A multifunctional internet of things sensing system

Abstract

This invention relates to the field of IoT sensing and data transmission, and discloses a multifunctional IoT sensing system. It includes: multiple IoT sensing terminals, a protocol abstraction and capability mapping gateway, a task intent management unit, a task compilation unit, an edge evidence evaluation unit, a fuzzy interval collaborative triggering unit, a hierarchical reporting and original fragment replenishment unit, and a multi-link transmission unit. The system generates a terminal execution profile based on the sensing task intent and terminal capability description information, calculates an evidence index for standardized sensing objects, and determines whether to directly generate a sensing summary or trigger supplementary sampling or heterogeneous sensing evidence by associated terminals based on the relationship between the evidence index and a threshold range. When a remote business platform needs to verify the data, it replenishes the original data fragments within the corresponding time window. This application can improve the sensing accuracy, edge detection reliability, and transmission efficiency in multi-scenario tasks.

Description

Technical Field

[0001] This invention relates to the field of low-altitude surveillance and target identification, and more specifically to a multifunctional Internet of Things (IoT) sensing system. Background Technology

[0002] IoT sensing systems are widely used in scenarios such as park security, equipment operation and maintenance, environmental monitoring, warehouse management and facility status early warning. By deploying various sensing terminals in the target area, they continuously collect, transmit and process on-site status information to achieve remote sensing and management of the target object's operating status, environmental changes and abnormal events.

[0003] In existing technologies, temperature, vibration, door magnet, current, displacement, or image sensing terminals are typically connected to a gateway. The gateway then performs protocol conversion and uploads the data to a platform for unified analysis. Most systems employ preset sampling periods and fixed reporting strategies. While some systems can support multiple terminal access, they focus more on unified access and data aggregation for heterogeneous terminals, with relatively insufficient mechanisms for task allocation, collaborative verification, and differentiated transmission among terminals.

[0004] Under the above structure, existing systems typically handle different sensing tasks using fixed sampling, fixed uploading, and unified processing methods. Furthermore, they lack a hierarchical judgment process that combines terminal status, data timeliness, and multi-terminal consistency after a single point of failure occurs. This leads to two problems during use: firstly, it is easy to upload a large amount of raw data for a long time, increasing the communication and processing burden; secondly, it is easy to make judgments based solely on local anomalies, affecting the reliability of the sensing results. Consequently, it is difficult to simultaneously ensure both sensing accuracy and transmission efficiency in multiple scenarios. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multifunctional Internet of Things (IoT) sensing system to solve the technical problems existing in the prior art.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A multifunctional Internet of Things (IoT) sensing system, comprising: Multiple IoT sensing terminals are used to collect raw sensing data and cache it locally; The Protocol Abstraction and Capability Mapping Gateway communicates with multiple IoT sensing terminals to parse heterogeneous protocol data uploaded by each IoT sensing terminal and convert it into standardized sensing objects in a unified format, while generating capability description information corresponding to each IoT sensing terminal. The task intent management unit is used to receive and perceive task intent descriptions. The task compilation unit is connected to the task intent management unit and the protocol abstraction and capability mapping gateway, respectively. It is used to generate corresponding terminal execution profiles for each IoT sensing terminal based on the sensing task intent description and capability description information. The terminal execution profile includes the basic sampling period, event triggering conditions and reporting granularity. The edge evidence evaluation unit is connected to the protocol abstraction and capability mapping gateway and is used to calculate the evidence index based on the standardized perception object. The fuzzy interval collaborative triggering unit is connected to the task compilation unit, the edge evidence evaluation unit, and the protocol abstraction and capability mapping gateway, respectively. It is used to compare the evidence index with the upper threshold and the lower threshold. When the evidence index is higher than the upper threshold, a perception summary of the target event is generated. When the evidence index is lower than the lower threshold, an invalid summary or a normal state summary is generated. When the evidence index is between the upper threshold and the lower threshold, the associated IoT sensing terminal is triggered to perform supplementary sampling or heterogeneous perception corroboration according to the terminal execution profile, and the evidence index is updated according to the supplementary sampling data or heterogeneous perception corroboration data. The hierarchical reporting and original fragment replenishment unit is connected to the fuzzy interval collaborative triggering unit, the protocol abstraction and capability mapping gateway, and the remote business platform, respectively. It is used to send the perception summary, evidence index and time slice identifier to the remote business platform. When it receives the original fragment replenishment request issued by the remote business platform based on the time slice identifier, it controls the corresponding IoT perception terminal to extract the original data fragment in the corresponding time window and send it to the remote business platform. The multi-link transmission unit is connected to the hierarchical reporting and original fragment replenishment unit and the remote service platform, respectively. It is used to select the first transmission link when sending the perception summary, evidence index and time slice identifier, and to select the second transmission link when sending the original data fragment.

[0007] Preferably, each of the IoT sensing terminals includes a sensing module, a local control module, a local cache module, and a communication interface; The sensing module is used to collect sensing data; The local control module is used to control the sampling period, triggering method, and local feature extraction process of the corresponding perception module based on the terminal's execution profile. The local caching module is used to cache the original sensing data in segments according to time order; The communication interface is used to complete the data transmission and reception and control command interaction with the protocol abstraction and capability mapping gateway.

[0008] Preferably, the capability description information includes the sensing type, sampling range, maximum sampling frequency, currently available link type, buffer capacity, remaining energy, and online status of the corresponding IoT sensing terminal; The protocol abstraction and capability mapping gateway converts the capability description information of each IoT sensing terminal into capability vectors, and sends the capability vectors to the task compilation unit so that the task compilation unit can allocate sensing tasks according to the actual capability differences of different IoT sensing terminals.

[0009] Preferably, the perception task intent description includes the target event type, perception dimension requirements, allowable decision delay, credibility requirements, and original data retention duration; The task compilation unit generates a terminal execution profile based on the perceived task intent description and capability description information; The terminal execution profile also includes local feature extraction rules, original fragment cache window and collaborative supplementary sampling trigger conditions. The terminal execution profiles corresponding to different IoT sensing terminals are set according to their capability vectors and their respective sensing roles.

[0010] Preferably, the edge evidence evaluation unit evaluates the standardized perception object according to a preset fusion rule; The evidence index is generated based on the following factors: the current health status of the IoT sensing terminal, the freshness of the data, the sensing consistency between adjacent IoT sensing terminals, and the degree of deviation from the historical baseline. The current health status is used to reflect the stability of the corresponding IoT sensing terminal; The data time freshness is used to reflect the temporal correlation between the current data and the time of the target event determination. The perception consistency is used to reflect the degree of mutual verification of the same event by multiple IoT sensing terminals. The historical baseline deviation is used to reflect the magnitude of change in the current perception result relative to the historical normal state.

[0011] Preferably, when generating the evidence index, the edge evidence evaluation unit first normalizes the current health status, data freshness, perceptual consistency, and historical baseline deviation to obtain corresponding evaluation values, and then integrates and calculates each evaluation value according to preset weights to obtain the evidence index. The preset weights are set based on the target event type and the priority of the perception task.

[0012] Preferably, when the evidence index is between the upper and lower thresholds, the fuzzy interval collaborative triggering unit selects a target IoT sensing terminal from multiple IoT sensing terminals to perform supplementary sampling or heterogeneous sensing corroboration based on a pre-established association mapping table and collaborative priority matrix. The association mapping table is established based on the spatial proximity, sensory complementarity, and business association relationships between IoT sensing terminals, while the collaborative priority matrix is ​​determined based on the available link status, remaining energy, buffer capacity, and the urgency of the current task.

[0013] Preferably, the perception summary includes a target event category identifier, key feature data, a corresponding evidence index, a time slice identifier, and a source IoT perception terminal identifier; After receiving the perception summary, the remote service platform determines whether to issue an original fragment replenishment request based on the evidence index, key feature data, and business review rules. Upon receiving an original fragment replenishment request, the hierarchical reporting and original fragment replenishment unit calls the local cache module of the corresponding IoT sensing terminal to extract the original data fragment from the cache time window corresponding to the time slice identifier and send it to the remote business platform.

[0014] Preferably, the time slice identifier is generated by combining the task identifier, the IoT sensing terminal identifier, the start timestamp, and the end timestamp; The local cache module uses a circular cache method to store the original perception data and retains the original perception data for the corresponding duration according to the original segment cache window set in the terminal execution profile. When the original fragment replenishment request sent by the remote business platform contains a time slice identifier, the IoT sensing terminal locates the corresponding cache interval based on the start timestamp and end timestamp in order to extract the original data fragment corresponding to the target event determination process.

[0015] Preferably, the first transmission link is a low-bandwidth, low-power link, and the second transmission link is a high-bandwidth link or a parallel transmission link. The multi-link transmission unit is also used to switch or reselect the first transmission link and the second transmission link according to the current link load status, link delay status and link availability status.

[0016] In summary, the present invention has the following main beneficial effects: This application, by setting up a task intent management unit, a task compilation unit, an edge evidence evaluation unit, and a fuzzy interval collaborative triggering unit, enables the system to avoid fixed uploading and uniform processing of data collected by various IoT sensing terminals. Instead, it first generates corresponding terminal execution profiles for different IoT sensing terminals based on the target event type, sensing dimension requirements, allowable decision latency, and credibility requirements. Then, it generates evidence indices by combining standardized sensing objects and performs hierarchical judgments based on the relationship between the evidence index and upper and lower thresholds. This transforms the sensing process from static collection to dynamic orchestration driven by tasks, achieving the goal of adaptively adjusting sampling, judgment, and collaboration methods for different monitoring tasks. This improves the system's identification of target events in complex scenarios and its sensing flexibility, avoiding the problems of inefficient resource allocation and insufficient adaptability of sensing results caused by fixed work of each terminal and a single processing flow in existing technologies.

[0017] By incorporating the current health status of the terminal, the freshness of the data, the perceptual consistency between adjacent IoT sensing terminals, and the deviation from the historical baseline into the evidence index generation process, and triggering the associated IoT sensing terminals to perform supplementary sampling or heterogeneous sensing corroboration according to the association mapping table and the collaborative priority matrix when the evidence index is in the fuzzy range, the system does not draw conclusions directly based on a single abnormal data from a single terminal. Instead, it first forms initial evidence at the edge and then performs directional supplementary evidence correction on gray zone events. This makes the judgment basis of the target event more complete, thereby reducing the false alarm rate and the false negative rate and improving the credibility of edge judgment. This avoids the problems of judgment distortion, increased energy consumption, and excessive communication burden caused by relying on single-point abnormal triggering, arbitrary selection of supplementary evidence objects, or simultaneous wake-up of all terminals in existing technologies.

[0018] By setting up hierarchical reporting and original fragment replenishment units, as well as multi-link transmission units, the system normally only sends perception summaries, evidence indices, and time-slice identifiers. When the remote business platform needs to review the data, it replenishes the original data fragments within the corresponding time window according to the time-slice identifiers. Different transmission links are selected for summary data and original fragment data, thus forming a hierarchical transmission mechanism between the edge side and the remote business platform, which prioritizes the transmission of summaries and replenishes data on demand. This achieves the goal of reducing the amount of data transmitted under normal circumstances, shortening the latency of regular reporting, and improving the efficiency of link utilization while ensuring review capabilities. This avoids the problems of high bandwidth consumption, high latency, difficulty in backtracking and positioning, and uneven utilization of multi-link resources caused by uploading all original data in existing technologies. Attached Figure Description

[0019] Figure 1 This is a system block diagram of the present invention.

[0020] Figure 2This is a schematic diagram of the fuzzy interval collaborative supplementary proof mechanism of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1 refer to Figure 1-2 A multifunctional Internet of Things (IoT) sensing system, comprising: Multiple IoT sensing terminals are used to collect raw sensing data and cache it locally; The Protocol Abstraction and Capability Mapping Gateway communicates with multiple IoT sensing terminals to parse heterogeneous protocol data uploaded by each IoT sensing terminal and convert it into standardized sensing objects in a unified format, while generating capability description information corresponding to each IoT sensing terminal. The task intent management unit is used to receive and perceive task intent descriptions. The task compilation unit is connected to the task intent management unit and the protocol abstraction and capability mapping gateway, respectively. It is used to generate corresponding terminal execution profiles for each IoT sensing terminal based on the sensing task intent description and capability description information. The terminal execution profile includes the basic sampling period, event triggering conditions and reporting granularity. The edge evidence evaluation unit is connected to the protocol abstraction and capability mapping gateway and is used to calculate the evidence index based on the standardized perception object. The fuzzy interval collaborative triggering unit is connected to the task compilation unit, the edge evidence evaluation unit, and the protocol abstraction and capability mapping gateway, respectively. It is used to compare the evidence index with the upper threshold and the lower threshold. When the evidence index is higher than the upper threshold, a perception summary of the target event is generated. When the evidence index is lower than the lower threshold, an invalid summary or a normal state summary is generated. When the evidence index is between the upper threshold and the lower threshold, the associated IoT sensing terminal is triggered to perform supplementary sampling or heterogeneous perception corroboration according to the terminal execution profile, and the evidence index is updated according to the supplementary sampling data or heterogeneous perception corroboration data. The hierarchical reporting and original fragment replenishment unit is connected to the fuzzy interval collaborative triggering unit, the protocol abstraction and capability mapping gateway, and the remote business platform, respectively. It is used to send the perception summary, evidence index and time slice identifier to the remote business platform. When it receives the original fragment replenishment request issued by the remote business platform based on the time slice identifier, it controls the corresponding IoT perception terminal to extract the original data fragment in the corresponding time window and send it to the remote business platform. The multi-link transmission unit is connected to the hierarchical reporting and original fragment replenishment unit and the remote service platform, respectively. It is used to select the first transmission link when sending the perception summary, evidence index and time slice identifier, and to select the second transmission link when sending the original data fragment.

[0023] This embodiment uses a comprehensive sensing scenario consisting of equipment rooms, passageways, and the outer boundary within a park as an example. Multiple IoT sensing terminals are deployed in different locations. Each IoT sensing terminal, depending on its installation location and sensing capabilities, undertakes one or more tasks, including temperature sensing, humidity sensing, vibration sensing, current sensing, door magnetic status sensing, displacement sensing, smoke sensing, or image feature extraction. Each IoT sensing terminal is communicatively connected to a protocol abstraction and capability mapping gateway, which is deployed in an edge computing node. The task intent management unit, task compilation unit, edge evidence evaluation unit, fuzzy interval collaborative triggering unit, hierarchical reporting and original fragment recovery unit, and multi-link transmission unit can all be deployed on the same edge server, or they can be distributed between the edge gateway and the edge server according to computing power and network conditions, but their logical connection relationship remains unchanged.

[0024] In this embodiment, each IoT sensing terminal includes a sensing module, a local control module, a local cache module, and a communication interface. The sensing module acquires raw sensing data. The local control module receives the terminal execution profile from the protocol abstraction and capability mapping gateway and controls the sensing module's basic sampling period, event triggering conditions, local feature extraction rules, and reporting granularity based on the terminal execution profile. The local cache module performs time-division caching of the raw sensing data, preferably using a circular cache method to avoid the cache space being occupied for a long time by a single event during continuous operation. The communication interface is used to complete terminal status reporting, sensing data transmission, execution profile reception, re-transmission request reception, and raw data fragment retransmission.

[0025] To enable IoT sensing terminals from different manufacturers, using different communication protocols, and with different data formats to participate in the same sensing task, in this embodiment, the protocol abstraction and capability mapping gateway performs protocol parsing on the data frames uploaded by each IoT sensing terminal and converts the parsing results into standardized sensing objects in a unified format. A standardized sensing object includes at least a task identifier, terminal identifier, acquisition time, sensing type, key feature data, terminal current health status, link status, and data source marker. Through this unified conversion, subsequent task compilation units and edge evidence evaluation units no longer depend on the underlying protocol format of specific terminals, but instead perform scheduling and judgment based on standardized sensing objects in a unified format.

[0026] In addition to standardized sensing objects, the protocol abstraction and capability mapping gateway also generates capability description information for each IoT sensing terminal. This capability description information includes at least the sensing type, sampling range, maximum sampling frequency, currently available link type, buffer capacity, remaining energy, and online status. To enable the task compilation unit to schedule different terminals in a unified manner, the protocol abstraction and capability mapping gateway further converts the capability description information into capability vectors. Each dimension of the capability vector corresponds to the terminal's sensing capability, link capability, buffer capability, and continuous working capability, respectively. These capability vectors are updated periodically or triggered when the terminal's status changes, thereby ensuring that the subsequent task compilation results remain consistent with the actual capabilities of the current terminal.

[0027] The task intent management unit receives perception task intent descriptions from remote business platforms. These descriptions do not merely provide a general monitoring target, but rather specify at least the target event type, perception dimension requirements, allowable decision latency, credibility requirements, and raw data retention duration. For example, when a remote business platform needs to monitor abnormal equipment temperature rise events, the target event type can be set to "abnormal equipment temperature rise," the perception dimension requirements can be set to three dimensions: temperature, current, and vibration, the allowable decision latency corresponds to the maximum time for the event to be preliminarily judged at the edge, the credibility requirement is used to limit the threshold setting range of the evidence index, and the raw data retention duration is used to constrain the cache window length of the local cache module. As another example, when a remote business platform needs to monitor abnormal opening events, the target event type can be changed to "abnormal opening," and the perception dimension requirements can be changed accordingly to "door magnet status," "displacement change," and "nearby vibration." Thus, it can be seen that the multi-functionality in this application is not a stack of multiple unrelated fixed functions, but rather the same system forming different perception orchestration and data processing flows driven by different task intents.

[0028] The task compilation unit generates corresponding terminal execution profiles for each IoT sensing terminal based on the perception task intent description and capability vector. The terminal execution profile includes at least the basic sampling period, event triggering conditions, and reporting granularity, and may further include local feature extraction rules, raw fragment cache windows, and collaborative supplementary sampling triggering conditions. The basic sampling period refers to the regular sampling interval of the terminal when no local anomalies are observed. Event triggering conditions refer to the conditions under which the terminal switches from a regular sampling state to a high-frequency sampling state or a local feature extraction state. Reporting granularity refers to whether the terminal reports raw data, feature data, or summary data. Local feature extraction rules refer to the key feature types output by the terminal after processing the sampling window locally, such as peak value, mean, slope, frequency domain energy, or the number of times the open / closed state changes. The raw fragment cache window refers to the length of time the terminal retains raw data fragments for each time slice in its local cache module. Collaborative supplementary sampling triggering conditions refer to the types of associated terminals allowed to participate in supplementary evidence collection, the triggering method, and the supplementary sampling time limit when the evidence index corresponding to a certain terminal falls into the fuzzy range. The reporting granularity is determined based on the target event type, allowable decision latency, the terminal's current link capability, and the need for remote verification.

[0029] To avoid task compilation merely remaining at the abstract task allocation level, this embodiment further stipulates that when forming a terminal execution profile, the task compilation unit first filters candidate terminals that meet the perception dimensions required for the target event type based on the capability vector. Then, it determines the priority terminals participating in the initial edge detection based on the allowable decision latency and the currently available link type. Next, it sets the original fragment cache window for each terminal based on the original data retention time and cache capacity. Finally, it establishes an association mapping table based on the spatial proximity and perceptual complementarity relationships between terminals for subsequent fuzzy interval collaborative triggering. In this way, the task compilation result is not simply the distribution of the same set of sampling periods, but rather the formation of terminal execution profiles with differentiated content for different terminals.

[0030] To further clarify the meaning of the perception roles described in the terminal execution profile, in this embodiment, the perception roles include at least a primary judgment role, a related supplementary evidence role, and an auxiliary corroborating role. The primary judgment role is used to form an initial evidence index, the related supplementary evidence role is used to perform supplementary sampling or heterogeneous perception corroboration when the evidence index is in an ambiguous range, and the auxiliary corroborating role is used to provide non-real-time supplementary features or background state information. The task compilation unit assigns corresponding perception roles to different terminals based on the target event type, capability vector, spatial location, and perception complementarity relationship.

[0031] The edge evidence assessment unit generates an evidence index based on standardized sensing objects. The evidence index reflects the credibility and completeness of current event clues. It is not solely based on whether a single sensor value exceeds its limits, but rather comprehensively considers the current health status of the IoT sensing terminal, data freshness, sensing consistency between adjacent IoT sensing terminals, and deviation from the historical baseline. To ensure a clear implementation path for this evaluation process, in this embodiment, the evidence index is calculated using the following formula:

[0032] in, Indicates the first Evidence index corresponding to a standardized perceived object. This indicates the current health status evaluation value of the terminal. This represents the data's time freshness rating. This represents the perceived consistency evaluation value. This indicates the deviation from the historical baseline. , , and Let each represent the weight of the corresponding evaluation value, and satisfy the following: ; The current health status evaluation value, data freshness evaluation value, perception consistency evaluation value, and historical baseline deviation evaluation value are all normalized before participating in the fusion calculation. The normalized values ​​all range from 0 to 1, thus ensuring that different evaluation dimensions have a unified dimension. To ensure that the terminal's current health status evaluation value has a clear implementation path, in this embodiment, the terminal's current health status evaluation value is calculated using the following formula: ; in, Indicates the first The current health status evaluation value of each terminal. This represents the online status evaluation value. This indicates the heart rate continuity assessment value. This indicates the evaluation value of the self-inspection result. This represents the power supply stability evaluation value. Indicates the integrity evaluation value of the communication message. This indicates the most recent calibration status evaluation value. to These represent the weights corresponding to each sub-item evaluation value, and the sum of all weights is 1. After normalization, each of the above sub-item evaluation values ​​ranges from 0 to 1. When the terminal's online status is stable, heartbeat is continuous, self-test passes, power supply is stable, messages are complete, and a valid calibration has recently been completed, Higher; conversely, reduce.

[0033] To give the data freshness evaluation value a clear meaning, in this embodiment, the data freshness evaluation value... It can be obtained in the following ways: ; in, Indicates the first The time difference between the acquisition time of a standardized sensing object and the current judgment time. This indicates the maximum effective time window corresponding to the allowable decision delay for this task. The closer the data acquisition time is to the current decision time, the longer the time window becomes. The closer it is to 1; when the time difference exceeds the allowable decision delay, Set it to 0 to avoid outdated data from continuing to participate in high-reliability judgments.

[0034] To ensure a clear implementation method for the perceived consistency evaluation value, in this embodiment, the perceived consistency evaluation value is calculated based on the degree of support for the current event by each associated terminal in the association mapping table, which can be specifically expressed as follows: ; in, Indicates the relationship with the first The number of terminals associated with each standardized sensing object. Indicates the first The terminal and the first The association weight between each associated terminal Indicates the first The level of support for the current event by each associated terminal. The value range is from 0 to 1. The degree of support... According to the The matching result between the local features output by each associated terminal and the current target event template is determined. A perfect match is assigned a value of 1, a non-match is assigned a value of 0, and a partial match is assigned a value between 0 and 1 based on similarity. Thus, when multiple associated terminals provide features that match the current event, the perceived consistency evaluation value increases; when the results from the associated terminals are inconsistent with the current object, the perceived consistency evaluation value decreases.

[0035] To ensure that the historical baseline deviation evaluation value is not merely a concept, in this embodiment, the edge evidence evaluation unit maintains a corresponding historical baseline for each task template and each terminal. The historical baseline is not a fixed constant, but rather is statistically derived from multiple valid historical samples of the terminal under normal conditions, the same task template, and within the same time period. Historical baseline deviation evaluation value It can be calculated as follows: ; in, This represents the key feature values ​​in the currently standardized sensing object. This represents the historical baseline feature value of the terminal under the corresponding task template. This indicates the allowable fluctuation range of the historical baseline. When the current key characteristic deviates significantly from its historical normal state... Increase; when the deviation is small or within the normal fluctuation range. Smaller.

[0036] The weights in the evidence index , , and The settings are not arbitrary, but pre-configured based on the target event type and the priority of the sensing task. For example, for alarm-type tasks with high timeliness requirements, the weight corresponding to data freshness is increased; for long-term operation monitoring tasks with high terminal stability requirements, the weight corresponding to the current health status of the terminal is increased. The sensing task priority is pre-given by the task template, including at least emergency, normal, and background analysis levels, with different priority levels corresponding to different weight configurations and allowable decision delay configurations. For the same task template, the preset weights remain fixed within a weight update cycle to ensure consistency in the determination of the evidence index within the same task cycle.

[0037] The fuzzy interval collaborative triggering unit compares the evidence index with an upper and lower threshold. Both the upper and lower thresholds are pre-defined by the task template, derived from historical labeled sample statistics, on-site debugging results, and operational false alarm / missed alarm tolerance settings. Since the evidence index, after normalization, ranges from 0 to 1, both the upper and lower thresholds are real numbers between 0 and 1, with the upper threshold being greater than the lower threshold. Specifically, when the evidence index is higher than the upper threshold, it indicates that the current standardized sensing object has high credibility, and a sensing summary of the target event can be directly generated; when the evidence index is lower than the lower threshold, it indicates that the current clue does not warrant further supplementary evidence, and an invalid summary or a normal state summary is generated; only when the evidence index is between the upper and lower thresholds is the current judgment considered to be in a fuzzy interval, requiring triggering associated terminals to perform supplementary sampling or heterogeneous sensing corroboration. Therefore, this application does not activate all terminals every time the initial judgment is uncertain, but only performs a conditional and directional supplementary evidence process for gray-zone events.

[0038] The ordinary state summary indicates that the current standardized sensing object does not meet the conditions for the target event to occur, but retains the current key feature data, evidence index, and time slice identifier for remote business platform to record. The invalid summary indicates that the current standardized sensing object neither meets the conditions for the target event to occur nor the conditions for record-keeping and verification.

[0039] To clarify the selection method for supplementary verification terminals, in this embodiment, the fuzzy interval collaborative triggering unit pre-stores an association mapping table and a collaborative priority matrix. The association mapping table indicates which terminals have supplementary verification relationships under the current task template. A supplementary verification relationship refers to the ability of the data type collected by one terminal to support or exclude the event judgment of another terminal. For example, in an abnormal temperature rise task, a supplementary verification relationship can be established between the temperature terminal, the current terminal, and the vibration terminal; in an abnormal opening task, a supplementary verification relationship can be established between the door magnetic terminal, the displacement terminal, and the adjacent vibration terminal. The collaborative priority matrix is ​​used to determine the actual priority wake-up order among multiple selectable supplementary verification terminals, and its calculation can take the following form: ; in, Indicates the first The priority score of each candidate terminal for supplementary certification. This indicates the strength of the association between the candidate terminal and the current event. This indicates the current available link status evaluation value for the candidate terminal. This indicates the remaining energy evaluation value of the candidate terminal. This indicates the evaluation value of the remaining cache space of the candidate terminal. , , and These represent the weights of the corresponding factors, and the sum of the weights is 1. The correlation strength... The value is generated by weighting the spatial proximity, perceptual complementarity, and business relevance between the candidate supplementary verification terminal and the main judgment terminal corresponding to the current event, and is normalized to a value between 0 and 1. The fuzzy interval collaborative triggering unit selects one or more candidate terminals with the highest priority scores to perform supplementary sampling or heterogeneous perception verification, and limits the maximum waiting time for the completion of supplementary verification to meet the allowable decision delay requirements of the task.

[0040] After supplementary sampling or heterogeneous sensory corroboration is completed, the marginal evidence assessment unit does not recalculate all evidence from scratch. Instead, it merges the initial evidence index with the collaborative supplementary evidence results to obtain an updated evidence index. The updated evidence index can be calculated as follows: ; in, This indicates the updated evidence index. This indicates the initial evidence index before supplementary evidence is provided. This indicates the collaborative evidence score regenerated based on the supplementary evidence data. This represents the fusion coefficient, with a value ranging from 0 to 1. For tasks that heavily rely on the reliability of the initial data... The value is relatively large; for tasks that rely more on heterogeneous corroboration, The value is relatively small. Through the above method, the edge side can incrementally correct evidence for events in ambiguous intervals without overturning the original judgment process.

[0041] When the updated evidence index is higher than the upper threshold, the fuzzy interval collaborative triggering unit generates a perceptual summary of the target event; when the updated evidence index is still within the fuzzy interval or falls below the lower threshold, an invalid summary, a normal state summary, or a summary awaiting review is generated according to the rules in the task template. It should be noted that the perceptual summary in this embodiment is not a simple truncation of the original data, but a structured summary constructed to meet the review needs of the remote business platform. The perceptual summary includes at least the target event category identifier, key feature data, the corresponding evidence index, the time slice identifier, and the source IoT sensing terminal identifier. The key feature data is determined by the local feature extraction rules in the terminal execution profile. The data type of key feature data differs under different task templates, such as the current peak value, the slope of change, the number of state flips, or multi-dimensional feature combination values, rather than all original sampling points.

[0042] To enable the remote service platform to accurately recall the original fragment based on the summary information, in this embodiment, the time slice identifier is generated by combining the task identifier, terminal identifier, start timestamp, and end timestamp. Specifically, the terminal establishes cache segments in its local cache module using fixed time windows or a fixed number of sampling points, and records the corresponding start and end timestamps for each cache segment. The time slice identifier can correspond to a single cache segment or a combination of multiple consecutive cache segments. After receiving the perception summary, the remote service platform determines whether to issue an original fragment replenishment request based on the evidence index, key feature data, and business review rules. The business review rules include at least one of the following situations: the evidence index is still in the fuzzy range, the perception results between associated terminals are inconsistent, the current health status of the source IoT perception terminal is lower than a preset health threshold, or the target event occurs during a manually set high-risk period. Once a replenishment request is issued, the request directly carries the time slice identifier, and the hierarchical reporting and original fragment replenishment unit control the corresponding IoT perception terminal to extract the original data fragment within the corresponding time window from the local cache module and send it to the remote service platform. Therefore, the remote business platform does not need to repeatedly issue vague time retrieval commands, but can accurately locate the original data segment corresponding to the current event by using the time slice identifier.

[0043] During data reporting and resubmission, the multi-link transmission unit selects different links based on data format and transmission requirements. In this embodiment, when sending the perception summary, evidence index, and time slice identifier, the first transmission link is preferentially selected. The first transmission link is preferably a low-bandwidth, low-power link, the purpose of which is to quickly complete the reporting of the initial edge judgment results with a low communication burden. When sending the original data fragment, the second transmission link is preferentially selected. The second transmission link is preferably a high-bandwidth link or a parallel transmission link, the purpose of which is to meet the need for the original fragment to be resubmitted in a short time. The first transmission link can be a narrowband wireless link, a low-power wide-area link, or other links suitable for low-speed reporting, and the second transmission link can be a cellular broadband link, a wireless LAN link, a wired Ethernet link, or a parallel transmission link composed of the aforementioned links. The multi-link transmission unit also continuously monitors the link load status, link delay status, and link availability status. When the first transmission link is temporarily unavailable, it can switch to a backup low-power link while maintaining the summary priority principle; when the second transmission link is congested, it can be split into multiple sub-fragments according to the current original fragment size and transmitted separately through parallel transmission links, and then reassembled by the remote service platform according to the time slice identifier.

[0044] The complete working process of this embodiment is as follows. The remote service platform first issues a description of the sensing task intent. After receiving the task intent, the task intent management unit sends the target event type, sensing dimension requirements, allowed decision latency, credibility requirements, and original data retention time to the task compilation unit. The protocol abstraction and capability mapping gateway synchronously provides the task compilation unit with the capability vectors of each IoT sensing terminal. The task compilation unit generates a terminal execution profile for each terminal based on the task intent and capability vectors, and issues the terminal execution profile to the corresponding terminal. The local control module controls the sensing module to perform basic sampling, triggered sampling, and local feature extraction based on the terminal execution profile, while simultaneously writing the original sensing data to the local cache module. The protocol abstraction and capability mapping gateway performs protocol parsing and standardization processing on the sensing results reported by each terminal, forming standardized sensing objects and sending them to the edge evidence evaluation unit. The edge evidence evaluation unit generates an evidence index based on the above formula. The fuzzy interval collaborative triggering unit determines whether to directly generate a sensing summary or trigger supplementary evidence based on the relationship between the evidence index and the threshold interval. If supplementary evidence is triggered, the fuzzy interval collaborative triggering unit selects the target supplementary evidence terminal based on the association mapping table and collaborative priority matrix, and updates the evidence index after supplementary evidence is completed. Subsequently, the hierarchical reporting and original fragment replenishment unit sends the perception summary, evidence index, and time slice identifier to the remote service platform via the first transmission link. If the remote service platform needs to verify, it initiates an original fragment replenishment request based on the time slice identifier. The corresponding terminal retrieves the original data fragment from its local cache module and sends it to the remote service platform via the second transmission link. This completes the entire closed-loop process from task generation to gray zone evidence replenishment, and from summary reporting to original fragment replenishment.

[0045] To further illustrate the multi-functional features and task compilation methods in this application, the following explanation uses two specific task templates to illustrate the differences in execution of the same system.

[0046] In the task of detecting abnormal temperature rise in equipment, the target event type received by the task intent management unit is abnormal temperature rise in equipment. The perception dimensions are required to be temperature, current, and vibration. The allowed decision latency is relatively short, and the reliability requirement is high. The original data retention period covers the continuous time period before and after the abnormal temperature rise. Under this task, the task compilation unit prioritizes the temperature terminal as the primary judgment terminal and sets the current terminal and vibration terminal as associated supplementary evidence terminals. In the terminal execution profile of the temperature terminal, the basic sampling period is relatively short, and the event triggering condition is set to the temperature rise rate exceeding a preset threshold or multiple consecutive temperature sampling values ​​deviating significantly from the historical baseline within a short period. Local feature extraction rules include temperature rise rate, temperature peak value, and temperature fluctuation amplitude. In the terminal execution profile of the current terminal, the event triggering condition is set to abnormal load changes or a sudden surge in instantaneous current. Local feature extraction rules include current peak value and current mean deviation. In the terminal execution profile of the vibration terminal, the event triggering condition is set to an abnormal increase in vibration spectrum energy or the occurrence of abnormal vibration that is asynchronous with equipment load changes. When the initial evidence index generated by the temperature terminal is in the fuzzy range, the fuzzy range collaborative triggering unit prioritizes waking up the current terminal and vibration terminal for supplementary evidence. If both the current terminal and the vibration terminal provide characteristics consistent with the abnormal temperature rise of the equipment, the updated evidence index increases, and an abnormal temperature rise perception summary is generated; if the supplementary evidence results do not support the abnormal temperature rise, the updated evidence index decreases, and only a normal state summary or a summary pending review is generated.

[0047] In the abnormal opening task, the target event type received by the task intent management unit changes to abnormal opening, and the perception dimensions are changed to door sensor status, displacement change, and nearby vibration. The decision latency is allowed to be shorter than in general state monitoring tasks, but the reliability requirement is higher than in ordinary state perception tasks. In this task, the task compilation unit prioritizes the door sensor terminal as the primary judgment terminal and sets the displacement terminal and nearby vibration terminal as associated supplementary verification terminals. In the terminal execution profile of the door sensor terminal, the event triggering condition is set to the door sensor opening / closing state switching during an unauthorized period, and the local feature extraction rules include the number of opening / closing state flips and the duration of the door opening. In the terminal execution profile of the displacement terminal, the event triggering condition is set to the door or cabinet displacement exceeding a preset range, and the local feature extraction rules include the displacement amount and displacement duration. In the terminal execution profile of the nearby vibration terminal, the event triggering condition is set to a short-term sudden increase in mechanical vibration, and the local feature extraction rules include the peak acceleration and the number of vibrations. At this time, the terminal execution profiles in the same system are significantly different from those in the abnormal temperature rise task, indicating that this application is not a fixed, single perception process, but rather dynamically generates different perception orchestrations and supplementary verification mechanisms according to the task intent.

[0048] Example 2 To further illustrate the coordination relationship between the standardized sensing object, time slice identifier, original fragment buffer window, and original fragment re-filling request in this application, the data organization and re-filling process are further explained below.

[0049] In this embodiment, the IoT sensing terminal generates a corresponding internal record item after each sampling or each local feature extraction window is completed. The internal record item includes at least the terminal identifier, task identifier, acquisition time, acquisition window number, key feature data index, and local cache address index. The local cache module writes the original sensing data in the order of acquisition time and maintains a time index table. After receiving the local feature data sent by the terminal, the protocol abstraction and capability mapping gateway encapsulates the terminal identifier, task identifier, acquisition time, sensing type, key feature data, and status field into a standardized sensing object. The status field includes at least the terminal's current health status, current link status, and execution profile version number. By carrying the execution profile version number in the standardized sensing object, ambiguity can be avoided where the edge side interprets the new data according to the old version rules after the terminal executes a profile update.

[0050] After the edge evidence assessment unit generates an evidence index based on the standardized sensing object, the sensing summary generated by the fuzzy interval collaborative triggering unit includes not only key feature data but also feature type markers and time slice identifiers corresponding to the key feature data. The feature type marker informs the remote service platform whether the key feature data is a peak value, mean value, frequency domain energy, state flip count, or displacement. The time slice identifier is used to uniquely locate the original segment. The generation rule for the time slice identifier is obtained by combining the task identifier, terminal identifier, start timestamp, and end timestamp in a fixed order. For a time slice identifier corresponding to a single cache segment, its start timestamp and end timestamp directly correspond to the boundary of a cache segment in the local cache module; for a time slice identifier spanning multiple cache segments, its start timestamp corresponds to the start boundary of the first cache segment, and its end timestamp corresponds to the end boundary of the last cache segment. The remote service platform does not need to understand the internal cache structure of each terminal; it only needs to include the time slice identifier in the replenishment request, and the hierarchical reporting and original segment replenishment unit can complete the local location and data extraction.

[0051] To avoid cache hit failures, in this embodiment, the original fragment cache window is set according to the following logic: The cache window length must at least cover the allowable decision latency; The cache window length must at least cover the maximum waiting time for collaborative verification of fuzzy intervals; The cache window length must at least cover the network round-trip latency required for the perception summary to be delivered to the remote service platform and trigger a replenishment request. When the task compilation unit executes the profiling at the compilation terminal, it incorporates the above three time components into the setting of the original fragment cache window, thereby ensuring that when the remote service platform issues a replenishment request, the original data fragments within the corresponding time window are still stored in the local cache module.

[0052] Example 3 To further illustrate the source of parameters and the feasibility of implementation, this embodiment further explains the formation process of weights, thresholds, historical baselines, and associated weights.

[0053] Weights used in the calculation of the evidence index , , and Generated from a task template configuration. Specifically, during the initial system deployment, for each target event type, real operational data is collected over a period of time, and combined with maintenance records or manual review results to form labeled event samples. The task compilation unit or independent configuration tool analyzes the impact of different evaluation factors on the false positive rate and false negative rate based on these labeled event samples, and generates the initial weights corresponding to the current task template. After the system goes live, if a sufficient number of completed review samples are accumulated, the weights in the task template can be updated without changing the overall technical concept of the claims.

[0054] The logic for setting the upper and lower thresholds is similar; essentially, it maps false positive tolerance and false negative tolerance to the boundary range of the evidence index, rather than arbitrarily specifying them based on experience. Since the evidence index ranges from 0 to 1, the upper and lower thresholds are also limited to this range, with the upper threshold always being greater than the lower threshold. Different task templates can use different threshold combinations, but the same task template remains fixed within a single threshold update cycle.

[0055] The historical baseline is formed as follows. For each terminal, during a period of stable operation where no target event has been confirmed, the system extracts key feature data from standardized sensing objects and stores it according to task template, time period, and operating condition. Then, based on similar samples, the system generates the historical baseline feature value and allowable fluctuation range for that terminal under the corresponding task template. Thus, the data source upon which the historical baseline deviation evaluation value relies is clear, and it can be updated as the terminal's long-term operating status changes.

[0056] The formation process of the association mapping table and association weights is as follows. During system deployment, the spatial proximity between terminals is first determined based on the installation topology. Then, the complementary relationship of perceptions is determined based on whether the perception types can mutually support target events. Finally, the business association relationship is determined based on whether there are common objects in the business scenario. On this basis, the protocol abstraction and capability mapping gateway generates association weights for each pair of candidate associated terminals. The closer the spatial distance, the more complementary the perception types, and the more consistent the objects, the higher the association weight. The association mapping table formed in the above way is not an abstract association concept, but a data structure that can be directly called by the fuzzy interval collaborative triggering unit.

[0057] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-functional Internet of Things (IoT) sensing system, comprising: Includes the following steps: Multiple IoT sensing terminals are used to collect raw sensing data and cache it locally; The Protocol Abstraction and Capability Mapping Gateway communicates with multiple IoT sensing terminals to parse heterogeneous protocol data uploaded by each IoT sensing terminal and convert it into standardized sensing objects in a unified format, while generating capability description information corresponding to each IoT sensing terminal. The task intent management unit is used to receive and perceive task intent descriptions. The task compilation unit is connected to the task intent management unit and the protocol abstraction and capability mapping gateway, respectively. It is used to generate corresponding terminal execution profiles for each IoT sensing terminal based on the sensing task intent description and capability description information. The terminal execution profile includes the basic sampling period, event triggering conditions and reporting granularity. The edge evidence evaluation unit is connected to the protocol abstraction and capability mapping gateway and is used to calculate the evidence index based on the standardized perception object. The fuzzy interval collaborative triggering unit is connected to the task compilation unit, the edge evidence evaluation unit, and the protocol abstraction and capability mapping gateway, respectively. It is used to compare the evidence index with the upper threshold and the lower threshold. When the evidence index is higher than the upper threshold, a perception summary of the target event is generated. When the evidence index is lower than the lower threshold, an invalid summary or a normal state summary is generated. When the evidence index is between the upper threshold and the lower threshold, the associated IoT sensing terminal is triggered to perform supplementary sampling or heterogeneous perception corroboration according to the terminal execution profile, and the evidence index is updated according to the supplementary sampling data or heterogeneous perception corroboration data. The hierarchical reporting and original fragment replenishment unit is connected to the fuzzy interval collaborative triggering unit, the protocol abstraction and capability mapping gateway, and the remote business platform, respectively. It is used to send the perception summary, evidence index and time slice identifier to the remote business platform. When it receives the original fragment replenishment request issued by the remote business platform based on the time slice identifier, it controls the corresponding IoT perception terminal to extract the original data fragment in the corresponding time window and send it to the remote business platform. The multi-link transmission unit is connected to the hierarchical reporting and original fragment replenishment unit and the remote service platform, respectively. It is used to select the first transmission link when sending the perception summary, evidence index and time slice identifier, and to select the second transmission link when sending the original data fragment.

2. The multi-functional IoT sensing system according to claim 1, wherein, Each of the IoT sensing terminals includes a sensing module, a local control module, a local cache module, and a communication interface; The sensing module is used to collect sensing data; The local control module is used to control the sampling period, triggering method, and local feature extraction process of the corresponding perception module based on the terminal's execution profile. The local caching module is used to cache the original sensing data in segments according to time order; The communication interface is used to complete the data transmission and reception and control command interaction with the protocol abstraction and capability mapping gateway.

3. The multi-functional IoT sensing system according to claim 2, wherein, The capability description information includes the sensing type, sampling range, maximum sampling frequency, currently available link type, buffer capacity, remaining energy, and online status of the corresponding IoT sensing terminal. The protocol abstraction and capability mapping gateway converts the capability description information of each IoT sensing terminal into capability vectors, and sends the capability vectors to the task compilation unit so that the task compilation unit can allocate sensing tasks according to the actual capability differences of different IoT sensing terminals.

4. A multifunctional IoT sensing system according to claim 3, characterized in that, The perception task intent description includes the target event type, perception dimension requirements, allowable decision delay, credibility requirements, and original data retention duration; The task compilation unit generates a terminal execution profile based on the perceived task intent description and capability description information; The terminal execution profile also includes local feature extraction rules, original fragment cache window and collaborative supplementary sampling trigger conditions. The terminal execution profiles corresponding to different IoT sensing terminals are set according to their capability vectors and their respective sensing roles.

5. A multifunctional Internet of Things sensing system according to claim 4, characterized in that, The edge evidence evaluation unit evaluates the standardized perception objects according to preset fusion rules; The evidence index is generated based on the following factors: the current health status of the IoT sensing terminal, the freshness of the data, the sensing consistency between adjacent IoT sensing terminals, and the degree of deviation from the historical baseline. The current health status is used to reflect the stability of the corresponding IoT sensing terminal; The data time freshness is used to reflect the temporal correlation between the current data and the time of the target event determination. The perception consistency is used to reflect the degree of mutual verification of the same event by multiple IoT sensing terminals. The historical baseline deviation is used to reflect the magnitude of change in the current perception result relative to the historical normal state.

6. A multifunctional Internet of Things sensing system according to claim 5, characterized in that, When generating the evidence index, the edge evidence evaluation unit first normalizes the current health status, data freshness, perceptual consistency, and historical baseline deviation to obtain the corresponding evaluation values. Then, it merges and calculates the evaluation values ​​according to the preset weights to obtain the evidence index. The preset weights are set based on the target event type and the priority of the perception task.

7. A multifunctional IoT sensing system according to claim 6, characterized in that, When the evidence index is between the upper and lower thresholds, the fuzzy interval collaborative triggering unit selects a target IoT sensing terminal from multiple IoT sensing terminals to perform supplementary sampling or heterogeneous sensing corroboration based on a pre-established association mapping table and collaborative priority matrix. The association mapping table is established based on the spatial proximity, sensory complementarity, and business association relationships between IoT sensing terminals, while the collaborative priority matrix is ​​determined based on the available link status, remaining energy, buffer capacity, and the urgency of the current task.

8. A multifunctional Internet of Things sensing system according to claim 7, characterized in that, The perception summary includes the target event category identifier, key feature data, corresponding evidence index, time slice identifier, and source IoT perception terminal identifier; After receiving the perception summary, the remote service platform determines whether to issue an original fragment replenishment request based on the evidence index, key feature data, and business review rules. Upon receiving an original fragment replenishment request, the hierarchical reporting and original fragment replenishment unit calls the local cache module of the corresponding IoT sensing terminal to extract the original data fragment from the cache time window corresponding to the time slice identifier and send it to the remote business platform.

9. A multifunctional Internet of Things sensing system according to claim 8, characterized in that, The time slice identifier is generated by combining the task identifier, the IoT sensing terminal identifier, the start timestamp, and the end timestamp. The local cache module uses a circular cache method to store the original perception data and retains the original perception data for the corresponding duration according to the original segment cache window set in the terminal execution profile. When the original fragment replenishment request sent by the remote business platform contains a time slice identifier, the IoT sensing terminal locates the corresponding cache interval based on the start timestamp and end timestamp in order to extract the original data fragment corresponding to the target event determination process.

10. A multifunctional Internet of Things sensing system according to claim 9, characterized in that, The first transmission link is a low-bandwidth, low-power link, and the second transmission link is a high-bandwidth link or a parallel transmission link; The multi-link transmission unit is also used to switch or reselect the first transmission link and the second transmission link according to the current link load status, link delay status and link availability status.

Images