Park public safety comprehensive management and control system based on Internet of Things

Abstract

The invention belongs to the technical field of public safety emergency management and control and Internet of Things fusion, and discloses a park public safety comprehensive management and control system based on the Internet of Things. Comprising a data acquisition module used for acquiring user event description information and campus Internet of Things equipment data in real time; the event processing and verification module is used for carrying out association analysis on the event list and campus Internet of Things data, obtaining event clusters containing credibility scores and determining the situation level of each event cluster by constructing a level decision tree, and the resource scheduling and guarantee module is used for triggering a network resource scheduling process according to the situation level of each event cluster. Dynamically creating an emergency network slice instance based on the generated resource scheduling request command, performing exclusive network resource allocation based on the emergency network slice instance, and outputting resource allocation completion confirmation; and the emergency response and monitoring module is used for confirming and executing response operation based on resource allocation, continuously monitoring the progress of the reported event and outputting a response execution report.

Description

Technical Field

[0001] This invention relates to the field of public safety emergency management and Internet of Things (IoT) integration technology, and more specifically, to an IoT-based integrated public safety management and control system for industrial parks. Background Technology

[0002] With the expansion of higher education, university campuses have become special public areas with dense populations and complex functions. The management of public safety on campus is directly related to the personal and property safety of teachers and students and the normal teaching and research order. It is an indispensable and rigid requirement for the intelligent management system of campuses.

[0003] Existing university campus public safety management systems typically employ a three-tier architecture: front-end sensing terminals, campus data gateways, and back-end alarm platforms. In this system, front-end sensing data is uniformly aggregated by the campus data gateway and transmitted to the back-end alarm platform. The platform usually treats a single report from a faculty or student, or an alarm from a single IoT device, as an independent processing unit. After verification, it triggers an alarm to respond to security incidents. However, this management system does not consider scenarios where the same real security incident is reported by multiple faculty or students successively. It struggles to automatically identify and correlate these multiple pieces of information pointing to the same event. For example, in a university's chemical engineering laboratory building corridor, slight smoke is generated due to dust accumulation and heat in the ventilation ducts. Faculty and students passing by submit reports via the campus safety app. The management system identifies these as multiple independent alarms, triggering multiple sets of emergency communication resource scheduling tasks in parallel. This requires command personnel to invest extra time in manual deduplication and correlation analysis, resulting in delays in issuing emergency response instructions.

[0004] In view of this, the present invention proposes an Internet of Things-based comprehensive management and control system for public safety in industrial parks to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a campus public safety integrated management and control system based on the Internet of Things, comprising: a data acquisition module, used to acquire user event description information and campus Internet of Things device data in real time; The event processing and verification module is used to extract user event description information to form an event list, perform correlation analysis between the event list and campus IoT data to obtain event clusters containing credibility scores, and analyze each event cluster by constructing a hierarchical decision tree to determine the status level of each event cluster. The resource scheduling and support module is used to trigger the network resource scheduling process according to the status level of each event cluster, dynamically create emergency network slice instances based on the generated resource scheduling request commands, allocate dedicated network resources based on the emergency network slice instances, and output a confirmation of resource allocation completion. The emergency response and monitoring module is used to confirm and execute response operations based on resource allocation, and to continuously monitor and report the progress of the event and output response execution reports.

[0006] Furthermore, the specific process of extracting user event description information to form an event list is as follows: Feature extraction and event type identification are performed on user event description information. Based on the event type identification results, the terminal location information and timestamp in the user event description information are combined to generate a reported event. Verify whether the geographical location of the reported event is within the campus geographical boundary based on the terminal location information. Mark the reported events that fail the verification as invalid events and remove them. Perform duplicate checks on the reported events that pass the location verification and remove duplicate reported events. Compile all the reported events that pass the verification to form an event list.

[0007] Furthermore, the specific process for obtaining event clusters containing credibility scores is as follows: The event list is associated with the campus IoT device data corresponding to each event in time and space. Semantic key features are extracted from the user event description information corresponding to the reported event, and physical status indicators are extracted from the associated campus IoT device data. The correlation between semantic key features and physical status indicators is calculated to obtain the consistency score of each reported event. A reported event association matrix is ​​constructed based on the event list. Each event cluster and its association strength are obtained through the reported event association matrix. The consistency scores of each reported event in the event cluster are aggregated to obtain the overall consistency score of each event cluster. The overall consistency score of the event cluster is weighted and fused with the corresponding association strength to obtain the credibility score of each event cluster.

[0008] Furthermore, the specific process of obtaining each event cluster and the correlation strength of each event cluster through the reported event correlation matrix is ​​as follows: The reported events in the event list are combined to form reported event pairs; the timestamp difference and geographic location difference of each reported event pair are calculated to obtain the temporal proximity and geographic coupling; the similarity of the distribution of user event description information of the reported event pairs is calculated to obtain the semantic relevance. The temporal proximity, geographical coupling, and semantic relevance of each reported event pair are weighted and fused to obtain the comprehensive relevance strength of each reported event pair; A matrix is ​​constructed using each reported event in the event list as its row and column indices. The overall correlation strength of each reported event pair is used as the corresponding matrix element to obtain the reported event correlation matrix. Cluster analysis is performed based on the reported event association matrix to identify different event clusters, and an event cluster identifier is assigned to each event cluster. The ratio of the sum of all comprehensive association strengths within each event cluster to the total number of reported event pairs is calculated to obtain the association strength of each event cluster.

[0009] Furthermore, the specific process of constructing a hierarchical decision tree is as follows: A hierarchical decision tree is constructed, which includes first-level nodes divided according to the dominant event type, second-level nodes divided according to the credibility score interval, third-level nodes divided according to the merged impact range level, and leaf nodes with different situation levels. The nodes at each level are connected through corresponding feature judgment conditions.

[0010] Furthermore, the specific process for determining the situation level of each event cluster is as follows: The most frequent event type within each event cluster is identified to determine the dominant event type of the cluster; the geometric union area of ​​the geographical locations of all reported events within each event cluster is calculated as the merging influence range of the corresponding event cluster. Based on the dominant event type, credibility score, and traversal level decision tree of the combined impact range of each event cluster, the situation level of each event cluster is obtained.

[0011] Furthermore, the specific process of triggering the network resource scheduling procedure based on the status level of each event cluster is as follows: A pre-defined campus network policy table containing several key business groups is provided. Each key business group is associated with a set of status levels, dominant event types, and network resource requirement specifications. The campus network policy table is queried based on the status level and dominant event type of each event cluster to determine the key business group and network resource requirement specifications corresponding to each event cluster. The key business groups, their network resource requirement specifications, and event cluster identifiers are encapsulated to generate resource scheduling request commands.

[0012] Furthermore, the specific process for dynamically creating emergency network slice instances is as follows: Based on preset mapping rules, the network resource requirement specifications in the resource scheduling request command are mapped to slice configuration parameters, which include virtual network function configuration, bandwidth reservation configuration and quality of service policy configuration. Based on the slice configuration parameters, the network function chain is deployed and network resources are allocated globally to complete the creation of emergency network slice instances.

[0013] Furthermore, the specific process for confirming the completion of output resource allocation is as follows: Based on the network resource requirement specifications, the operational status of the created emergency network slice instances is verified, and the emergency network slice instances that pass the verification are marked as ready. Analyze the key business groups in the resource scheduling request command to obtain the specific resource requirement parameters of each key business group; Based on the specific resource requirements of each key business group, a resource allocation strategy is formulated in the emergency network slice instance in the ready state; the resource allocation strategy is converted into an executable configuration command; the emergency network slice instance allocates dedicated network resources to the key business groups based on the configuration command and generates a resource allocation completion confirmation.

[0014] Furthermore, the specific process of executing response actions and continuously monitoring and reporting event progress to output a response execution report is as follows: Once resource allocation is confirmed, activate all critical business applications located at the field terminals and command center; establish communication connections between the field terminals and command center through emergency network slicing instances, and execute emergency response operations. During the execution of emergency response operations, the system synchronously monitors the real event progress status corresponding to each event cluster and the running status of the emergency network slice instance, and collects relevant monitoring data; based on the monitoring data, it automatically generates response execution reports.

[0015] The technical effects and advantages of the IoT-based integrated management and control system for public safety in industrial parks according to this invention are as follows: 1. This invention preprocesses a large amount of scattered reported information by performing strict geographical boundary verification and duplicate detection on reported events, effectively filtering out invalid and redundant information. By constructing a correlation matrix of reported events and performing cluster analysis to form event clusters and credibility scores, and by comprehensively analyzing the correlation strength between the credibility scores and each event cluster, it achieves deep fusion and cross-validation of multi-source information. It can intelligently identify the authenticity of events and automatically aggregate multiple reports describing the same event, solving the problems of false alarms, missed alarms and alarm storms caused by the failure to correlate multi-source information. 2. This invention constructs a hierarchical decision tree comprising three levels of nodes: dominant event type, credibility score, and merged impact scope. It also pre-defines the status level for leaf nodes, enabling automated, objective, and standardized assessment of the severity and impact scope of event clusters. This provides a scientific and quantitative basis for the allocation of emergency response resources. Furthermore, by dynamically creating emergency network slice instances based on event status levels and according to the specific resource requirements of key business groups, it provides deterministic quality of service assurance for core emergency operations, ensuring the unimpeded flow of critical commands in complex network environments. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the IoT-based integrated management and control system for public safety in a park, as described in this invention. Figure 2 This is a schematic diagram of the process for obtaining a credibility score in this invention. Detailed Implementation

[0017] 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. Example

[0018] Please see Figure 1 and Figure 2 As shown in the figure, the main design contents of the IoT-based integrated management and control system for public safety in the park described in this embodiment are as follows: The existing university campus public safety management and control system adopts a three-layer architecture of front-end sensing terminal - campus data gateway - back-end alarm platform. Each reported public safety information is matched with alarm information and then processed as an independent alarm event. However, in the university's campus public safety management system, the public safety information received by the campus data gateway is an independent data unit. For example, public safety information A is "smoke at the entrance of Room 215 of the Experimental Building," and public safety information B is "smoke concentration exceeding the standard detected by the smoke sensor." The background alarm platform matches the keyword "smoke" in public safety information A with fire-related rules to generate an independent alarm. Similarly, it generates another independent alarm when the smoke concentration value in public safety information B exceeds the set threshold. The platform only performs isolated analysis on each information unit and does not establish a spatiotemporal correlation between public safety information A and public safety information B. It cannot identify that the two point to the same real event, which not only causes alarm redundancy and resource scheduling conflicts, but also increases the information screening and judgment burden of command personnel.

[0019] Based on this, an IoT-based integrated public safety management and control system for the park is designed, including: The data acquisition module is used to acquire user event description information and campus IoT device data in real time.

[0020] Users submit event descriptions via mobile devices, which include event text descriptions, terminal location information, and timestamps. Campus IoT device data is continuously collected and uploaded by various sensing devices deployed throughout the campus. These sensing devices include video surveillance cameras, access control systems, smoke sensors, temperature and humidity sensors, infrared detectors, people counting sensors, and emergency alarm buttons. Each type of sensing device generates data according to a preset sampling period or event triggering mechanism.

[0021] The event processing and verification module is used to extract user event description information to form an event list, perform correlation analysis between the event list and campus IoT data to obtain event clusters containing credibility scores, and analyze each event cluster by constructing a hierarchical decision tree to determine the status level of each event cluster. The specific process of extracting user event description information to form an event list is as follows: Feature extraction and event type identification are performed on user event description information. Based on the identified event type, the terminal location information and timestamp in the user event description information are combined to generate a reported event. Text preprocessing operations (removing punctuation, whitespace, and stop words, etc.) are performed on the event text description in the user event description information to obtain a text sequence. The text sequence is then segmented to extract keywords and entity relationships. The extracted keywords and entity relationships are compared with an event type keyword database to identify the event type and determine its type. For example, a user event description might include the event text "Smoke was seen coming out near the east gate of the library, possibly indicating a fire", the terminal location information "latitude and longitude coordinates (116.345, 39.987)", and the timestamp "2025-10-20 14:30:00". The event text description is segmented, and nouns and verbs are retained as key features, resulting in "'library' 'east gate' 'nearby' 'saw' 'smoke' 'emerging' 'fire'"; The keyword "fire" in the event type keyword library has the highest matching degree. The event type "fire" is used as the identification result. The event type "fire" in the identification result is combined with the terminal location information and timestamp in the user event description information to form a structured reporting event.

[0022] Verify whether the geographical location of the reported event is within the campus geographical boundary based on the terminal location information. Reported events that fail verification are marked as invalid and removed. Load the polygon coordinates of the campus geographical boundary. If the terminal location information point is not inside the campus geographical boundary polygon, the reported event record is marked as invalid and removed from the subsequent processing flow. Reported events that pass location verification are retained and continue to be processed.

[0023] Duplicate submissions are detected for reported events that have passed location verification, and duplicate submissions are removed. All verified reported events are compiled into an event list. Duplicate submissions are identified by comparing the timestamp differences, terminal location information similarity, and user event description text similarity among retained reported events. If the similarity of multiple reported events in time, location, and text description all exceeds a preset threshold, they are determined to be duplicate events. Identified duplicate submissions are removed, while non-duplicate submissions are retained. All verified reported events are then organized into an ordered set according to timestamp order, forming the event list.

[0024] For example, reporting event A: event type (fire); terminal location information ((116.345, 39.987) / east gate of the library); timestamp (2023-10-20 14:30:00); event text description (smoke was seen near the east gate of the library); Reported Event B: Event Type (Fire); Terminal Location Information ((116.3455, 39.9872) / Near the East Gate of the Library); Timestamp (2023-10-20 14:32:15); Event Text Description (Smoke is coming from the library area); Reported Event C: Event Type (Public Security Incident); Terminal Location Information (116.350, 39.985) / Gymnasium); Timestamp (2023-10-20 14:31); Event Text Description (Fighting in the Gymnasium); Through repeatability testing, reported event A and reported event B are highly similar in event type, location, and time, and are therefore judged as duplicate reported events describing the same event; reported event A and reported event C are different in event type and are both retained.

[0025] The specific process for obtaining event clusters containing credibility scores is as follows: The event list is associated with the corresponding campus IoT device data. For each reported event in the event list, the terminal location information and timestamp of the reported event are used to query the campus IoT device data. The data records of campus IoT devices whose geographical location is within a specific distance range (e.g., less than 50 meters) from the terminal location information of the reported event, and whose data reporting timestamp is within a specific time window (e.g., ±5 minutes) from the timestamp of the reported event, are then linked together to form a spatiotemporally corresponding data pair.

[0026] Semantic key features are extracted from the user event description information corresponding to the reported events, and physical status indicators are extracted from the associated campus IoT device data; nouns and verbs are retained in the event text description information of the user event description information, and words related to the event type are filtered out to form a list of semantic key features; for example, when the event type is "fire", words such as "smoke", "fire", and "burning" are extracted as semantic key features; Obtain device type and physical status readings from associated campus IoT devices; map the device type to the corresponding physical status index; for example, the device status parameter of a smoke sensor device type is mapped to the "smoke concentration" physical status index, and the device status parameter of a video surveillance camera device type is mapped to the "flame recognition status" physical status index; use the specific value or status identifier of the device status parameter as the physical status index value.

[0027] Calculate the correlation between semantic key features and physical state indicators to obtain the consistency score of each reported event; perform semantic matching judgment between each word in the semantic key feature list and the type or state value of the physical state indicator (e.g., "smoke" and "smoke concentration is higher than normal range"); if there is semantic correlation, it is considered a valid match; count the number of valid matches to obtain the matching ratio; use the matching ratio as the consistency score of the reported event, with a value range of [0,1].

[0028] A reported event association matrix is ​​constructed based on the event list. The association matrix is ​​used to obtain each event cluster and the association strength of each event cluster. The consistency scores of each reported event in the event cluster are aggregated to obtain the overall consistency score of each event cluster. For each event cluster, the consistency scores of all reported events in the event cluster are collected, and the arithmetic mean of these consistency scores is calculated as the overall consistency score of the event cluster.

[0029] The overall consistency score of an event cluster is weighted and fused with its corresponding association strength to obtain a credibility score for each event cluster. The credibility score reflects the authenticity and reliability of the event cluster. The overall consistency score and its corresponding association strength are normalized, and the credibility score for each event cluster is calculated using a weighted summation formula.

[0030] It should be explained that the weights of the overall consistency score and the correlation strength are initialized based on historical data and handling experience. The overall consistency score is direct evidence to verify the authenticity of user reports and serves as the dominant factor, while the correlation strength reflects the spatiotemporal and semantic closeness between reported events within an event cluster and is used to assist in correction. Therefore, the weight of the overall consistency score should be relatively high.

[0031] The specific process of obtaining each event cluster and the correlation strength of each event cluster through the reported event correlation matrix is ​​as follows: The reported events in the event list are combined to form reported event pairs. During the combination process, duplicate pairings are avoided (for example, if reported event A and reported event B have already formed a pair, then no further pairing of reported event B and reported event A will be generated). The timestamp difference and geographic location difference of each reported event pair are calculated to obtain temporal proximity and geographic coupling. Temporal proximity is used to represent the closeness of two reported events in the time dimension, and the timestamp difference is the absolute difference between the timestamps of the two reported events. It is obtained by "temporal proximity = 1 ÷ (1 + timestamp difference)". The smaller the time difference, the higher the temporal proximity.

[0032] Geographic coupling degree is used to represent the degree of coupling between two reported events in the spatial dimension. The geographical difference is the Euclidean distance between the terminal location information of the two reported events. For each pair of reported events, the terminal location information of the two reported events is extracted, and the geographical distance between the two terminal location information is calculated. The geographic coupling degree is obtained by the formula "geographic coupling degree = 1 ÷ (1 + geographical distance)"; where the smaller the geographical distance, the higher the geographic coupling degree.

[0033] Calculate the similarity of user event description information distribution in reported event pairs to obtain semantic relevance. Semantic relevance is used to represent the semantic similarity of user event description information between two reported events. Extract event text descriptions from user event description information in reported event pairs. Preprocess the event text descriptions (word segmentation and stop word removal, etc.) and extract keywords. Calculate text similarity based on the degree of keyword overlap and use it as semantic relevance. The semantic relevance value range is [0,1]. The higher the semantic relevance value, the higher the degree of semantic similarity.

[0034] We perform weighted fusion of temporal proximity, geographical coupling, and semantic relevance of each reported event pair to obtain the comprehensive association strength of each reported event pair. The comprehensive association strength represents the overall degree of association between a pair of reported events in the three dimensions of time, space, and semantics. The higher the value, the more likely the two reported events are to belong to the same real event.

[0035] It should be explained that the weights assigned to temporal proximity, geographical coupling, and semantic relevance are dynamically allocated based on the event type of the reported event. For example, for a "fire" event, the spread of the fire is strongly correlated with the physical location, so geographical coupling has a high weight; while for a "rumor" event, the spread and evolution of the quoted text are strongly correlated, so hidden semantic relevance has a high weight.

[0036] A matrix is ​​constructed using each reported event in the event list as its row and column indices. The overall correlation strength of each reported event pair is used as the corresponding matrix element to obtain the reported event correlation matrix. It should be explained that if two reported events do not form a valid pair (e.g., themselves and themselves), the corresponding matrix element is assigned a value of 0.

[0037] Cluster analysis is performed based on the reported event association matrix to identify different event clusters, and an event cluster identifier is assigned to each event cluster. The comprehensive association strength in the reported event association matrix is ​​converted into a distance metric. Each reported event is initialized as a separate event cluster for iteration. In each iteration, the average link distance between all event clusters is calculated, and the two event clusters with the smallest average link distance are merged. The merging process is repeated until the minimum average link distance between all event clusters exceeds a predetermined clustering distance threshold. The resulting clusters are the identified different event clusters.

[0038] The correlation strength of each event cluster is obtained by calculating the ratio of the sum of all overall correlation strengths within each event cluster to the total number of reported event pairs. The correlation strength reflects the overall structural tightness among the reported events within an event cluster.

[0039] The specific process of constructing a hierarchical decision tree is as follows: Construct a hierarchical decision tree, a hierarchical tree structure, to classify event clusters in multiple dimensions and ultimately determine their status level.

[0040] The hierarchical decision tree includes first-level nodes divided according to the dominant event type. Each dominant event type corresponds to a first-level node branch. The branch path is triggered by the event type matching condition. The corresponding branch path is entered only when the dominant event type of the event cluster matches the branch condition.

[0041] The second-level node is based on the confidence score interval division. The second-level node is used to divide the event cluster according to the confidence score interval. The confidence score interval is a predefined set of numerical intervals covering the range [0,1].

[0042] For example, the credibility score intervals are: low credibility score interval [0, 0.3), medium credibility score interval [0.3, 0.7), and high credibility score interval [0.7, 1.0]. Each credibility score interval corresponds to a second-level node branch. The judgment process compares the credibility score of the event cluster with each interval, determines the interval to which it belongs, and enters the corresponding branch path.

[0043] The third-level node is based on the classification of the scope of impact of the event cluster. The third-level node is used to classify the scope of impact of the event cluster according to its classification level. For example, the scope of impact of the event cluster can be divided into three levels: small (area < 500 square meters), medium (500 square meters ≤ area ≤ 2000 square meters), and large (area > 2000 square meters). Each scope of impact of the event cluster corresponds to a third-level node branch. The judgment process compares the scope of impact of the event cluster with the threshold of each level to determine its level and enters the corresponding branch.

[0044] The system includes leaf nodes with different risk levels, and nodes at each level are connected by corresponding feature judgment conditions. Leaf nodes are the endpoints of the hierarchical decision tree inference path, representing the final classification result. Each leaf node has a risk level, which includes four states: "low risk," "medium risk," "high risk," and "emergency."

[0045] The specific process for determining the situation level of each event cluster is as follows: Analyze the most frequent event types within each event cluster to determine the dominant event type of the cluster; count the number of times each event type appears in the event cluster to form an event type frequency list; and find the most frequent event type from the event type frequency list.

[0046] It should be explained that if there are multiple event types with the same highest frequency within a certain event cluster, the event type corresponding to the earliest reported event with the earliest timestamp will be selected as the dominant event type of that event cluster.

[0047] Calculate the geometric union area of ​​the geographic locations of all reported events within each event cluster, and use it as the merged influence range of the corresponding event cluster. The merged influence range is a quantified value of the size of the spatial area covered by the event cluster. Using the terminal location information of all reported events within the event cluster as input data, the Graham scan algorithm is used to calculate the convex hull of these coordinate points, generating the smallest convex polygon that encloses all points. The area of ​​this convex polygon is used as the geometric union area. The geometric union area is used as the merged influence range of the event cluster.

[0048] Based on the dominant event type, credibility score, and merged impact range of each event cluster, the tiered decision tree is traversed to obtain the situational level of each event cluster. The traversal process starts from the root node of the tiered decision tree, matches the corresponding branch in the first-level node according to the dominant event type, and enters the branch path; selects the corresponding branch in the second-level node according to the interval where the credibility score falls; selects the corresponding branch in the third-level node according to the level to which the merged impact range belongs; finally, a leaf node is reached, and the situational level of the leaf node is the situational level of the event cluster.

[0049] The resource scheduling and support module is used to trigger the network resource scheduling process based on the status level of each event cluster, dynamically create emergency network slice instances based on the generated resource scheduling request commands, allocate dedicated network resources based on the emergency network slice instances, and output a confirmation of resource allocation completion.

[0050] The specific process of triggering network resource scheduling based on the status level of each event cluster is as follows: The pre-set campus network policy table contains several key business groups. Each key business group is associated with a set of status levels, dominant event types, and network resource requirements. The campus network policy table contains several policy records, and each policy record corresponds to a key business group (such as "video surveillance feedback", "emergency voice dispatch", "personnel location tracking", "alarm information push", and "command center large screen display").

[0051] Each policy record contains four fields: the associated critical business group, the allowed status level, the allowed dominant event type, and the corresponding network resource requirement specification. The network resource requirement specification represents the set of technical parameters of the network service capabilities required by the critical business group during operation. Network resource requirements include bandwidth requirements, latency requirements, and reliability requirements.

[0052] Based on the status level and dominant event type of each event cluster, the campus network policy table is queried to determine the key business groups and network resource requirement specifications corresponding to each event cluster. For each event cluster with a determined status level, using the status level and dominant event type of the event cluster as query conditions, each policy record in the campus network policy table is traversed sequentially to determine whether the status level in the policy record is completely consistent with the status level of the event cluster, and whether the dominant event type in the policy record is completely consistent with the dominant event type of the event cluster. When both conditions are met, the policy record is considered to be successfully matched, and its associated key business groups and network resource requirement specifications are extracted.

[0053] The critical business groups, their network resource requirements, and event cluster identifiers are encapsulated to generate a resource scheduling request command. This command contains three fields: an event cluster identifier field, a critical business group field, and a network resource requirement specification field. The obtained critical business group, network resource requirement specification, and event cluster identifier are then filled into the corresponding fields.

[0054] The specific process for dynamically creating emergency network slice instances is as follows: Based on preset mapping rules, the network resource requirement specifications in the resource scheduling request command are mapped to slice configuration parameters. The slice configuration parameters include virtual network function configuration, bandwidth reservation configuration, and quality of service policy configuration. The mapping rules contain the correspondence between each network resource requirement specification and the slice configuration parameters.

[0055] Virtual network function configuration is used to determine the types of virtual network functions that need to be deployed in the emergency network slice instance and their chain order; bandwidth reservation configuration is used to set the bandwidth allocation value on the end-to-end path according to the minimum bandwidth requirement; service quality policy configuration is used to generate differential service code point markers, queue scheduling policies and traffic shaping parameters according to the maximum allowable latency, the maximum allowable packet loss rate and the service quality priority level.

[0056] Based on the various requirements in the network resource requirements specification, preset mapping rules are used to convert them one by one. Reliability requirements are converted into virtual network function configurations, bandwidth requirements are converted into bandwidth reservation configurations, and latency and reliability requirements are converted into quality of service policy configurations, forming complete slice configuration parameters.

[0057] For example, the network resource requirement specification in a resource scheduling request command is: Bandwidth requirement: 100 Mbps; Latency requirement: 50 ms; Reliability requirement: 99.9%; Reliability requirements are translated into "deploying virtual network function components with redundant backup capabilities on different physical servers"; bandwidth requirements are translated into "allocating asymmetric guaranteed bandwidth according to business traffic characteristics"; and latency requirements are mapped to "configuring high-priority queues and setting latency thresholds".

[0058] The generated slice configuration parameters are as follows: Virtual network function configuration (instantiate two virtualized user plane function components and configure them as primary and backup modes; instantiate one virtualized firewall component); Bandwidth reservation configuration (uplink guaranteed bandwidth: 20 Mbps, downlink guaranteed bandwidth: 100 Mbps); Quality of Service policy configuration (service level identifier is set to critical task instruction level, scheduling policy adopts strict priority queue with bandwidth limit, maximum latency 50 ms; maximum jitter 10 ms and maximum packet loss rate 0.001%).

[0059] Based on the slice configuration parameters, the deployment of network function chains and the global allocation of network resources are executed to complete the creation of emergency network slice instances. Virtual network function components are instantiated on distributed computing nodes, and logical connections between components are established according to the service chain sequence defined in the virtual network function configuration, forming a complete data processing path. Global allocation of network resources involves updating the forwarding rules of switches and routers through the software-defined network controller, allocating independent label switching paths or tunnel identifiers to emergency network slice instances. Simultaneously, the scheduling rules in the Quality of Service (QoS) policy configuration are applied to network edge devices to complete the creation of emergency network slice instances.

[0060] The specific process for confirming the completion of output resource allocation is as follows: The operational status of the created emergency network slice instance is verified based on the network resource requirement specifications. Test traffic is injected into the emergency network slice instance, and performance probing tools are used to measure the actual end-to-end bandwidth, end-to-end latency, and packet loss rate. The measured bandwidth values ​​are compared with the bandwidth requirements, the measured latency values ​​are compared with the latency requirements, and the link availability and reliability requirements calculated based on the packet loss rate are compared. When all measured values ​​meet or exceed the requirements of the network resource requirement specifications, the emergency network slice instance passes the operational status verification.

[0061] The emergency network slice instance that has passed verification will be marked as ready, and the value of the status field of the emergency network slice instance will be updated from "Creating" to "Ready".

[0062] Analyze the key business groups in the resource scheduling request command to obtain the specific resource requirement parameters for each key business group. Based on the type of key business group, obtain the specific resource requirement parameters for the corresponding key business group (required guaranteed bandwidth, maximum latency, maximum allowed concurrent connections, priority flag value in the quality of service policy, data encryption requirements, and transmission protocol type).

[0063] For example, for the "video surveillance backhaul" critical business group, the specific resource requirements parameters include a guaranteed bandwidth of no less than 20Mbps, a maximum latency of no more than 100ms, and a service quality policy priority marked as EF (accelerated forwarding).

[0064] Based on the specific resource requirements of each key business group, a resource allocation strategy is formulated in the emergency network slice instance in the ready state; an independent logical channel is allocated for each key business group within the emergency network slice instance, and a unique virtual LAN identifier is assigned to each logical channel; according to the guaranteed bandwidth in the specific resource requirements parameters, a corresponding minimum guaranteed bandwidth is allocated to each logical channel; and according to the maximum latency, a corresponding service priority and queue scheduling strategy are set for each logical channel.

[0065] The resource allocation strategy is translated into executable configuration commands; the executable configuration commands are represented as protocol data instructions that can be recognized by the target network device.

[0066] The emergency network slice instance allocates dedicated network resources to critical business groups based on configuration commands and generates a resource allocation completion confirmation. The generated configuration commands are distributed to the virtual network function components of the emergency network slice instance through the management channel. The virtual network function components execute the received configuration commands, create corresponding virtual LAN identifier mappings for each critical business group on the data plane, implement traffic policing to ensure bandwidth, and perform queue scheduling according to the set business priorities. After all configuration commands have been executed and the resource allocation results have been verified, a resource allocation completion confirmation containing an event cluster identifier and a resource readiness timestamp is generated.

[0067] The emergency response and monitoring module is used to confirm and execute response operations based on resource allocation, and to continuously monitor and report the progress of the event and output response execution reports.

[0068] The specific process of executing response actions, continuously monitoring and reporting event progress, and outputting a response execution report is as follows: Based on the confirmation of resource allocation completion, various key business applications located at the field terminals and the command center are activated; the field terminals are mobile or fixed terminals deployed in the area where the incident occurred; the command center is the workstation within the campus security command and dispatch area; based on the confirmation message of resource allocation completion, various key business applications (such as video surveillance applications and alarm notification applications) are activated.

[0069] Communication connections between field terminals and the command center are established through emergency network slicing instances, and emergency response operations are executed. The specific behavior patterns of the emergency response operations are determined based on the dominant event type and situation level of the event cluster. For example, when the dominant event type is "fire", the emergency response operations include automatically retrieving video surveillance footage of the incident area, pushing the alarm location to security personnel, and activating the broadcast system to play evacuation notices. When the dominant event type is "fighting", the emergency response operations include activating surrounding cameras to track and record, notifying the nearest patrol personnel to rush to the scene, and recording the content of voice intercom communications. The log information of all emergency response operations is recorded in real time in the event handling log storage area.

[0070] During the synchronous monitoring of emergency response operations, the system tracks the real-world event progress status of each event cluster and the operational status of the emergency network slice instances, collecting relevant monitoring data. The real-world event progress status reflects the actual handling situation at the event site. The monitoring data includes: reading location update information uploaded by on-site terminals, receiving text or voice feedback submitted by security personnel, and parsing the behavior recognition results output by the video analysis module; the operating status of emergency network slice instances reflects the performance of the network services they carry; bandwidth utilization, traffic statistics, and monitoring the trends of end-to-end latency and packet loss rate.

[0071] Response execution reports are automatically generated based on monitoring data. The incident handling result summary includes the incident start time, response initiation time, police arrival time, description of on-site handling actions, and final status determination; network resource usage statistics include peak bandwidth usage, average latency, cumulative packet loss, and service quality policy compliance rate; the response timeline records include the incident reporting time, incident cluster generation time, resource scheduling request command generation time, emergency network slice instance creation completion time, resource allocation completion confirmation generation time, and the execution time of each emergency response operation.

[0072] In this embodiment, by performing strict geographical boundary verification and duplicate detection on reported events, preprocessing of a large amount of scattered reported information is achieved, effectively filtering out invalid and redundant information. By constructing a reported event association matrix and performing cluster analysis to form event clusters and credibility scores, and comprehensively analyzing the correlation strength between the credibility scores and each event cluster, deep fusion and cross-validation of multi-source information is achieved. This enables intelligent identification of event authenticity and automatic aggregation of multiple reports describing the same event, solving the problems of false alarms, missed alarms, and alarm storms caused by the failure to associate multi-source information.

[0073] By constructing a hierarchical decision tree comprising three levels of nodes—dominant event type, credibility score, and merged impact scope—and pre-setting status levels for leaf nodes, the system achieves automated, objective, and standardized assessment of the severity and impact scope of event clusters. This provides a scientific and quantitative basis for the allocation of emergency response resources. Furthermore, by dynamically creating emergency network slice instances based on event status levels and according to the specific resource requirements of key business groups, the system provides deterministic quality of service assurance for core emergency operations, ensuring the unimpeded flow of critical commands in complex network environments.

[0074] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0075] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only one method, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0076] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

[0077] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An IoT-based park public safety integrated management and control system, characterized in that, The campus public safety integrated management and control system based on the Internet of Things comprises: a data acquisition module, configured to acquire user event description information and campus Internet of Things device data in real time; an event processing and verification module, configured to extract the user event description information to form an event list, associate and analyze the event list and the campus Internet of Things data, obtain an event cluster containing a credibility score, analyze each event cluster by constructing a hierarchical decision tree, and determine a situation level of each event cluster; a resource scheduling and guarantee module, configured to trigger a network resource scheduling process according to the situation level of each event cluster, dynamically create an emergency network slice instance based on a generated resource scheduling request command, allocate exclusive network resources based on the emergency network slice instance, and output a resource allocation completion confirmation; an emergency response and monitoring module, configured to execute a response operation based on the resource allocation completion confirmation, and continuously monitor the progress of the reported event and output a response execution report. 2.The IoT-based park public safety integrated management and control system according to claim 1, characterized in that, The specific process of extracting the user event description information to form the event list is as follows: feature extraction and event type identification are performed on the user event description information, the event type based on the identification result is combined with terminal location information and a timestamp in the user event description information, and a reported event is generated; whether the geographical position of the reported event is located within the campus geographical boundary is verified based on the terminal location information, and the reported event that fails to pass the verification is marked as an invalid event and is removed; repeated detection is performed on the reported event that passes the location verification, the reported event that is repeatedly submitted is removed, and the reported event that passes all the verifications is compiled to form the event list. 3.The garden public safety integrated management and control system based on Internet of Things according to claim 2, characterized in that, The specific process of obtaining the event cluster containing the credibility score is as follows: each reported event in the event list is associated with the campus Internet of Things device data corresponding to the time and space, semantic key features are extracted from the user event description information corresponding to the reported event, and physical state indicators are extracted from the associated campus Internet of Things device data; the consistency score of each reported event is obtained by calculating the correlation between the semantic key features and the physical state indicators; a reported event association matrix is constructed based on the event list, each event cluster and the association strength of each event cluster are obtained through the reported event association matrix, the consistency scores of each reported event in the event cluster are aggregated to obtain the overall consistency score of each event cluster, and the overall consistency score of the event cluster and the corresponding association strength are weighted and fused to obtain the credibility score of each event cluster. 4.The garden public safety integrated management and control system based on Internet of Things according to claim 3, characterized in that, The specific process of obtaining each event cluster and the association strength of each event cluster through the reported event association matrix is as follows: each reported event in the event list is combined to form a reported event pair, the timestamp difference and the geographical position difference of each reported event pair are calculated to obtain the time sequence proximity and the geographical coupling degree; the semantic association degree is obtained by calculating the distribution similarity of the user event description information of the reported event pair; the time sequence proximity, the geographical coupling degree and the semantic association degree of each reported event pair are weighted and fused to obtain the comprehensive association strength of each reported event pair; a matrix is constructed with each reported event in the event list as a row and a column index, and the comprehensive association strength of each reported event pair is taken as a corresponding matrix element to obtain the reported event association matrix. Based on the reported event association matrix, clustering analysis is performed to identify different event clusters, and an event cluster identifier is assigned to each event cluster; the sum of the comprehensive correlation strengths of all events in each event cluster is calculated, and the ratio of the sum to the total number of reported events is obtained to obtain the correlation strength of each event cluster. 5.The IOT-based park public safety integrated management and control system according to claim 1, characterized in that, The specific process of constructing the hierarchical decision tree is as follows: The hierarchical decision tree is constructed, which includes a first-level node divided according to the dominant event type, a second-level node divided based on the credibility score interval, a third-level node divided based on the merged influence range, and a leaf node provided with different situation levels, and each level node is connected through the corresponding feature judgment condition. 6.The garden public safety integrated management and control system based on Internet of Things according to claim 5, characterized in that, The specific process of determining the situation level of each event cluster is as follows: The event type with the highest frequency in each event cluster is counted to determine the dominant event type of the event cluster; the geometric union area of the geographic positions of all reported events in each event cluster is calculated as the merged influence range of the corresponding event cluster; Based on the dominant event type, credibility score, and merged influence range of each event cluster, the hierarchical decision tree is traversed to obtain the situation level of each event cluster. 7.The garden public safety integrated management and control system based on Internet of Things according to claim 6, characterized in that, The specific process of triggering the network resource scheduling process according to the situation level of each event cluster is as follows: A campus network policy table containing several key service groups is preset, each key service group is associated with a group of situation levels, dominant event types, and network resource demand specifications; based on the situation level and dominant event type of each event cluster, the campus network policy table is queried to determine the corresponding key service group and network resource demand specification of each event cluster; the key service group and its network resource demand specification and event cluster identifier are encapsulated to generate a resource scheduling request command. 8.The garden public safety integrated management and control system based on Internet of Things according to claim 7, characterized in that, The specific process of dynamically creating an emergency network slice instance is as follows: Based on the preset mapping rule, the network resource demand specification in the resource scheduling request command is mapped to the slice configuration parameter, which includes virtual network function configuration, bandwidth reservation configuration, and quality of service policy configuration; Based on the slice configuration parameter, the deployment of the network function chain and the global allocation of network resources are executed to complete the creation of the emergency network slice instance. 9.The IOT-based park public safety integrated management and control system according to claim 8, characterized in that, The specific process of outputting the resource allocation completion confirmation is as follows: Based on the network resource demand specification, the running state of the created emergency network slice instance is verified, and the emergency network slice instance that passes the verification is marked as ready state; The key service groups in the resource scheduling request command are analyzed to obtain the specific resource demand parameters of each key service group; Based on each key service group and the corresponding specific resource demand parameter, a resource allocation strategy is formulated in the emergency network slice instance in the ready state; the resource allocation strategy is converted into an executable configuration command; the emergency network slice instance allocates exclusive network resources to the key service group based on the configuration command, and generates a resource allocation completion confirmation. 10.The IOT-based park public safety integrated management and control system according to claim 9, characterized in that, The specific process of executing the response operation and continuously monitoring the progress of the reported event to output a response execution report is as follows: Based on the resource allocation completion confirmation, start various key business applications located at the on-site terminal and the command center; establish a communication connection between the on-site terminal and the command center through the emergency network slice instance, and execute the emergency response operation; Synchronous monitoring of the progress state of the real event corresponding to each event cluster and the running state of the emergency network slice instance during the execution of the emergency response operation, and collecting relevant monitoring data; Automatic generation of a response execution report based on the monitoring data.

Images