long term. Adjustable priority weight coefficients, such as prioritizing coverage in urban areas and cost in mountainous areas, maximize monitoring effectiveness and minimize costs, avoiding waste.

[0112] In one embodiment, the multi-source node planning data includes at least topographic data of the target area, monitoring task requirement data, node deployment cost data, and monitoring coverage requirement data. The node constraints include at least cost constraints, coverage constraints, dynamic adaptation constraints, and efficiency constraints. Determining the weight coefficients of the node constraints based on the multi-source node planning data includes: S902, determine the mapping relationship between the multi-source node planning data and the node constraints; S904, Based on the mapping relationship, extract the quantitative indicators corresponding to the node constraints and perform normalization processing.

[0113] The mapping relationship refers to the precise correspondence between multi-source node planning data and node constraints, used to clarify which multi-source data supports which type of node constraint, the support method, and the strength of the correlation. For example, obstacle distribution in topographic data is strongly correlated with coverage constraints, and equipment purchase price in resource allocation data is strongly correlated with cost constraints; these are the core basis for subsequent extraction of quantitative indicators. Quantitative indicators are concrete numerical parameters that transform abstract node constraints into calculable and comparable values. Each node constraint corresponds to at least one quantitative indicator, and the indicator value must be determined based on the mapped multi-source node planning data; it is the numerical standard for measuring whether the constraint is satisfied. Normalization is a mathematical processing method that transforms quantitative indicators of different dimensions and orders of magnitude into dimensionless values ​​within a unified interval (usually [0, 1]). Its purpose is to eliminate the interference of dimensional differences on subsequent weight calculations and multi-objective optimization, ensuring that indicators of different constraints are comparable and additive.

[0114] Specifically, the multi-source node planning data is classified and constraint decomposed. Based on the results of data classification and constraint decomposition, a mapping table between multi-source data and node constraints is established. For example, topographic data can be associated with coverage constraints and efficiency constraints. For each type of node constraint, the indicator definition, calculation method, and threshold range are determined based on the mapped multi-source data.

[0115] For example: Extraction of quantitative indicators for cost constraints: The core quantitative indicators are: total life cycle cost (C_total) and cost per unit monitoring area (C_unit). Calculation method: ; in: (Equipment procurement cost) = Unit price of equipment × Quantity of nodes; (Construction cost) = Construction cost coefficient × Terrain complexity × Number of nodes; (Operation and maintenance cost) = Annual operation and maintenance fee × Operation and maintenance cycle; ; in: The threshold range for the target monitoring area can be set based on the project budget, such as... .

[0116] Extraction of quantitative indicators for coverage constraint: The core quantitative indicator is: monitoring coverage ( Key point coverage ( ); Calculation method: ; in: The effective coverage area of ​​the monitoring range of all nodes (excluding areas obstructed by obstacles). ; in: The number of key monitoring points covered. This represents the total number of key locations. Threshold range: set based on task requirements, such as... .

[0117] Extraction of quantitative indicators for efficiency constraints: The core quantitative indicator is: average data transmission latency ( ), edge node computing power load rate ( ) Calculation method: ; in: Data transmission latency (strongly correlated with node spacing and link bandwidth); Address latency issues at edge nodes; The total number of nodes;

[0118] in: This represents the actual computing power consumed by the edge nodes. This represents the maximum computing power limit for a node. Threshold range: set based on the real-time requirements of the task, such as... , .

[0119] Extraction of quantitative indicators for dynamic adaptation constraints: The core quantitative indicator is: node expansion margin ( ), terrain change adaptability ( ) Calculation method: ; in: The number of nodes that can be added in advance; This represents the initial number of nodes deployed. ; in: This represents the annual frequency of topographic variation. The influence coefficient of terrain change on node monitoring; Threshold range: set based on environmental stability, such as... , .

[0120] The calculated core quantitative indicators are normalized. Generally, these indicators can be divided into two categories: benefit indicators and cost indicators. Benefit indicators are generally better the larger they are, while cost indicators are generally better the smaller they are. Therefore, the Min-Max normalization method can be used to convert all quantitative indicators into dimensionless values ​​in the interval [0, 1].

[0121] Benefit-oriented indicators (the higher the indicator value, the better, such as coverage rate, expansion margin): ; Cost-related metrics (lower values ​​are better, such as transmission latency and total lifecycle cost): ; in: The original value of the indicator; The minimum value of the indicator; The maximum value of the indicator; This is the normalized value.

[0122] S906, the objective weights of the node constraints are calculated based on the normalized quantitative indicators using the entropy weight method.

[0123] Among them, the entropy weight method is typically a purely data-driven objective weighting method. It measures the dispersion and information content of an indicator by calculating the information entropy value of a normalized quantified indicator. The higher the dispersion and the greater the information content, the higher the corresponding entropy weight (objective weight), and vice versa. This method involves no subjective human intervention; the weighting result is entirely determined by the characteristics of the indicator data. The objective weight is the node constraint weight calculated based on the entropy weight method. It reflects the importance of each constraint quantified indicator at the data level. It is a weight value determined solely by the dispersion and information content of the indicator, independent of human experience, providing an objective benchmark for subsequent comprehensive weight optimization.

[0124] Specifically, the normalized quantitative index matrix is ​​as follows: ,in: : Number of node planning samples in the target area (e.g., samples of different node layout schemes); These correspond to four types of node constraints: cost, coverage, efficiency, and dynamic adaptation. : No. In the planning sample, the first Normalized value of quantitative index for class constraints (range of values) ).

[0125] To eliminate the influence of the indicator dimensions (already normalized) and the sample size, we first calculate the indicator weight of each sample under the corresponding constraints, using the following formula: ; in Minimum value (e.g.) ), used to avoid due to This renders subsequent logarithmic calculations meaningless.

[0126] Information entropy is a core indicator for measuring the amount of information in a metric. The smaller the entropy value, the higher the dispersion of the metric, the stronger its ability to differentiate planning schemes, and the greater the amount of information. The formula is: ; in The entropy normalization coefficient ensures the entropy value is normalized. .

[0127] Calculate the first Difference coefficient of class constraints The coefficient of difference reflects the distinguishing power of an indicator and is negatively correlated with the entropy value. The smaller the entropy value, the larger the coefficient of difference, and the more significant the influence of the indicator on the planning scheme. The formula is:

[0128] Calculate the first Objective weights of class constraints The difference coefficients are normalized to obtain the objective weights of each constraint, ensuring that the sum of all objective weights is 1. The formula is as follows: ; Result characteristics: If the quantitative index of a certain constraint varies greatly in different planning samples (e.g., the coverage index fluctuates from 60% to 98% in different site layout schemes), then big, The high value indicates that this constraint is more important for node planning at the data level.

[0129] Example results: If the objective weights of the four types of constraints are calculated as follows: , , , This indicates that the quantitative indicator of coverage has the highest degree of dispersion and has the greatest impact on node planning at the data level.

[0130] S908 uses the analytic hierarchy process (AHP) to construct a judgment matrix by combining the topographic data of the target area, the monitoring task requirements data, the node deployment cost data, and the monitoring coverage requirement data, and obtains the subjective correction weights.

[0131] The Analytic Hierarchy Process (AHP) is a subjective weighting method that breaks down complex multi-objective decision-making problems into hierarchical parts. It combines expert experience with real-world scenario data, quantifying the importance of each factor through pairwise comparisons. Its core is transforming qualitative judgments into quantitative weights, adapting to the personalized weight settings required in node planning that incorporate regional characteristics. The judgment matrix is ​​the core component of AHP. It's a square matrix constructed by comparing the pairwise importance of node constraints at the same level with multi-source data from the target region (e.g., a 4×4 judgment matrix for 4 constraints). Matrix elements represent the importance ratio of one constraint to another, forming the basis for calculating subjectively corrected weights. Subjectively corrected weights are calculated using AHP, combining actual data such as the target region's topography, monitoring task requirements, deployment costs, and coverage requirements to construct the judgment matrix. This weight incorporates the personalized needs of the regional scenario and expert experience, used to modify the objective weights obtained by the entropy weight method in a scenario-based manner, ensuring that the final weights both align with data characteristics and match actual node planning needs.

[0132] Specifically, combining target area data and expert experience, qualitative scenario requirements are transformed into quantitative weights. The first step defines a three-layer structure: the target layer determines subjectively adjusted weights; the criteria layer comprises four types of constraints: cost, coverage, efficiency, and dynamic adaptation; and the solution layer uses multi-source node planning data, including topography and monitoring task requirements. The second step constructs a judgment matrix. Constraint priorities are first determined based on regional characteristics (e.g., coverage in urban core areas, cost in remote mountainous areas). Constraints are then compared pairwise using a 1-9 scale (1 = equally important, 5 = significantly important, etc.) to form the judgment matrix. The third step performs a consistency check. This involves simplifying the calculation matrix, calculating consistency indicators, and comparing them with empirical values. If CR < 0.1, the matrix is ​​valid; otherwise, the values ​​are adjusted. Finally, the matrix processing results are normalized to obtain subjectively adjusted weights that fit the regional scenario, ensuring that the weights match the personalized needs of node planning.

[0133] S910, Based on a preset weighted fusion coefficient, the objective weights and the subjective correction weights are fused to obtain combined weights; S912, normalize and verify the combined weights to determine the weight coefficients of the node constraints.

[0134] Specifically, based on the project requirements of the distributed monitoring network node planning (such as data reliability priority and scenario adaptability requirements), a weighted fusion coefficient is preset (for example, the objective weight ratio coefficient is set as α, and the subjective correction weight ratio coefficient is 1-α, where α typically ranges from 0.4 to 0.6, and can be set according to actual conditions; in some embodiments of this disclosure, the weighted fusion coefficient is not absolutely limited). For the four types of node constraints—cost, coverage, efficiency, and dynamic adaptation—fusion calculations are performed separately. The preset α is multiplied by the objective weight of the corresponding constraint, and then (1-α) is multiplied by the subjective correction weight of that constraint. The two results are added together to obtain the initial combined weight of each constraint. First, the sum of the initial combined weights of all constraints is calculated, and then the initial combined weight of each constraint is divided by this sum to complete the normalization process. Subsequently, the processed weights are verified to ensure that "all weights are in the range [0,1] and the sum is 1," ensuring the standardization and rationality of the weights. If the normalization verification passes, the verified combined weights are directly used as the final weight coefficients of the node constraints; if it fails (e.g., the sum deviates too much from 1), the preset weighted fusion coefficients are readjusted, and the above fusion and verification steps are repeated until the final weight coefficients that meet the requirements are obtained.

[0135] In this embodiment, by establishing a mapping relationship between multi-source node planning data and constraints, the correlation between constraints and planning data is clarified, avoiding the problem of setting quantitative indicators out of thin air or being out of touch with actual needs. All quantitative indicators are derived from real data, ensuring that the input data for weight calculation is relevant and traceable. Abstract constraints such as cost and coverage are transformed into calculable quantitative indicators such as life cycle cost and monitoring coverage, solving the pain point of constraints being difficult to quantify and evaluate; at the same time, normalization processing eliminates the dimensional differences between different indicators, unifying all indicators to the [0,1] interval, making multi-dimensional constraints that were originally not directly comparable comparable and additive.

[0136] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0137] Based on the same inventive concept, this disclosure also provides a distributed monitoring network architecture deployment apparatus for implementing the distributed monitoring network architecture deployment method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more embodiments of the distributed monitoring network architecture deployment apparatus provided below can be found in the limitations of the distributed monitoring network architecture deployment method described above, and will not be repeated here.

[0138] In one embodiment, such as Figure 3 As shown, a distributed monitoring network architecture deployment device 1000 is provided, including: a location determination module 1002, a node planning module 1004, a breakpoint synchronization module 1006, and an encryption module 1008, wherein: The location determination module 1002 is used to acquire multi-source node planning data associated with the target area, and determine the node planning location in the target area based on the multi-source node planning data and pre-set node constraints. The node planning module 1004 is used to determine distributed nodes based on the planned node locations. The distributed nodes include edge nodes and sensing nodes. The sensing nodes are used to collect monitoring data and transmit it to the edge nodes. The edge nodes are used for inference and data caching. The breakpoint synchronization module 1006 is used to deploy a data inference model in the edge node and determine a verification-based breakpoint synchronization scheme for the edge node based on the data acquisition scenario. The data acquisition scenario includes a high-frequency data acquisition scenario and a low-bandwidth data acquisition scenario. The data inference model is used to verify the data acquired by the sensing node. The encryption module 1008 is used to encrypt the monitoring data using the spatiotemporal information involved in the data transmission process of the edge node and the sensing node.

[0139] In one embodiment of the device, the breakpoint synchronization module 1006 includes: A data table synchronization module is used to respond to data acquisition scenarios other than high-frequency data acquisition scenarios and / or low-bandwidth data acquisition scenarios. The edge node calculates a content fingerprint based on the received monitoring data, and merges the monitoring data based on the content fingerprint and a preset similarity threshold to generate at least one first data block. The content fingerprint is used to uniquely represent the content characteristics of the corresponding monitoring data. The edge node assigns a unique identifier to each generated first data block and generates a bidirectional index table based on the unique identifier, the content fingerprint of the first data block, the data start and end times, the data block size, the synchronization status, and the check hash value. The bidirectional index table is stored in both the edge node and the cloud node of the distributed monitoring network. After generating the first data block, the edge node pushes the newly added entries of the bidirectional index table to the cloud node. The cloud node compares the local bidirectional index table... If no table entry with the same unique identifier exists, a new table entry corresponding to the unique identifier is added, and the synchronization status is marked as "synchronizing". After receiving the first data block, the cloud node performs integrity verification using the content fingerprint and check hash value. In response to the successful integrity verification, the synchronization status is updated to "synchronized", and the status change information is sent back to the edge node. The edge node receives the status change information and synchronously updates the synchronization status of the corresponding table entry in its local bidirectional index table. In response to the interruption of data synchronization between the edge node and the cloud node in the distributed monitoring network, the edge node caches the data in the bidirectional index table corresponding to the newly added first data block locally and changes the synchronization status to "not synchronized". In response to the resumption of data synchronization between the edge node and the cloud node in the distributed monitoring network, the unsynchronized first data block is synchronized based on the synchronization status.

[0140] In one embodiment of the device, the breakpoint synchronization module 1006 further includes: A window synchronization module is used to respond to situations where the data acquisition scenario is a high-frequency data acquisition scenario and / or a low-bandwidth data acquisition scenario. The edge node determines an initial reference value based on the received monitoring data, and calculates the difference between the initial reference value and each monitoring data point, generating an incremental difference sequence composed of the initial reference value and the difference. The edge node divides the incremental difference sequence into several consecutive sliding window data segments based on a preset sliding window length, and calculates a check hash value for each sliding window data segment. The window reference value of the first sliding window data segment is the initial reference value. In two adjacent sliding window data segments, the window reference value of the latter sliding window data segment is the last value of the monitoring data corresponding to the former sliding window data segment. The window baseline value, window start and end timestamps, and check hash value are encapsulated to obtain a window data packet. The edge node transmits the window data packet to the cloud node of the distributed monitoring network. After receiving the window data packet, the cloud node reverses the process based on the window baseline value and the incremental differential sequence to obtain the monitoring data, and verifies the integrity of the window data segment using the check hash value. In response to the interruption of data synchronization between the edge node and the cloud node in the distributed monitoring network, the edge node buffers the untransmitted window data packet. In response to the recovery of data synchronization between the edge node and the cloud node in the distributed monitoring network from the interruption, the edge node sends the buffered window data packets to the cloud node sequentially. The cloud node receives and splices the window data packets based on the window start and end timestamps.

[0141] In one embodiment of the device, the encryption module 1008 includes: The data binding module is used to generate a spatiotemporal fingerprint containing the unique identifier of the sensing node, the physical location of the deployment, the data collection time and the characteristics of the data content when the sensing node collects monitoring data, and bind the spatiotemporal fingerprint to the collected monitoring data.

[0142] The data transmission module is used to transmit the bound data to the edge node. The edge node receives the bound data, verifies the bound data, and in response to the verification passing, the edge node synchronizes the bound data to the cloud node in the distributed monitoring network.

[0143] In one embodiment of the device, the sensing nodes and the edge nodes form a mesh cluster. Each node within the mesh cluster autonomously discovers and forms an adjacency network using a preset self-organizing network protocol. The bound data is a data packet to be transmitted. The data transmission module is further configured to forward the data packet to be transmitted within the mesh cluster using adjacency routing. Each time the data packet passes through a forwarding node, the forwarding node verifies the spatiotemporal fingerprint in the data packet to be transmitted. In response to successful verification, the forwarding node adds its own spatiotemporal fingerprint to the fingerprint sequence of the data packet to be transmitted, forming a relay-style trajectory chain. The forwarding nodes include sensing nodes and edge nodes, and the fingerprint sequence contains multiple forwarding nodes. The spatiotemporal fingerprint of a point is used to determine the target receiving node of the data packet to be transmitted. After receiving the data packet carrying the relay trajectory chain, the target receiving node performs node peer verification on the relay trajectory chain. The node peer verification includes comparing the timestamp continuity of the spatiotemporal fingerprints of adjacent nodes and the physical correlation between the node deployment location and the transmission path. The target receiving node includes edge nodes and sensing nodes. If the node peer verification passes, the target receiving node receives and caches the monitoring data in the data packet to be transmitted. If the verification fails, it is determined that the data packet to be transmitted has a transmission anomaly or node illegal access risk. The target receiving node intercepts the data packet to be transmitted and broadcasts the anomaly information to the cluster.

[0144] In one embodiment of the device, multiple edge nodes constitute an edge node cluster. The data transmission module is further configured to enable the edge node cluster to receive monitoring data bound to the spatiotemporal fingerprint, fragment the monitoring data to generate several data fragments, and assign a unique fragment identifier, fragment index, and fragment verification hash value to each data fragment. The edge node cluster binds each data fragment to a corresponding fragment spatiotemporal fingerprint fragment to form distributed trajectory fragments. The distributed trajectory fragments include: a unique fragment identifier, a fragment index, fragment data, fragment spatiotemporal fingerprint fragment, and fragment verification hash value. The edge node cluster distributes the distributed trajectory fragments to different edge nodes within the cluster, with each distributed trajectory fragment backed up to at least two non-adjacent edge nodes. When a cloud node in the distributed monitoring network initiates a data acquisition request, it retrieves all corresponding distributed trajectory fragments from the edge node cluster based on the complete trajectory chain association identifier. Through fragment index sorting, fragment verification hash value verification, and fragment spatiotemporal fingerprint segment continuity comparison, the legality verification of the trajectory fragments and the reconstruction of the complete spatiotemporal trajectory chain are completed to obtain monitoring data.

[0145] In one embodiment of the device, the position determination module 1002 includes: The weight coefficient determination module is used to determine the weight coefficients of node constraints based on the multi-source node planning data. The node constraints include at least cost constraints, coverage constraints, dynamic adaptation constraints, and efficiency constraints.

[0146] The site selection method determination module is used to determine the node site selection method based on the topographic data and monitoring task requirement data in the multi-source node planning data of the target area. The node site selection method includes multi-target ant colony algorithm, hierarchical site selection method, and follower-type site selection method.

[0147] The initial location determination module is used to calculate the initial planned node locations in the target area based on the determined node location method and the weight coefficients of the node constraint conditions.

[0148] The verification module is used to perform compliance verification on the initial node planning position based on the node constraints, and determine the final node planning position in the target area after the verification is passed.

[0149] In one embodiment of the device, the multi-source node planning data includes at least the topographic data of the target area, monitoring task requirement data, node deployment cost data, and monitoring coverage requirement data. The node constraints include at least cost constraints, coverage constraints, dynamic adaptation constraints, and efficiency constraints. The weight coefficient determination module is further used to determine the mapping relationship between the multi-source node planning data and the node constraints; based on the mapping relationship, extract the quantitative indicators corresponding to the node constraints and perform normalization processing; calculate the objective weight of the node constraints based on the normalized quantitative indicators using the entropy weight method; construct a judgment matrix using the analytic hierarchy process (AHP) combined with the topographic data of the target area, monitoring task requirement data, node deployment cost data, and monitoring coverage requirement data to obtain subjective correction weights; fuse the objective weights and the subjective correction weights based on a preset weighted fusion coefficient to obtain combined weights; and perform normalization verification on the combined weights to determine the weight coefficients of the node constraints.

[0150] Each module in the aforementioned distributed monitoring network architecture deployment device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0151] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores node planning data, monitoring data, etc. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements a distributed monitoring network architecture deployment method.

[0152] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the computer device to which the present disclosure is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0153] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0154] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0155] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0156] It should be noted that all data involved in this disclosure (including but not limited to data used for analysis, stored data, and displayed data) are information and data that have been authorized by the user or fully authorized by all parties.

[0157] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this disclosure can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this disclosure may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this disclosure may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0158] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0159] The embodiments described above are merely illustrative of several implementations of this disclosure, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent disclosure. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this disclosure, and these all fall within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the appended claims.

Claims

1. A method for deploying a distributed monitoring network architecture, characterized in that, The method includes: Obtain multi-source node planning data associated with the target area, and determine the node planning positions in the target area based on the multi-source node planning data and pre-set node constraints. Based on the planned locations of the nodes, distributed nodes are determined. The distributed nodes include edge nodes and sensing nodes. The sensing nodes are used to collect monitoring data and transmit it to the edge nodes. The edge nodes are used for inference and data caching. A data inference model is deployed in the edge node, and a verification-based breakpoint synchronization scheme for the edge node is determined based on the data acquisition scenario. The data acquisition scenario includes high-frequency data acquisition scenario and low-bandwidth data acquisition scenario. The data inference model is used to verify the data acquired by the sensing node. The monitoring data is encrypted using the spatiotemporal information involved in the data transmission process of the edge nodes and the sensing nodes.

2. The method according to claim 1, characterized in that, The verification-based breakpoint synchronization scheme for determining the edge nodes based on the data acquisition scenario includes: In response to the data acquisition scenario being a data acquisition scenario other than a high-frequency data acquisition scenario and / or a low-bandwidth data acquisition scenario, the edge node calculates a content fingerprint based on the received monitoring data, and merges the monitoring data based on the content fingerprint and a preset similarity threshold to generate at least one first data block. The content fingerprint is used to uniquely characterize the content features of the corresponding monitoring data. The edge node assigns a unique identifier to each generated first data block, and generates a bidirectional index table based on the unique identifier, the content fingerprint of the first data block, the data start and end time, the data block size, the synchronization status and the check hash value. The bidirectional index table is stored in the edge node and the cloud node of the distributed monitoring network, respectively. After the edge node generates the first data block, it pushes the new entry of the bidirectional index table to the cloud node. The cloud node compares the local bidirectional index table. If there is no entry with the same unique identifier, it adds an entry corresponding to the unique identifier and marks the synchronization status as synchronizing. The new entry matches the generated first data block. After receiving the first data block, the cloud node performs integrity verification using the content fingerprint and the verification hash value. In response to the integrity verification being successful, it updates the synchronization status to synchronized and sends the status change information back to the edge node. The edge node receives the status change information and synchronously updates the synchronization status of the corresponding entry in its local bidirectional index table. In response to the interruption of data synchronization between the edge node and the cloud node in the distributed monitoring network, the edge node caches the data in the bidirectional index table corresponding to the newly added first data block locally and changes the synchronization status to unsynchronized; In response to the recovery of data synchronization between the edge nodes and cloud nodes in the distributed monitoring network from interruption, the unsynchronized first data block is synchronized based on the synchronization status.

3. The method according to claim 1, characterized in that, The verification-based breakpoint synchronization scheme for determining the edge nodes based on the data acquisition scenario includes: In response to the data acquisition scenario being a high-frequency data acquisition scenario and / or a low-bandwidth data acquisition scenario, the edge node determines an initial reference value based on the received monitoring data, and calculates the difference based on the initial reference value and each monitoring data to generate an incremental difference sequence composed of the initial reference value and the difference. The edge node divides the incremental differential sequence into several consecutive sliding window data segments based on a preset sliding window length, and calculates a verification hash value for each sliding window data segment; wherein, the window reference value of the first sliding window data segment is the initial reference value; in two adjacent sliding window data segments, the window reference value of the latter sliding window data segment is the last value of the monitoring data corresponding to the former sliding window data segment; The window data packet is obtained by encapsulating the window baseline value, window start and end timestamps, and check hash value. The edge node transmits the window data packet to the cloud node of the distributed monitoring network; after receiving the window data packet, the cloud node reverses the process based on the window baseline value and the incremental differential sequence to obtain the monitoring data, and verifies the integrity of the window data segment through the verification hash value. In response to an interruption in data synchronization between the edge node and the cloud node in the distributed monitoring network, the edge node caches untransmitted window data packets; In response to the recovery of data synchronization between the edge node and the cloud node in the distributed monitoring network from the interruption, the edge node sequentially sends the cached window data packets to the cloud node, and the cloud node receives and splices the window data packets based on the window start and end timestamps.

4. The method according to claim 1, characterized in that, The encryption of monitoring data using the spatiotemporal information involved in data transmission between the edge nodes and the sensing nodes includes: When the sensing node collects monitoring data, it generates a spatiotemporal fingerprint containing its own unique device identifier, deployment physical location, data collection time, and data content characteristics, and binds the spatiotemporal fingerprint to the collected monitoring data. The bound data is transmitted to the edge node. The edge node receives the bound data, verifies the bound data, and if the verification is successful, the edge node synchronizes the bound data to the cloud node in the distributed monitoring network.

5. The method according to claim 4, characterized in that, The sensing nodes and the edge nodes form a mesh cluster. Each node in the mesh cluster autonomously discovers and forms a neighboring network through a preset self-organizing network protocol. The bound data is a data packet to be transmitted. The bound data is transmitted to the edge node, and the edge node receives the bound data and verifies it, including: The data packets to be transmitted are forwarded in the Mesh cluster by adjacency routing. Each time a forwarding node passes through a forwarding node, the forwarding node verifies the spatiotemporal fingerprint in the data packets to be transmitted. In response to successful verification, the forwarding node adds its own spatiotemporal fingerprint to the fingerprint sequence of the data packet to be transmitted, forming a relay-style trajectory chain. The forwarding node includes sensing nodes and edge nodes, and the fingerprint sequence contains the spatiotemporal fingerprints of multiple forwarding nodes. The target receiving node of the data packet to be transmitted is determined. After the target receiving node receives the data packet to be transmitted carrying the relay trajectory chain, it performs node peer verification on the relay trajectory chain. The node peer verification includes comparing the timestamp continuity of the spatiotemporal fingerprints of adjacent nodes and the physical correlation between the node deployment location and the transmission path. The target receiving node includes edge nodes and sensing nodes. If the node peer verification passes, the target receiving node receives and caches the monitoring data in the data packet to be transmitted; if the verification fails, it is determined that the data packet to be transmitted has a transmission abnormality or node illegal access risk, the target receiving node intercepts the data packet to be transmitted, and broadcasts the abnormal information to the cluster.

6. The method according to claim 4, characterized in that, Multiple edge nodes constitute an edge node cluster. The bound data is transmitted to the edge nodes, and each edge node receives and verifies the bound data. Upon successful verification, the edge node synchronizes the bound data to cloud nodes in the distributed monitoring network. This process includes: The edge node cluster receives monitoring data bound to the spatiotemporal fingerprint, performs fragmentation processing on the monitoring data to generate several data fragments, and assigns a unique fragment identifier, fragment index and fragment verification hash value to each data fragment. The edge node cluster binds each data shard with its corresponding spatiotemporal fingerprint fragment to form a distributed trajectory fragment. The distributed trajectory fragment includes: a unique shard identifier, a shard index, shard data, a spatiotemporal fingerprint fragment, a shard verification hash value, and a complete trajectory chain association identifier. The edge node cluster stores the distributed trajectory fragments in a distributed manner to different edge nodes within the edge node cluster, wherein each distributed trajectory fragment is backed up to at least two non-adjacent edge nodes; When a cloud node in a distributed monitoring network initiates a data acquisition request, it retrieves all the corresponding distributed trajectory fragments from the edge node cluster based on the complete trajectory chain association identifier. Through fragment index sorting, fragment verification hash value verification, and fragment spatiotemporal fingerprint segment continuity comparison, the legality verification of the trajectory fragments and the reconstruction of the complete spatiotemporal trajectory chain are completed to obtain the monitoring data.

7. The method according to claim 1, characterized in that, The step of determining the node planning positions in the target area based on the multi-source node planning data and pre-set node constraints includes: Based on the multi-source node planning data, the weight coefficients of the node constraints are determined. The node constraints include at least cost constraints, coverage constraints, dynamic adaptation constraints, and efficiency constraints. Based on the topographic data and monitoring task requirement data in the multi-source node planning data of the target area, a node site selection method is determined. The node site selection method includes multi-target ant colony algorithm, hierarchical site selection method, and follow-up site selection method. Based on the determined node location method and the weight coefficients of the node constraints, the initial node planning positions in the target area are calculated. Based on the node constraints, the initial planned node positions are verified for compliance. Once the verification is passed, the final planned node positions in the target area are determined.

8. The method according to claim 7, characterized in that, The multi-source node planning data includes at least the topographic data of the target area, monitoring task requirement data, node deployment cost data, and monitoring coverage requirement data. The node constraints include at least cost constraints, coverage constraints, dynamic adaptation constraints, and efficiency constraints. Determining the weight coefficients of the node constraints based on the multi-source node planning data includes: Determine the mapping relationship between the multi-source node planning data and the node constraints; Based on the mapping relationship, extract the quantitative indicators corresponding to the node constraints and perform normalization processing; The objective weights of the node constraints are calculated using the entropy weight method based on normalized quantitative indicators. By using the analytic hierarchy process, a judgment matrix is ​​constructed by combining the topographic data of the target area, the monitoring task requirements data, the node deployment cost data, and the monitoring coverage requirements data to obtain subjective correction weights. Based on a preset weighted fusion coefficient, the objective weights and the subjectively modified weights are fused to obtain a combined weight; The combined weights are normalized and verified to determine the weight coefficients of the node constraints.

9. The method according to claim 1, characterized in that, The data reasoning models include: random forest model, logistic regression model, and support vector machine model.

10. The method according to claim 1 or 9, characterized in that, The data reasoning model is trained using any of the following methods: It is trained based on the local historical monitoring data of each edge node; It is trained based on deviation cases that do not conform to the actual working conditions in each edge node; It is trained based on the positive and negative samples in each edge node.

Images