Pedestrian flow management method, system and device and storage medium

Abstract

The invention discloses a visitor flow rate management method, system and device and a storage medium, and relates to the technical field of communication, and the method comprises the steps: collecting a control plane signaling flow between a core network of a 5G private network and a base station in a sliding time window through responding to the updating of the sliding time window; extracting real-time signaling behavior characteristics of a user according to the control plane signaling flow; the deviation between the real-time signaling behavior characteristics of the user and a behavior baseline is calculated, and people flow abnormity is identified according to the deviation; the behavior baseline is dynamically updated according to historical signaling behavior characteristics of the user. By adopting the embodiment of the invention, the abnormal people flow can be identified with low time delay and high reliability by utilizing the localized signaling closed-loop characteristic of the 5G private network.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method, system, device and storage medium for managing pedestrian traffic in 5G private networks. Background Technology

[0002] Currently, some solutions for detecting and controlling abnormal situations in crowds collect communication behavior data from user terminals via the public network and combine this data with big data platforms to analyze crowd gatherings or individuals exhibiting unusual behavior.

[0003] However, public network communication transmission not only suffers from latency, making it difficult to meet the requirements of second-level detection; but also faces the risk of privacy leaks in data transmission under public network environment, lacking industry-specific security isolation. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method, system, device and storage medium for managing pedestrian flow, which can be applied in 5G private networks and can utilize the localized signaling closed-loop characteristics of 5G private networks to identify abnormal pedestrian flow with low latency and high reliability.

[0005] In a first aspect, embodiments of the present invention provide a method for managing pedestrian flow, applied in a 5G private network, the method comprising: In response to the sliding time window update, the control plane signaling flow between the core network and base station of the 5G private network within the sliding time window is collected; Based on the control plane signaling flow, extract the user's real-time signaling behavior characteristics; The deviation between the user's real-time signaling behavior characteristics and the behavior baseline is calculated, and abnormal crowd flow is identified based on the deviation; the behavior baseline is dynamically updated based on the user's historical signaling behavior characteristics.

[0006] As an improvement to the above scheme, the step of responding to the sliding time window update and collecting the control plane signaling flow between the core network and the base station of the 5G private network within the sliding time window includes: In response to the sliding time window update, the system collects RRC signaling between the terminal and the base station, NAS signaling between the terminal and the core network of the 5G private network, and NGAP signaling between the base station and the core network within the sliding time window. Based on the timing relationship between the RRC signaling, the NAS signaling, and the NGAP signaling, the control plane signaling flow is obtained.

[0007] As an improvement to the above scheme, if the real-time signaling behavior features include registration features, then the step of extracting the user's real-time signaling behavior features based on the control plane signaling flow includes: Based on the control plane signaling flow, obtain the distribution of registration reason types for terminal access to the 5G private network; Compute the registration success rate and average registration interval of terminals accessing the 5G private network; The user's registration characteristics are obtained based on the distribution of registration reason types, the registration success rate, and the average registration interval.

[0008] As an improvement to the above scheme, the real-time signaling behavior features include spatial transformation features. Therefore, the step of extracting the user's real-time signaling behavior features based on the control plane signaling flow includes: Based on the control plane signaling flow, the frequency of terminal handover to base station, handover failure rate, and average dwell time of the terminal at the base station are obtained; The user's switching characteristics are obtained based on the frequency, the switching failure rate, and the average dwell time. Based on the control plane signaling flow, the number of tracking area updates and the frequency of change of the terminal in the core network of the 5G private network are obtained to obtain the user's mobility characteristics. Based on the switching characteristics and the movement characteristics, the user's spatial transformation characteristics are obtained.

[0009] As an improvement to the above scheme, the real-time signaling behavior characteristics include service activity characteristics. Therefore, the step of extracting the user's real-time signaling behavior characteristics based on the control plane signaling flow includes: Based on the control plane signaling flow, obtain the number of sessions established by the terminal in the 5G private network, the average duration of the sessions, and the reasons for the sessions released. Based on the number of sessions established, the average duration of each session, and the reason for session release, the user's business activity characteristics are obtained.

[0010] As an improvement to the above scheme, the real-time signaling behavior characteristics include network quality characteristics. Therefore, the step of extracting the user's real-time signaling behavior characteristics based on the control plane signaling flow includes: Based on the control plane signaling flow, the reference signal sent from the base station to the terminal and the RRC connection hold time between the terminal and the base station are collected. Based on the reference signal, calculate a first average value and a first fluctuation rate of the received power of the reference signal, and calculate a second average value and a second fluctuation rate of the received quality of the reference signal; The user's network quality characteristics are obtained based on the first average value, the first volatility, the second average value, the second volatility, and the RRC connection hold time.

[0011] As an improvement to the above scheme, the real-time signaling behavior characteristics include abnormal signaling ratio characteristics. Therefore, the step of extracting the user's real-time signaling behavior characteristics based on the control plane signaling flow includes: Based on the control plane signaling flow, calculate the rejection signaling ratio and the abnormal code ratio in the total signaling; The user's abnormal signaling ratio characteristic is obtained based on the rejection signaling ratio and the abnormal code ratio.

[0012] As an improvement to the above scheme, the step of calculating the deviation of the user's real-time signaling behavior characteristics from the behavior baseline, and identifying abnormal crowd flow based on the deviation, includes: Based on the user's demographic tags, determine the target demographic category mentioned by the user, and obtain the typical characteristics of each subcategory within the target demographic category; Calculate the first deviation value between the real-time signaling behavior characteristics and each of the typical characteristics; Calculate the second deviation between each real-time signaling behavior feature and the mean of the user behavior baseline under the corresponding feature; Based on the first deviation value and the second deviation value, abnormal crowd flow is identified.

[0013] As an improvement to the above scheme, methods for extracting typical features corresponding to population categories include: Based on the user's demographic category label, the historical signaling behavior features are hierarchically aggregated to obtain the category features corresponding to each demographic category; Principal component analysis is used to embed and map the category features to obtain low-dimensional behavior vectors; The low-dimensional behavior vectors are clustered to obtain groups with similar behaviors, which are then used as subcategories. Calculate the central prototype vector of the behaviorally similar groups, and use the central prototype vector as the typical feature of the subdivision category.

[0014] As an improvement to the above scheme, the method for constructing the behavioral baseline includes: Collect historical control plane signaling streams of users within a preset continuous time period, and extract historical signaling behavior features based on the historical control plane signaling streams; Calculate the mean and variance of the historical signaling behavior characteristics; Based on the mean and variance, a baseline of user behavior is constructed for each user.

[0015] As an improvement to the above scheme, the calculation of the mean and variance of the historical signaling behavior characteristics includes: Using statistical methods, the first mean and first variance of each of the aforementioned historical signaling behavior characteristics are calculated; Extract the time tags of the historical signaling behavior features, and obtain the feature time series data based on the time tags and the historical signaling behavior features; The feature time series data is input into a pre-trained recurrent neural network, and the periodic patterns within a preset continuous time period are captured by predicting signaling behavior to obtain the second mean and the second variance. Based on a preset confidence level, the first mean and the second mean are weighted and fused, and the first variance and the second variance are weighted and fused to obtain the mean and variance of historical signaling behavior features, respectively; the confidence level is the prediction error of signaling behavior.

[0016] Secondly, embodiments of the present invention provide a people flow management system applied in a 5G private network, the people flow management system comprising: The data acquisition module is used to collect the control plane signaling flow between the core network and the base station of the 5G private network within the sliding time window in response to the sliding time window update. The feature extraction module is used to extract real-time signaling behavior features of users based on the control plane signaling flow; An anomaly identification module is used to calculate the deviation between the user's real-time signaling behavior characteristics and the behavior baseline, and to identify abnormal crowd flow based on the deviation; the behavior baseline is dynamically updated based on the user's historical signaling behavior characteristics.

[0017] Thirdly, embodiments of the present invention provide a crowd flow management device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the crowd flow management method as described above.

[0018] Fourthly, embodiments of the present invention provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the crowd flow management method as described above.

[0019] Compared with existing technologies, the present invention discloses a method, system, device, and storage medium for managing pedestrian flow. This method collects control plane signaling flow between the core network and base station of a 5G private network within a sliding time window in response to the window's update. Based on the control plane signaling flow, it extracts real-time signaling behavior characteristics of users. It calculates the deviation between these characteristics and a behavioral baseline, and identifies abnormal pedestrian flow based on this deviation. The behavioral baseline is dynamically updated based on the user's historical signaling behavior characteristics. By employing embodiments of the present invention, the localized signaling closed-loop characteristics of the 5G private network can be utilized to identify abnormal pedestrian flow with low latency and high reliability. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the steps of a pedestrian flow management method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a people flow management system provided in an embodiment of the present invention; Figure 3This is a schematic diagram of the structure of a people flow management device provided in an embodiment 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] In the description and claims, it should be understood that the terms "first," "second," etc., used in the description and claims are only for the purpose of distinguishing the description of the same technical features, and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated, nor necessarily the order of description or chronological order. The terms are interchangeable where appropriate. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature.

[0023] Currently in the transportation sector, the detection and control of abnormal crowd situations generally rely on traffic security personnel to identify and handle abnormal crowds through on-site patrols or by receiving backend early warning information. The backend early warning information comes from crowd analysis of video stream data, which needs to be transmitted over the public network.

[0024] However, public network communication suffers from latency, and video transmission and cloud analysis also incur delays. Furthermore, public networks can experience congestion in large-scale crowd scenarios, potentially leading to the loss or delay of critical anomaly information, making it difficult to meet the second-level detection requirements in traffic scenarios. More importantly, data transmission over public networks carries the risk of privacy breaches.

[0025] Based on the above considerations, this invention provides a method for managing pedestrian flow in a 5G private network. It should be noted that this invention constructs a 5G private network in the target scenario, and performs pedestrian flow analysis securely and efficiently using signaling data flowing within the private network.

[0026] Please see Figure 1 In this embodiment, the pedestrian flow management method is specifically executed through steps S1 to S4: S1. In response to the sliding time window update, collect the control plane signaling flow between the core network and the base station of the 5G private network within the sliding time window.

[0027] It should be noted that the control plane signaling flow is control command data transmitted between the core network and the base station in the 5G private network, rather than service data directly used by users. It is mainly used to manage network interaction processes such as terminal access, session establishment, and mobility control.

[0028] The flow of people is characterized by real-time changes. In this embodiment of the invention, a sliding time window is used for dynamic updating to ensure that the signaling flow collected each time is the latest time interval, providing accuracy in the time dimension for subsequent real-time analysis. Furthermore, the control plane signaling flow can directly reflect the interaction process between the terminal and the network, and can indirectly correlate the user's spatial location and flow status.

[0029] S2. Extract the user's real-time signaling behavior features based on the control plane signaling flow.

[0030] Control plane signaling flow is raw, unprocessed data that cannot be directly used for people flow analysis. By extracting real-time signaling behavior features, signaling data can be directly correlated with user behavior, transforming disordered signaling flow into analyzable and comparable key indicators, thereby improving the accuracy of people flow analysis.

[0031] S3. Calculate the deviation between the user's real-time signaling behavior characteristics and the behavior baseline, and identify abnormal crowd flow based on the deviation; the behavior baseline is dynamically updated based on the user's historical signaling behavior characteristics. Different users have different behavioral habits, and anomalies cannot be directly determined solely by real-time signaling behavior characteristics. It is necessary to compare these characteristics with the user's normal behavioral baseline to avoid misjudgments caused by differences in user behavior and ensure the specificity of deviation assessments. Furthermore, the behavioral baseline is dynamically updated to adapt to long-term changes in user behavior, further improving the accuracy of deviation assessments.

[0032] Unlike existing technologies that analyze video or image data, this invention relies solely on the existing core network and base station equipment of the 5G private network, requiring no additional hardware and is unaffected by environmental factors such as lighting or obstruction. Furthermore, unlike solutions that obtain terminal data through public network platforms, this invention can achieve the subdivision of micro-areas and adapt to personalized pedestrian flow management needs within a given scenario.

[0033] In the above solution, relying on the localized deployment characteristics of 5G private networks, signaling data does not need to rely on public network transmission, which can ensure the stability of data collection and analysis. At the same time, the dynamic baseline can adapt to the behavioral differences of different users and is suitable for traffic management in various 5G private network scenarios.

[0034] As a preferred implementation, step S1, in response to the sliding time window update, involves collecting the control plane signaling flow between the core network and the base station of the 5G private network within the sliding time window, including: In response to the sliding time window update, the system collects RRC signaling between the terminal and the base station, NAS signaling between the terminal and the core network of the 5G private network, and NGAP signaling between the base station and the core network within the sliding time window. Based on the timing relationship between the RRC signaling, the NAS signaling, and the NGAP signaling, the control plane signaling flow is obtained.

[0035] In some preferred embodiments, control plane signaling streams are collected using mirrored traffic or interface replication, and local windowed caching is performed on edge nodes to facilitate subsequent data retrieval or analysis. More preferably, the window size of the windowed cache is consistent with the length of the sliding time window; in other embodiments, the windowed cache uses the number of records as the caching standard.

[0036] The sliding time window is updated according to a preset step size. Each update is a trigger event that automatically switches the data collection time interval from the previous window period to the current window period, ensuring that each collected control plane signaling stream corresponds to the latest fixed duration interval, providing a unified time reference for the timing association of subsequent signaling streams.

[0037] RRC (Radio Resource Control) signaling is a type of control signaling transmitted over the air interface between a terminal and a 5G private network base station. It is mainly used to manage the wireless connection between the terminal and the base station, such as establishing / releasing RRC connections, configuring radio resources, and reporting signal measurement results, reflecting the wireless access status of the terminal.

[0038] NAS (Non-Access Stratum) signaling is a type of control signaling transmitted between the terminal and the 5G private network core network. It does not rely on the radio air interface and is mainly used to handle core services such as user registration, session management, and mobility management, reflecting the interaction status between the terminal and the core network.

[0039] NGAP (Next Generation Application Protocol) signaling is a type of control signaling transmitted between 5G private network base stations and the core network through the NG interface. It is mainly used to transmit interactive instructions between the base station and the core network. It is generated by direct interaction between the base station and the core network and serves as an intermediate bridge connecting RRC signaling and NAS signaling, reflecting the instruction forwarding status.

[0040] The three types of signaling collected separately are scattered and fragmented data. By using time sequence relationships, the fragmented signaling can be integrated into a complete business chain, ensuring that the control plane signaling flow can accurately reconstruct a user's complete communication behavior, and providing a complete and coherent data source for subsequent extraction of user real-time signaling behavior characteristics.

[0041] RRC signaling, NAS signaling, and NGAP signaling fully cover the entire interactive link from the terminal and base station to the core network, which can avoid misjudgment of user behavior due to missing signaling. The control plane signaling flow based on the embodiments of the present invention can accurately restore user behavior.

[0042] As a preferred implementation, the real-time signaling behavior characteristics include at least one of registration characteristics, spatial transformation characteristics, service activity characteristics, network quality characteristics, and abnormal signaling ratio characteristics.

[0043] In some preferred embodiments, the above features are combined to obtain real-time signaling behavior features.

[0044] Further, preferably, for registration features, step S2, extracting the user's real-time signaling behavior features based on the control plane signaling flow, includes: Based on the control plane signaling flow, obtain the distribution of registration reason types for terminal access to the 5G private network; Compute the registration success rate and average registration interval of terminals accessing the 5G private network; The user's registration characteristics are obtained based on the distribution of registration reason types, the registration success rate, and the average registration interval.

[0045] In the complete terminal registration process, the terminal sends a registration request via NAS signaling, and the base station forwards the registration request to the core network via NGAP signaling. If the core network accepts the registration request, it will return a registration acceptance signaling to the terminal; otherwise, it will return a registration rejection signaling.

[0046] When a terminal sends registration request-related signaling to the core network, it includes a registration reason field indicating the type of registration reason, such as initial registration upon power-on, registration triggered by location update, registration triggered by session reconstruction, or periodic registration. The distribution of registration reason types can be obtained by analyzing the proportion of each type within a sliding time window, reflecting the driving factors behind the terminal's registration behavior.

[0047] The registration success rate is related to the access stability of the 5G private network. The average registration interval can accurately reflect the registration frequency of a single terminal. If the average registration interval is low, it may indicate that the terminal access is unstable or that the terminal moves frequently.

[0048] Registration is the first step for a terminal to access a 5G private network. Only when a terminal successfully registers can a user carry out activities within the private network. This embodiment of the invention accurately reflects the terminal's access status and activity level within the private network through registration features, and can further determine whether there is active user flow or whether the user flow is new or existing.

[0049] As a preferred implementation, for spatial transformation features, step S2, extracting real-time signaling behavior features of the user based on the control plane signaling flow, includes: Based on the control plane signaling flow, the frequency of terminal handover to base station, handover failure rate, and average dwell time of the terminal at the base station are obtained; The user's switching characteristics are obtained based on the frequency, the switching failure rate, and the average dwell time. Based on the control plane signaling flow, the number of tracking area updates and the frequency of change of the terminal in the core network of the 5G private network are obtained to obtain the user's mobility characteristics. Based on the switching characteristics and the movement characteristics, the user's spatial transformation characteristics are obtained.

[0050] When terminal movement weakens the signal, it triggers RRC connection reconfiguration signaling to request a handover. The base stations coordinate handover resources via NGAP signaling. The frequency of terminal base station handover reflects the frequency of terminal movement within the private network; the handover failure rate reflects the link stability between base stations within the private network and is related to the network support capabilities during terminal movement; the average dwell time reflects the degree to which the terminal's spatial location remains fixed within the private network.

[0051] The tracing area (TA) is the smallest geographical unit used by the core network to locate a terminal. The number of tracing area updates, compared to the frequency of terminal switching base stations, can reflect the geographical movement of the terminal on a macro-spatial scale; the frequency of change reflects the speed of the terminal's movement.

[0052] In the above scheme, small-range movements of the terminal can be captured by switching features, and large-range movements of the terminal can be captured by movement features, ultimately forming a complete spatial transformation description.

[0053] As a preferred implementation, for service activity characteristics, step S2, extracting real-time signaling behavior characteristics of users based on the control plane signaling flow, includes: Based on the control plane signaling flow, obtain the number of sessions established by the terminal in the 5G private network, the average duration of the sessions, and the reasons for the sessions released. Based on the number of sessions established, the average duration of each session, and the reason for session release, the user's business activity characteristics are obtained.

[0054] When a terminal needs to use data services, it sends a PDU (Packet Data Unit) session establishment request to the core network in the form of NAS signaling. The core network coordinates resources with the base station through NGAP signaling and finally returns a session acceptance signaling or rejection signaling.

[0055] When a terminal or network terminates a session, it generates a session release request or session release command. The reason for the session release can be obtained through the reason code in the signaling. The reason for the session release is the basis for determining whether the session termination is abnormal.

[0056] In the above scheme, the business activity characteristics are directly related to the intensity of user terminal business demand, usage habits and network support status within the private network, and are the core basis for judging whether the flow of people business behavior is abnormal.

[0057] As a preferred implementation, regarding network quality characteristics, step S2, extracting real-time signaling behavior characteristics of users based on the control plane signaling flow, includes: Based on the control plane signaling flow, the reference signal sent from the base station to the terminal and the RRC connection hold time between the terminal and the base station are collected. Based on the reference signal, calculate a first average value and a first fluctuation rate of the received power of the reference signal, and calculate a second average value and a second fluctuation rate of the received quality of the reference signal; The user's network quality characteristics are obtained based on the first average value, the first volatility, the second average value, the second volatility, and the RRC connection hold time.

[0058] The reference signal is a standardized signal continuously broadcast by the base station to the terminal in the 5G private network. It is used by the terminal to measure the wireless channel quality and report back to the base station. The terminal reports the reception status of the reference signal to the base station in the form of RRC signaling. The 5G private network can obtain the original measurement value of the reference signal by parsing the signaling, reflecting the signal coverage strength and stability at the terminal's location.

[0059] A higher Reference Signal Receiving Power (RSRP) value indicates a stronger signal received by the terminal, while a lower value indicates a weaker signal. The first average value is the arithmetic mean of all RSRP measurements within the sliding time window, reflecting the overall signal strength level. The first fluctuation value is the dispersion of RSRP measurements within the sliding time window, reflecting the stability of the signal strength.

[0060] The Reference Signal Receiving Quality (RSRQ) is used to evaluate the signal-to-noise ratio (SNR) of a signal. A higher value indicates better signal quality, while a lower value indicates more severe interference. The second average value is the arithmetic mean of the RSRQ measurements within the sliding time window, reflecting the overall level of signal quality. The second fluctuation value is the dispersion of the RSRQ measurements within the sliding time window, reflecting the stability of the signal quality.

[0061] RRC connection hold time directly reflects the stability of the terminal connection.

[0062] In the above scheme, network quality characteristics can reflect the characteristics of wireless link quality and connection stability of the terminal in the 5G private network, directly related to the network coverage quality and communication continuity of the user's location, and can determine the network's ability to support the flow of people.

[0063] As a preferred implementation, for abnormal signaling ratio characteristics, step S2, extracting real-time signaling behavior characteristics of the user based on the control plane signaling flow, includes: Based on the control plane signaling flow, calculate the rejection signaling ratio and the abnormal code ratio in the total signaling; The user's abnormal signaling ratio characteristic is obtained based on the rejection signaling ratio and the abnormal code ratio.

[0064] Full signaling refers to the sum of all control plane signaling interactions between terminals and base stations, terminals and core networks, and base stations and core networks in a 5G private network, including various types of signaling such as registration, handover, session establishment, and location update.

[0065] Rejection and exception codes in signaling interaction are direct indicators of abnormal interaction between the terminal and the network. The abnormal signaling ratio feature in this embodiment of the invention quantifies the proportion of abnormalities in the signaling flow from two dimensions: interaction result and abnormal cause, avoiding the omission of abnormality due to a single perspective.

[0066] In this embodiment of the invention, all real-time signaling behavior features are extracted from the inherent control plane signaling flow of the 5G private network. No additional positioning equipment needs to be deployed, and existing private network resources can be reused, reducing the cost and complexity of the solution implementation.

[0067] As a preferred implementation, step S3, calculating the deviation between the user's real-time signaling behavior characteristics and the behavior baseline, and identifying abnormal crowd flow based on the deviation, includes: Based on the user's demographic tags, determine the target demographic category mentioned by the user, and obtain the typical characteristics of each subcategory within the target demographic category; Calculate the first deviation value between the real-time signaling behavior characteristics and each of the typical characteristics; Calculate the second deviation between each real-time signaling behavior feature and the mean of the user behavior baseline under the corresponding feature; Based on the first deviation value and the second deviation value, abnormal crowd flow is identified.

[0068] It should be noted that the aforementioned population category labels are macro-level classification identifiers pre-labeled for users, such as drivers, criminals, and patients, used to distinguish users' core identities or behavioral scenarios; subcategories refer to smaller groups obtained through clustering under the target population category, reflecting more refined behavioral commonalities among the same population, such as further dividing drivers into regular drivers and high-risk drivers.

[0069] The first deviation value is the difference between the user's real-time signaling behavior characteristics and the typical characteristics of each sub-category under the target population category. The standardized distance between the two is the first deviation value, which reflects the degree of deviation between real-time behavior and the typical pattern of the same population.

[0070] The second deviation value is the difference between a single real-time signaling behavior feature and the mean value of the corresponding feature in the user's own behavior baseline, reflecting the degree of deviation between real-time behavior and the user's own historical norms.

[0071] In some preferred embodiments, by A comprehensive anomaly score is calculated to identify abnormal pedestrian flow. Among these, and These are the preset weighting coefficients. The first deviation, The real-time signaling behavior characteristics at time t, This represents the mean of the baseline user behavior for the corresponding feature. This is the second deviation. This is a typical characteristic.

[0072] In the above scheme, considering that the baseline of a single user may have individual-specific biases, the scheme combines real-time features with typical group features to capture group-perspective anomalies, supplementing the limitations of the user's own baseline, and making anomaly identification more comprehensive.

[0073] Further, preferably, the method for extracting typical features corresponding to population categories includes: Based on the user's demographic category label, the historical signaling behavior features are hierarchically aggregated to obtain the category features corresponding to each demographic category; Principal component analysis is used to embed and map the category features to obtain low-dimensional behavior vectors; The low-dimensional behavior vectors are clustered to obtain groups with similar behaviors, which are then used as subcategories. Calculate the central prototype vector of the behaviorally similar groups, and use the central prototype vector as the typical feature of the subdivision category.

[0074] In some preferred embodiments, a combination of principal component analysis and autoencoder is used to compress and embed multidimensional features, represented as follows: Z is a low-dimensional behavior vector that can capture the implicit correlations between behavioral features.

[0075] Furthermore, based on the low-dimensional behavior vector Z, group clustering is performed using K-Means or DBSCAN algorithms to obtain groups with similar behaviors. .

[0076] based on Calculate the central prototype vector for each behaviorally similar group to obtain the typical features of the sub-category. Preferably, each prototype is the portrait template prototype for the sub-category.

[0077] In this embodiment of the invention, user profile templates are pre-defined based on the user's role or usage scenario in the 5G private network. Different user types have different core behavioral patterns, which may lead to differences in the key features emphasized by different profile templates. By focusing on key features through profile templates, only core indicators strongly correlated with user type can be retained, making it more targeted. For example, separate profile templates are established for high-risk drivers, elderly caregivers, and emergency response personnel.

[0078] In the above scheme, the progressive processing from macro categories to micro subcategories ensures that typical features not only conform to the commonalities of the population but also distinguish the specific scenarios. Furthermore, the typical features of the subcategories can provide an accurate benchmark for the calculation of the first deviation value, making the comparison between real-time behavior and the typical group more meaningful.

[0079] As a preferred embodiment, the method for constructing the behavioral baseline includes: Collect historical control plane signaling streams of users within a preset continuous time period, and extract historical signaling behavior features based on the historical control plane signaling streams; Calculate the mean and variance of the historical signaling behavior characteristics; Based on the mean and variance, a baseline of user behavior is constructed for each user.

[0080] It should be noted that the historical control plane signaling flow originates from the signaling storage system of the private network. New control plane signaling flows will be continuously and incrementally stored in the signaling storage system. In this embodiment of the invention, historical signaling within this period is collected retrospectively to provide raw data for extracting historical signaling behavior features.

[0081] It is understandable that historical signaling behavior characteristics are of the same type as real-time signaling behavior characteristics, reflecting the characteristics of a user's communication behavior over a period of time. The extraction logic of historical signaling behavior characteristics is the same as the method steps described in the above embodiments, and will not be repeated here.

[0082] The average value of historical signaling behavior characteristics over a preset continuous period reflects the normal level of the characteristics; the dispersion of historical signaling behavior characteristics, i.e., the variance, reflects the normal fluctuation range of the characteristics. The mean and variance can characterize the normal range of the current historical signaling behavior characteristics, and the constructed user behavior baseline fluctuates incrementally as data accumulates.

[0083] In the above scheme, a personalized, normal behavior benchmark that supports dynamic updates is established for each user to avoid misjudgments caused by a uniform baseline; and by combining the mean and variance, a clear user behavior baseline standard is provided for deviation calculation.

[0084] Further, preferably, the calculation of the mean and variance of the historical signaling behavior characteristics includes: Using statistical methods, the first mean and first variance of each of the aforementioned historical signaling behavior characteristics are calculated; Extract the time tags of the historical signaling behavior features, and obtain the feature time series data based on the time tags and the historical signaling behavior features; The feature time series data is input into a pre-trained recurrent neural network, and the periodic patterns within a preset continuous time period are captured by predicting signaling behavior to obtain the second mean and the second variance. Based on a preset confidence level, the first mean and the second mean are weighted and fused, and the first variance and the second variance are weighted and fused to obtain the mean and variance of historical signaling behavior features, respectively; the confidence level is the prediction error of signaling behavior.

[0085] The first mean and first variance calculated by traditional statistical methods can reflect the overall average level and overall fluctuation range of signaling characteristics in historical data throughout the entire time period, but they cannot capture the periodic patterns in the time dimension.

[0086] In this embodiment of the invention, a recurrent neural network is used to learn the periodic patterns of signaling behavior. Specifically, the learning process is based on feature time-series data, learning the long-term dependencies and implicit periods in the sequence to predict future signaling behavior, and inversely inferring the expected behavior level under different time scenarios to obtain a second mean and a second variance, which can accurately capture the behavioral fluctuations brought about by time periods.

[0087] For example, the expected signaling behavior at time t+1 is predicted using a recurrent neural network, denoted as: ;in, It is a non-linear mapping function for the network, used to capture implicit periodic patterns and context dependencies within continuous time periods.

[0088] The second mean generated by the final recurrent neural network is represented as follows: The second variance is expressed as .

[0089] pass The first and second means are combined to obtain the mean; through The first and second variances are fused. The confidence level is dynamically determined by the prediction error of the recurrent neural network.

[0090] In the above scheme, the confidence level is obtained through the prediction error of the recurrent neural network, which reflects whether the recurrent neural network accurately captures the periodic pattern of the current time scene. The final mean and variance can not only retain the long-term stability of the statistical method, but also incorporate the captured time periodic characteristics, and more accurately adapt to the signaling behavior.

[0091] The pedestrian flow management method provided by this invention relies on the localized deployment characteristics of 5G private networks. Signaling data does not need to rely on public network transmission, which can ensure the stability of data collection and analysis. At the same time, the dynamic baseline can adapt to the behavioral differences of different users, making it suitable for pedestrian flow management in various 5G private network scenarios.

[0092] This invention provides a people flow management system applied in a 5G private network. Please refer to [link / reference]. Figure 2 The pedestrian flow management system includes a data acquisition module 11, a feature extraction module 12, and an anomaly detection module 13, wherein: Data acquisition module 11 is used to collect control plane signaling flow between the core network and base station of 5G private network within the sliding time window in response to the sliding time window update; Feature extraction module 12 is used to extract real-time signaling behavior features of users based on the control plane signaling flow; The anomaly identification module 14 is used to calculate the deviation between the user's real-time signaling behavior characteristics and the behavior baseline, and to identify abnormal crowd flow based on the deviation; the behavior baseline is dynamically updated based on the user's historical signaling behavior characteristics.

[0093] In a preferred embodiment, the data acquisition module 11 is specifically used to: in response to the sliding time window update, acquire the RRC signaling between the terminal and the base station, the NAS signaling between the terminal and the core network of the 5G private network, and the NGAP signaling between the base station and the core network within the sliding time window; Based on the timing relationship between the RRC signaling, the NAS signaling, and the NGAP signaling, the control plane signaling flow is obtained.

[0094] In a preferred embodiment, when the real-time signaling behavior characteristic is a registration characteristic, the feature extraction module 12 is specifically used for: Based on the control plane signaling flow, obtain the distribution of registration reason types for terminal access to the 5G private network; Compute the registration success rate and average registration interval of terminals accessing the 5G private network; The user's registration characteristics are obtained based on the distribution of registration reason types, the registration success rate, and the average registration interval.

[0095] In a preferred embodiment, when the real-time signaling behavior characteristics are spatial transformation characteristics, the feature extraction module 12 is specifically used for: Based on the control plane signaling flow, the frequency of terminal handover to base station, handover failure rate, and average dwell time of the terminal at the base station are obtained; The user's switching characteristics are obtained based on the frequency, the switching failure rate, and the average dwell time. Based on the control plane signaling flow, the number of tracking area updates and the frequency of change of the terminal in the core network of the 5G private network are obtained to obtain the user's mobility characteristics. Based on the switching characteristics and the movement characteristics, the user's spatial transformation characteristics are obtained.

[0096] In a preferred embodiment, when the real-time signaling behavior characteristics are those of active services, the feature extraction module 12 is specifically used for: Based on the control plane signaling flow, obtain the number of sessions established by the terminal in the 5G private network, the average duration of the sessions, and the reasons for the sessions released. Based on the number of sessions established, the average duration of each session, and the reason for session release, the user's business activity characteristics are obtained.

[0097] In a preferred embodiment, when the real-time signaling behavior characteristics are network quality characteristics, the feature extraction module 12 is specifically used for: Based on the control plane signaling flow, the reference signal sent from the base station to the terminal and the RRC connection hold time between the terminal and the base station are collected. Based on the reference signal, calculate a first average value and a first fluctuation rate of the received power of the reference signal, and calculate a second average value and a second fluctuation rate of the received quality of the reference signal; The user's network quality characteristics are obtained based on the first average value, the first volatility, the second average value, the second volatility, and the RRC connection hold time.

[0098] In a preferred embodiment, when the real-time signaling behavior characteristic is an abnormal signaling ratio characteristic, the feature extraction module 12 is specifically used for: Based on the control plane signaling flow, calculate the rejection signaling ratio and the abnormal code ratio in the total signaling; The user's abnormal signaling ratio characteristic is obtained based on the rejection signaling ratio and the abnormal code ratio.

[0099] In a preferred embodiment, the anomaly detection module 13 is specifically used for: Based on the user's demographic tags, determine the target demographic category mentioned by the user, and obtain the typical characteristics of each subcategory within the target demographic category; Calculate the first deviation value between the real-time signaling behavior characteristics and each of the typical characteristics; Calculate the second deviation between each real-time signaling behavior feature and the mean of the user behavior baseline under the corresponding feature; Based on the first deviation value and the second deviation value, abnormal crowd flow is identified.

[0100] Further, preferably, the method for extracting typical features corresponding to population categories includes: Based on the user's demographic category label, the historical signaling behavior features are hierarchically aggregated to obtain the category features corresponding to each demographic category; Principal component analysis is used to embed and map the category features to obtain low-dimensional behavior vectors; The low-dimensional behavior vectors are clustered to obtain groups with similar behaviors, which are then used as subcategories. Calculate the central prototype vector of the behaviorally similar groups, and use the central prototype vector as the typical feature of the subdivision category.

[0101] As a preferred embodiment, the method for constructing the behavioral baseline includes: Collect historical control plane signaling streams of users within a preset continuous time period, and extract historical signaling behavior features based on the historical control plane signaling streams; Calculate the mean and variance of the historical signaling behavior characteristics; Based on the mean and variance, a baseline of user behavior is constructed for each user.

[0102] Further, preferably, the calculation of the mean and variance of the historical signaling behavior characteristics includes: Using statistical methods, the first mean and first variance of each of the aforementioned historical signaling behavior characteristics are calculated; Extract the time tags of the historical signaling behavior features, and obtain the feature time series data based on the time tags and the historical signaling behavior features; The feature time series data is input into a pre-trained recurrent neural network, and the periodic patterns within a preset continuous time period are captured by predicting signaling behavior to obtain the second mean and the second variance. Based on a preset confidence level, the first mean and the second mean are weighted and fused, and the first variance and the second variance are weighted and fused to obtain the mean and variance of historical signaling behavior features, respectively; the confidence level is the prediction error of signaling behavior.

[0103] The pedestrian flow management system provided by this invention relies on the local deployment characteristics of 5G private networks. Signaling data does not need to rely on public network transmission, which can ensure the stability of data collection and analysis. At the same time, the dynamic baseline can adapt to the behavioral differences of different users, making it suitable for pedestrian flow management in various 5G private network scenarios.

[0104] Please see Figure 3 , Figure 3 This is a structural block diagram of a pedestrian flow management device provided in an embodiment of the present invention. The pedestrian flow management device includes a processor 31, a memory 32, and a computer program stored in the memory 32 and executable on the processor 31. When the processor 31 executes the computer program, it implements the steps in the above-described pedestrian flow management method embodiments, such as steps S1 to S3.

[0105] For example, the computer program can be divided into one or more modules / units, which are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the pedestrian flow management device.

[0106] The pedestrian flow management device may include, but is not limited to, a processor 31 and a memory 32. Those skilled in the art will understand that the schematic diagram is merely an example of a pedestrian flow management device and does not constitute a limitation on the device. It may include more or fewer components than illustrated, or combine certain components, or use different components. For example, the pedestrian flow management device may also include input / output devices, network access devices, buses, etc.

[0107] The processor 31 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor 31 is the control center of the pedestrian flow management device, connecting all parts of the device via various interfaces and lines.

[0108] The memory 32 can be used to store the computer programs and / or modules. The processor 31 implements various functions of the crowd management device by running or executing the computer programs and / or modules stored in the memory 32 and calling the data stored in the memory 32. The memory 32 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0109] If the modules / units integrated into the crowd flow management device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by the processor 31, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0110] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for managing pedestrian flow, characterized in that, The pedestrian flow management method, applied in 5G private networks, includes: In response to the sliding time window update, the control plane signaling flow between the core network and base station of the 5G private network within the sliding time window is collected; Based on the control plane signaling flow, extract the user's real-time signaling behavior characteristics; The deviation between the user's real-time signaling behavior characteristics and the behavior baseline is calculated, and abnormal crowd flow is identified based on the deviation; the behavior baseline is dynamically updated based on the user's historical signaling behavior characteristics.

2. The method for managing pedestrian flow as described in claim 1, characterized in that, The response to the sliding time window update, collecting the control plane signaling flow between the core network and base station of the 5G private network within the sliding time window, includes: In response to the sliding time window update, the system collects RRC signaling between the terminal and the base station, NAS signaling between the terminal and the core network of the 5G private network, and NGAP signaling between the base station and the core network within the sliding time window. Based on the timing relationship between the RRC signaling, the NAS signaling, and the NGAP signaling, the control plane signaling flow is obtained.

3. The method for managing pedestrian flow as described in claim 1, characterized in that, The real-time signaling behavior features include registration features. Therefore, extracting the user's real-time signaling behavior features based on the control plane signaling flow includes: Based on the control plane signaling flow, obtain the distribution of registration reason types for terminal access to the 5G private network; Compute the registration success rate and average registration interval of terminals accessing the 5G private network; The user's registration characteristics are obtained based on the distribution of registration reason types, the registration success rate, and the average registration interval.

4. The method for managing pedestrian flow as described in claim 1, characterized in that, The real-time signaling behavior features include spatial transformation features. Therefore, extracting the user's real-time signaling behavior features based on the control plane signaling flow includes: Based on the control plane signaling flow, the frequency of terminal handover to base station, handover failure rate, and average dwell time of the terminal at the base station are obtained; The user's switching characteristics are obtained based on the frequency, the switching failure rate, and the average dwell time. Based on the control plane signaling flow, the number of tracking area updates and the frequency of change of the terminal in the core network of the 5G private network are obtained to obtain the user's mobility characteristics. Based on the switching characteristics and the movement characteristics, the user's spatial transformation characteristics are obtained.

5. A method for managing pedestrian flow as described in claim 1, characterized in that, The real-time signaling behavior characteristics include service activity characteristics. Therefore, extracting the user's real-time signaling behavior characteristics based on the control plane signaling flow includes: Based on the control plane signaling flow, obtain the number of sessions established by the terminal in the 5G private network, the average duration of the sessions, and the reasons for the sessions released. Based on the number of sessions established, the average duration of each session, and the reason for session release, the user's business activity characteristics are obtained.

6. The method for managing pedestrian flow as described in claim 1, characterized in that, The real-time signaling behavior features include network quality features. Therefore, extracting the user's real-time signaling behavior features based on the control plane signaling flow includes: Based on the control plane signaling flow, the reference signal sent from the base station to the terminal and the RRC connection hold time between the terminal and the base station are collected. Based on the reference signal, calculate a first average value and a first fluctuation rate of the received power of the reference signal, and calculate a second average value and a second fluctuation rate of the received quality of the reference signal; The user's network quality characteristics are obtained based on the first average value, the first volatility, the second average value, the second volatility, and the RRC connection hold time.

7. The method for managing pedestrian flow as described in claim 1, characterized in that, The real-time signaling behavior features include abnormal signaling ratio features. Therefore, extracting the user's real-time signaling behavior features based on the control plane signaling flow includes: Based on the control plane signaling flow, calculate the rejection signaling ratio and the abnormal code ratio in the total signaling; The user's abnormal signaling ratio characteristic is obtained based on the rejection signaling ratio and the abnormal code ratio.

8. The method for managing pedestrian flow as described in claim 1, characterized in that, The step of calculating the deviation between the user's real-time signaling behavior characteristics and the behavior baseline, and identifying abnormal crowd flow based on the deviation, includes: Based on the user's demographic tags, determine the target demographic category mentioned by the user, and obtain the typical characteristics of each subcategory within the target demographic category; Calculate the first deviation value between the real-time signaling behavior characteristics and each of the typical characteristics; Calculate the second deviation between each real-time signaling behavior feature and the mean of the user behavior baseline under the corresponding feature; Based on the first deviation value and the second deviation value, abnormal crowd flow is identified.

9. A method for managing pedestrian flow as described in claim 8, characterized in that, Methods for extracting typical features corresponding to population categories include: Based on the user's demographic category label, the historical signaling behavior features are hierarchically aggregated to obtain the category features corresponding to each demographic category; Principal component analysis is used to embed and map the category features to obtain low-dimensional behavior vectors; The low-dimensional behavior vectors are clustered to obtain groups with similar behaviors, which are then used as subcategories. Calculate the central prototype vector of the behaviorally similar groups, and use the central prototype vector as the typical feature of the subdivision category.

10. A method for managing pedestrian flow as described in claim 1 or 8, characterized in that, The method for constructing the behavioral baseline includes: Collect historical control plane signaling streams of users within a preset continuous time period, and extract historical signaling behavior features based on the historical control plane signaling streams; Calculate the mean and variance of the historical signaling behavior characteristics; Based on the mean and variance, a baseline of user behavior is constructed for each user.

11. A method for managing pedestrian flow as described in claim 10, characterized in that, The calculation of the mean and variance of the historical signaling behavior characteristics includes: Using statistical methods, the first mean and first variance of each of the aforementioned historical signaling behavior characteristics are calculated; Extract the time tags of the historical signaling behavior features, and obtain the feature time series data based on the time tags and the historical signaling behavior features; The feature time series data is input into a pre-trained recurrent neural network, and the periodic patterns within a preset continuous time period are captured by predicting signaling behavior to obtain the second mean and the second variance. Based on a preset confidence level, the first mean and the second mean are weighted and fused, and the first variance and the second variance are weighted and fused to obtain the mean and variance of historical signaling behavior features, respectively; the confidence level is the prediction error of signaling behavior.

12. A pedestrian flow management system, characterized in that, The crowd flow management system, applied in 5G private networks, includes: The data acquisition module is used to collect the control plane signaling flow between the core network and the base station of the 5G private network within the sliding time window in response to the sliding time window update. The feature extraction module is used to extract real-time signaling behavior features of users based on the control plane signaling flow; The deviation calculation module is used to calculate the deviation between the user's real-time signaling behavior characteristics and the user behavior baseline; the user behavior baseline is dynamically updated based on the user's historical signaling behavior characteristics within the 5G private network. An anomaly detection module is used to identify abnormal pedestrian flow based on the deviation.

13. A pedestrian flow management device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the crowd flow management method as described in any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the pedestrian flow management method as described in any one of claims 1 to 11.

Images