A method and system for regional edge number identification

Abstract

The application relates to the field of mobile communication security technology and discloses a regional edge number identification method and system, which comprises the following steps: acquiring signaling data of a mobile terminal and generating signaling trajectory data after preprocessing; acquiring all base stations in a key area and their position distribution according to the signaling trajectory data; calculating the dynamic weight of each base station relative to the center point of the area based on all the base stations in the key area; constructing a daily frequency space-time feature matrix based on the signaling trajectory data and the dynamic weight of the base stations; constructing a monthly frequency space-time feature matrix based on the signaling trajectory data; calculating a number edge index based on the daily frequency space-time feature matrix and the monthly frequency space-time feature matrix, and accurately identifying the edge number based on the number edge index. The application realizes accurate identification of the edge number of the key area based on the characteristics of the mobile phone signaling data of the user in the communication data, combined with multi-dimensional feature analysis of the user, a dynamic base station weight model and four-dimensional feature collaborative decision-making.

Description

Technical Field

[0001] This invention relates to the field of mobile communication security technology, and more specifically to a method and system for identifying regional edge numbers. Background Technology

[0002] Currently, in scenarios such as emergency management and protection of critical facilities, the demand for accurate identification of abnormal behavior around specific areas is becoming increasingly prominent. Traditional monitoring methods mainly rely on fixed-location cameras, access control systems, or sensor networks, which have significant limitations: on the one hand, they cannot effectively cover open or semi-open areas without access control, making it difficult to detect abnormal loitering or repeated approaching behavior; on the other hand, they still rely on manual review of a large number of images or card swipe records for post-event investigation, resulting in insufficient real-time early warning capabilities; furthermore, in areas lacking dedicated monitoring equipment, such as parks, squares, and open areas, it is impossible to effectively obtain continuous personnel movement trajectories.

[0003] Big data in communications offers a new solution to these problems—signaling data generated through interaction between mobile terminals and base stations can reflect the location and movement characteristics of people in real time. However, existing signaling data analysis schemes still have significant shortcomings: for example, area-based personnel identification methods based on dwell time are almost ineffective during low-activity periods such as nighttime; in addition, traditional GPS electronic fence technology has a large positioning error (usually exceeding 200 meters) in areas at the edge of base station signal coverage. When people move within the area boundary, signal fluctuations or drift can easily cause trajectory interruptions and misjudgments, making it impossible to effectively identify suspicious behaviors such as loitering or frequent approaching. Overall, existing methods struggle to distinguish between terminal numbers whose actual location is at the physical edge (such as near fences or boundaries) and those whose behavioral patterns exhibit edge characteristics (such as frequent short visits), and also lack a fusion and discrimination mechanism for signal strength attenuation characteristics and spatiotemporal behavioral patterns.

[0004] Therefore, how to accurately identify edge numbers (i.e., unauthorized or potentially high-risk personnel frequently approaching important areas) from massive signaling data has become a pressing technical challenge in the field of emergency monitoring and regional security protection. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for identifying regional edge numbers. By analyzing the signaling trajectory data of user mobile phones, analyzing the key area range and the distribution of internal base stations, constructing spatiotemporal trajectory features, exploring based on the features, calculating edge indexes, and effectively identifying suspicious numbers at the edge of the region.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for identifying regional edge numbers includes:

[0008] S1. Obtain signaling data from the mobile terminal, preprocess it, and generate signaling trajectory data;

[0009] S2. Based on the signaling trajectory data, obtain all base stations and their location distribution within the key area;

[0010] S3. Based on all base stations within the key area, calculate the dynamic weight of each base station relative to the center point of the area.

[0011] S4. Based on the signaling trajectory data and the base station dynamic weights, construct a daily frequency spatiotemporal feature matrix;

[0012] S5. Based on the signaling trajectory data, construct a monthly frequency spatiotemporal feature matrix;

[0013] S6. Based on the daily frequency spatiotemporal feature matrix and the monthly frequency spatiotemporal feature matrix, calculate the number edge index, and accurately identify edge numbers based on the number edge index.

[0014] Preferably, S1 includes:

[0015] Acquire signaling data from the mobile terminal and perform data cleaning on the signaling data to filter out data with abnormal trajectory status;

[0016] After cleaning the terminal signaling trajectory data, the trajectory of each number is sorted in chronological order to obtain the initial trajectory data;

[0017] The initial trajectory data is subjected to outlier noise removal, data aggregation, and trajectory completion to obtain the signaling trajectory data.

[0018] Preferably, S2 includes:

[0019] Based on the map, obtain the coverage area of ​​key areas, select the center point of the area based on the coverage area, and extract the latitude and longitude of the center point of the area;

[0020] Based on the signaling trajectory data, signaling trajectory data for the past month is extracted, and the distance between each trajectory and the latitude and longitude of the regional center point is calculated using the trajectory latitude and longitude. Trajectory points whose distances meet preset conditions are selected, and the number and location distribution of base stations in key areas are statistically analyzed based on the selected trajectory points.

[0021] Preferably, the dynamic weight w i The calculation formula is:

[0022] w i =exp(k*(d max -d i ))

[0023] Where, d i Let d be the distance from base station i to the center point of the region.max The value is the maximum distance from all base stations in the region to the center point of the region, and k is the attenuation coefficient.

[0024] Preferably, S4 includes:

[0025] Based on signaling trajectory data and base station dynamic weights within the regional period, the number of times each number connects to each base station in the key area and the total number of times it connects to all base stations in the key area on a single day are counted for each number.

[0026] The connection density index (CDI) of a base station is constructed using the following formula:

[0027] CDI=Σ(con i *w i ) / con k

[0028] Where, con i The number of times each number connects to base station i in a day, w i Here, con represents the dynamic weight parameters for base station i. k This represents the total number of times the number connected to all base stations in the key area within one day.

[0029] Calculate the standard deviation (SDS) of base station handover intervals:

[0030]

[0031] Among them, T j T represents the time interval between the j-th base station handover for each number within the key area, where N is the total number of base station handovers within the period; avg This represents the average number of base station handovers.

[0032] Preferably, S5 includes:

[0033] Based on signaling trajectory data within the regional period, the number of days connected to all base stations in the key area within a single month and the maximum number of consecutive days connected to all base stations in the key area within a single month are calculated for each number.

[0034] Calculate the duration of low-frequency visits (LFD):

[0035] LFD = day totle / max(1,day con )

[0036] Among them, day totle This represents the number of days the phone number appeared in the designated area within a month; day con This refers to the maximum number of consecutive days that the number appears in the key area within a month;

[0037] Calculate the spacetime density index SDI:

[0038] SDI=-Σ(p t *lnp t )

[0039] Where, p t Let t be the percentage of the number's occurrences in different hourly segments t within a month, relative to the total number of occurrences, where t = 1, 2, 3, ..., 24.

[0040] Preferably, the formula for calculating the edge index (EI) is as follows:

[0041] EI=α*CDI+β*SDS+γ*LFD-δ*SDI

[0042] Wherein, α, β, γ and δ are all adjustable weighting coefficients, obtained from statistical analysis of the data.

[0043] Preferably, accurate identification of edge numbers is based on the edge number index, including:

[0044] If EI > K, the corresponding number is judged as an edge number or a passing number;

[0045] If EI≤K, the corresponding number is judged as a non-marginal number or a permanent number, where K is a preset threshold.

[0046] A regional edge number identification system, comprising:

[0047] Signaling trajectory data acquisition module: used to acquire signaling data from mobile terminals, preprocess it, and generate signaling trajectory data;

[0048] Base station location analysis module: used to obtain all base stations and their location distribution within the key area based on the signaling trajectory data;

[0049] Base station weight dynamic allocation module: used to calculate the dynamic weight of each base station relative to the center point of the area based on all base stations in the key area;

[0050] Daily frequency spatiotemporal feature construction module: used to construct a daily frequency spatiotemporal feature matrix based on the signaling trajectory data and the base station dynamic weights;

[0051] Monthly frequency spatiotemporal feature construction module: used to construct a monthly frequency spatiotemporal feature matrix based on the signaling trajectory data;

[0052] Edge system calculation and identification module: used to calculate the number edge index based on the daily frequency spatiotemporal feature matrix and the monthly frequency spatiotemporal feature matrix, and to accurately identify edge numbers based on the number edge index.

[0053] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method and system for identifying regional edge numbers, which has the following effects:

[0054] 1. This invention, based on various characteristics of user mobile phone signaling data in communication data, combined with multi-dimensional user feature analysis, dynamic base station weight model, and four-dimensional feature collaborative decision-making, achieves accurate identification of edge numbers in key areas. It solves the problem of edge signal perception in concealed key areas and reduces the false alarm rate for high-frequency visitors (delivery drivers or couriers).

[0055] 2. This invention is based on communication data, feature engineering, and expert system rules. It does not require expensive hardware materials or the deployment of hardware facilities such as cameras and sensor networks. It directly utilizes operator signaling data, reducing implementation costs. Edge computing supports dynamic parameter optimization (such as automatically increasing the β coefficient at night), adapting to different scenarios such as office areas, nuclear power plants, and border checkpoints. It can also be widely used in smart city management, technical reconnaissance, and security control.

[0056] 3. This invention has the advantages of high accuracy, real-time performance and wide applicability, which can effectively improve the efficiency and accuracy of edge number identification and contribute to the development and progress of smart city management, technical investigation and security control.

[0057] 4. This invention combines big data in communications with multi-dimensional feature strategies, and utilizes advanced data analysis techniques and algorithms to achieve the identification and tracking of edge numbers in key areas. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0059] Figure 1 The present invention provides a flowchart of a method for identifying regional edge numbers.

[0060] Figure 2 This is a schematic diagram illustrating the base station weight allocation within a key area provided by the present invention.

[0061] Figure 3 This is an architecture diagram for multi-dimensional feature fusion computation provided by the present invention.

[0062] Figure 4 This is a schematic diagram of a regional edge number recognition system provided by the present invention. Detailed Implementation

[0063] 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.

[0064] This invention discloses a method for identifying regional edge numbers, such as... Figure 1 As shown, it includes:

[0065] S1. Obtain signaling data from the mobile terminal, preprocess it, and generate signaling trajectory data;

[0066] S2. Based on the signaling trajectory data, obtain all base stations and their location distribution within the key area;

[0067] S3. Based on all base stations in the key area, calculate the dynamic weight of each base station relative to the center point of the area, so that the base station weight is dynamically allocated according to the location of the base station.

[0068] S4. Construct a daily frequency spatiotemporal feature matrix based on signaling trajectory data and base station dynamic weights;

[0069] S5. Construct a monthly frequency spatiotemporal feature matrix based on signaling trajectory data;

[0070] S6. Based on the daily and monthly spatiotemporal feature matrices, calculate the number edge index and accurately identify edge numbers based on the number edge index.

[0071] The specific implementation process of each step of the present invention will be described in detail below.

[0072] In this embodiment, S1 includes:

[0073] Acquire signaling data from mobile terminals and perform data cleaning, deleting data with empty or zero key fields or abnormal trajectory status caused by base station data parsing errors or abnormal terminal device signals, in order to remove data noise;

[0074] The cleaned terminal signaling trajectory data is sorted in chronological order to obtain the initial trajectory data;

[0075] The initial trajectory data is subjected to outlier noise removal, the ping-pong switching data is used for trajectory point aggregation, and the initial trajectory data with few trajectory points or missing data fields is used for trajectory completion.

[0076] Preferably, the signaling trajectory data mainly includes trajectory location data, which contains different trajectory points. Each trajectory point is a sample data, containing information such as mobile phone number, time, latitude and longitude.

[0077] In this embodiment, based on signaling trajectory data, all base stations and their geographical distribution within a key area are identified. Specifically:

[0078] Use a map marker tool to plot the designated key area, and select a center point based on the coverage area. The center point can be defined according to the population density within the area, such as administrative buildings, important office areas, etc. Extract the latitude and longitude of the center point, and measure the distance d from the center point to the farthest boundary point. max .

[0079] Based on signaling trajectory data, the distance of each trajectory point from the latitude and longitude of the center point of the area is calculated, and points with a distance less than or equal to d are filtered out. max All signaling trajectories within the region are used for subsequent analysis.

[0080] Based on the selected signaling data, the number of base station IDs, the latitude and longitude location of each base station, and the distance from the center point of the area are counted.

[0081] In this embodiment, based on the base stations in the key area, the dynamic weight of each base station relative to the center point of the area is calculated, so that the base station weight is dynamically allocated according to the location of the base station.

[0082] like Figure 2 As shown, the specific rules for the dynamic allocation of base station weights are as follows:

[0083] Based on the principle that "the farther a base station is from the center point of the area, the lower its weight," a base station weight calculation rule is set. A dynamic attenuation coefficient k is introduced, with a value range of [0.05, 0.1]. In this embodiment, the default value of k is 0.08.

[0084] The specific calculation formula is as follows:

[0085] w i =exp(k*(d max -d i ))

[0086] Where, d i Let d be the distance from base station i to the center point of the region. max This represents the maximum distance from all base stations within the region to the region's center point, where k is the attenuation coefficient, and the value of k varies for different regions.

[0087] If the distance from the farthest boundary of the region to the center point is 500m, the distance from base station A to the center point is 500m, and the distance from base station B to the center point is approximately 0m, then according to the base station weight calculation formula, the weight of base station A is: w A =exp(0.08*(500-500))=e 0 =1, the weight of base station B is: w B =exp(0.08*(500-0))=e -40 ≈0.

[0088] In this embodiment, the daily frequency spatiotemporal feature matrix is ​​constructed as follows:

[0089] Based on the signaling trajectory and base station dynamic weights within the regional period, the number of times each number connects to each base station in the region on a single day and the total number of times all base stations in the region connect on a single day are counted for each number.

[0090] Based on the number of times a user connects to different base stations and the base station weight, a "Base Station Connection Density Index (CDI)" is constructed to obtain the CDI score ratio of base stations in the connection area within a single day.

[0091] CDI=Σ(con i *w i ) / con k

[0092] Where, con i The number of times each number connects to base station i in a day, w i Here, con represents the dynamic weight parameters for base station i. k This represents the total number of times the number connected to all base stations in the key area within one day.

[0093] A higher CDI (Connectivity Density Index) indicates that users are more likely to be located at the edge of the area; a lower CDI indicates that users are more likely to be located at the center of the area.

[0094] For example, if a person connects to base station A 0 times, base station B 2 times, and base station C 8 times on a certain day, then CDI = (0*1 + 2*0 + 8*0.8) / 10 = 0.64, where the weights of base stations A, B, and C are 1, 0, and 0.8 respectively. Based on this feature, the user has a 64% probability of being active on the edge of the region.

[0095] Based on the signaling trajectory data within the regional period, the time difference between adjacent trajectory points is calculated as the dwell time; the average dwell time of each user appearing in all trajectory points within the region on a single day is calculated; and the standard deviation of the dwell time within the region on a single day for each user is calculated as the "Standard Deviation of Base Station Handover Interval (SDS)" feature.

[0096]

[0097] Among them, T j T represents the time interval between the j-th base station handover for each number within the key area, where N is the total number of base station handovers within the period; avg This represents the average number of base station handovers.

[0098] A higher standard deviation of base station handover interval (SDS) indicates that the user's base station handover interval fluctuates greatly within the area, which is consistent with the behavioral characteristics of marginal personnel who linger and stay briefly; a lower standard deviation of base station handover interval (SDI) indicates that the user's base station handover interval fluctuates little within the area, and is mostly a person within the area who moves regularly.

[0099] For example, if a person connects to base stations within a certain area and stays for a certain period of time on a certain day, it can be expressed as: [A->B: 120s, B->C: 90s, C->B: 110s]. The mean stay time is (120+90+110) / 3 = 106.67s, the variance is (120-106.67)^2+(90-106.67)^2+(110-106.67)^2 = 466.67, and the standard deviation of SDI is 15.28s. This indicates that the user's handover interval fluctuates moderately and there are short stays in the area.

[0100] In this embodiment, the monthly frequency spatiotemporal feature matrix is ​​constructed as follows:

[0101] Based on the signaling trajectory within the regional period, the maximum number of days each user connects to all base stations in the key area within a single month and the maximum number of consecutive days connected to all base stations in the key area within a single month are counted. The "Low Frequency Continuous Days (LFD)" feature is constructed to obtain the ratio of the number of consecutive days connected to base stations in the area within a single month.

[0102] LFD = day totle / max(1,day con )

[0103] Among them, day totle This represents the number of days the phone number appeared in the designated area within a month; day con This refers to the maximum number of consecutive days that the number appears in the key area within a month;

[0104] The higher the LFD value for consecutive low-frequency days, the more frequently users visit the key area, but the visits are more scattered, and the greater the probability that they are marginal or passersby; the lower the LFD value for consecutive low-frequency days, the less likely they are marginal or passersby.

[0105] Low-frequency continuous days (LFD) can capture scattered visit patterns and identify individuals with high suspicion on the margins, but it may lead to misjudgments. For example, delivery drivers who briefly visit key areas every day have low LFD values ​​and are more likely to be local residents.

[0106] Therefore, in addition to this, a suppression feature related to the distribution of visit time, the "spatial-temporal density index (SDI)," is introduced to suppress LFD misjudgments caused by high-frequency regular visits.

[0107] SDI=-∑(p t *lnp t )

[0108] Where, p t Let SDI be the percentage of the number's occurrences in different hourly segments t within a month, relative to the total number of occurrences, where t = 1, 2, 3, ..., 24. SDI can also be expressed as:

[0109] SDI = -[p1*lnp1 + p2*lnp2 + ... + p 24 *lnp 24 ]

[0110] The Spatiotemporal Density Index (SDI) reflects the distribution of users' visits to key areas across different time periods. A higher SDI value indicates a more even distribution of visits, with a greater tendency towards people within the area; a lower SDI value indicates a more dispersed distribution of visits, with a greater tendency towards people on the periphery of the area.

[0111] For example, if a delivery driver delivers to a key area for 22 days in a month, with a maximum consecutive delivery period of 22 days, then according to the LFD calculation formula, LFD = 22 / max(1,22) = 1.0. This indicates that although the driver has frequent and continuous visits, they do not meet the marginal dispersion characteristic. If a worker within a certain area works 22 consecutive days, the LFD value is also 1.0. Therefore, they cannot be effectively separated, and the SDI feature is introduced.

[0112] If the delivery driver's delivery time in the past month is concentrated between 11:00-12:00 with a probability P1 of 0.45 and between 12:00-13:00 with a probability of 0.55, then according to the SI calculation formula, the SDI value is 0.69, indicating that his behavior pattern is highly concentrated (delivering food during lunchtime).

[0113] Therefore, by using LFD+SDI to make joint decisions, we can characterize users' marginal behaviors and improve the accuracy of edge recognition.

[0114] In this embodiment, as Figure 3 As shown, based on the daily and monthly frequency spatiotemporal feature matrices, the number edge index is calculated as follows:

[0115] By fusing four-dimensional features—CDI, SDS, LFD, and SDI—and using data with labels for resident personnel within the region for feature fusion weight adjustment, as well as final EI threshold adjustment, the system ensures that resident personnel's mobile phone numbers are filtered out when the EI value is less than a specified threshold. The dynamic threshold K is typically set as the mean of the top 15% of EI values ​​in historical data plus twice the standard deviation.

[0116] The formula for calculating the edge index (EI) is as follows:

[0117] EI=α*CDI+β*SDS+γ*LFD-δ*SDI

[0118] Wherein, α, β, γ and δ are all adjustable weighting coefficients, obtained from statistical analysis of the data.

[0119] If EI > K, the corresponding number is judged as an edge number or a passing number;

[0120] If EI≤K, the corresponding number is identified as a non-marginal number or a permanent number, and its information is entered into the internal personnel database.

[0121] Using a set of permanent residents in a region for a given month, the weighting coefficients (such as α, β, γ, and δ) of the edge number identification system are optimized. The specific process for setting each coefficient in the system is as follows:

[0122] Obtain the list of permanent residents and their mobile phone numbers in a key area for a certain month, and extract the terminal signaling data for that month as fitting (or training) data;

[0123] Based on S2-S5, the monthly data is processed to extract the daily and monthly frequency spatiotemporal feature matrices for each number.

[0124] Based on S6, the edge index of phone numbers is fitted using spatiotemporal trajectory features. The weighting coefficients α, β, γ, and δ within the edge computing system are statistically analyzed and adjusted, along with the threshold K of the final edge index EI. This ensures that when the edge index EI ≤ K, it can maximize the coverage of mobile phone numbers of resident residents within that month, thus solidifying the weight values ​​within the edge number calculation system for that region. Specifically: Based on the previously identified resident residents in a historical month as labels, the CDI, SDS, LFD, and SDI values ​​for that month are calculated. By setting grid parameters, the values ​​of parameters α, β, γ, δ, and K are enumerated, where each parameter ranges from 0 to 1. Each calculated mobile phone number satisfying EI ≤ K is compared with the resident resident labels. The α, β, γ, δ, and K values ​​with the highest overlap are the final fixed parameter values.

[0125] In this embodiment, a regional edge number identification system is disclosed, such as... Figure 4 As shown, it includes:

[0126] Signaling trajectory data acquisition module: used to acquire signaling data from mobile terminals, preprocess it, and generate signaling trajectory data;

[0127] Base station location analysis module: used to obtain all base stations and their location distribution within a key area based on signaling trajectory data;

[0128] Base station weight dynamic allocation module: used to calculate the dynamic weight of each base station relative to the center point of the area based on all base stations in the key area;

[0129] Daily frequency spatiotemporal feature construction module: used to construct a daily frequency spatiotemporal feature matrix based on signaling trajectory data and base station dynamic weights;

[0130] Monthly frequency spatiotemporal feature construction module: used to construct a monthly frequency spatiotemporal feature matrix based on signaling trajectory data;

[0131] Edge system calculation and identification module: It is used to calculate the edge index of numbers based on the daily frequency spatiotemporal feature matrix and the monthly frequency spatiotemporal feature matrix, and to accurately identify edge numbers based on the edge index.

[0132] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0133] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for identifying regional edge numbers, characterized in that, include: S1. Obtain signaling data from the mobile terminal, preprocess it, and generate signaling trajectory data; S2. Based on the signaling trajectory data, obtain all base stations and their location distribution within the key area; S3. Based on all base stations within the key area, calculate the dynamic weight of each base station relative to the center point of the area. S4. Construct a daily frequency spatiotemporal feature matrix based on the signaling trajectory data and the base station dynamic weights; S5. Based on the signaling trajectory data, construct a monthly frequency spatiotemporal feature matrix; S6. Based on the daily frequency spatiotemporal feature matrix and the monthly frequency spatiotemporal feature matrix, calculate the number edge index, and accurately identify edge numbers based on the number edge index; S4 includes: Based on signaling trajectory data and base station dynamic weights within the regional period, the number of times each number connects to each base station in the key area and the total number of times it connects to all base stations in the key area on a single day are counted for each number. The connection density index (CDI) of a base station is constructed using the following formula: in, The number of times each number connects to base station i in a day. Let i be the dynamic weight parameter of base station i. This represents the total number of times the number connected to all base stations in the key area within one day. Calculate the standard deviation (SDS) of base station handover intervals: in, The time interval for the j-th base station handover for each number in the key area, and N is the total number of base station handovers within the period; This represents the average value during base station handover. S5 includes: Based on signaling trajectory data within the regional period, the number of days connected to all base stations in the key area within a single month and the maximum number of consecutive days connected to all base stations in the key area within a single month are calculated for each number. Calculate the duration of low-frequency visits (LFD): in, This represents the number of days within a month that the phone number appeared in key areas. This refers to the maximum number of consecutive days that the number appears in the key area within a month; Calculate the spacetime density index SDI: in, Let t be the percentage of the number's occurrences in different hourly segments t within a month, where t = 1, 2, 3, ..., 24.

2. The method for identifying regional edge numbers according to claim 1, characterized in that, S1 includes: Acquire signaling data from the mobile terminal and perform data cleaning on the signaling data to filter out data with abnormal trajectory status; After cleaning the terminal signaling trajectory data, the trajectory of each number is sorted in chronological order to obtain the initial trajectory data; The initial trajectory data is subjected to outlier noise removal, data aggregation, and trajectory completion to obtain the signaling trajectory data.

3. The method for identifying regional edge numbers according to claim 1, characterized in that, S2 include: Based on the map, obtain the coverage area of ​​key areas, select the center point of the area based on the coverage area, and extract the latitude and longitude of the center point of the area; Based on the signaling trajectory data, signaling trajectory data for the past month is extracted, and the distance between each trajectory and the latitude and longitude of the regional center point is calculated using the trajectory latitude and longitude. Trajectory points whose distances meet preset conditions are selected, and the number and location distribution of base stations in key areas are statistically analyzed based on the selected trajectory points.

4. The method for identifying regional edge numbers according to claim 1, characterized in that, Dynamic weights The calculation formula is: in, Let i be the distance from base station i to the center point of the region. This represents the maximum value from all base stations within the region to the region's center point. This is the attenuation coefficient.

5. The method for identifying regional edge numbers according to claim 1, characterized in that, Number Edge Index The calculation formula is: in, , , and All of these are adjustable weighting coefficients, obtained from statistical data analysis.

6. The method for identifying regional edge numbers according to claim 5, characterized in that, Accurate identification of edge numbers based on the edge number index, including: If EI > K, the corresponding number is judged as an edge number or a passing number; If EI≤K, the corresponding number is judged as a non-marginal number or a permanent number, where K is a preset threshold.

7. A regional edge number identification system, characterized in that, To implement the method of any one of claims 1-6, comprising: Signaling trajectory data acquisition module: used to acquire signaling data from mobile terminals, preprocess it, and generate signaling trajectory data; Base station location analysis module: used to obtain all base stations and their location distribution within the key area based on the signaling trajectory data; Base station weight dynamic allocation module: used to calculate the dynamic weight of each base station relative to the center point of the area based on all base stations in the key area; Daily frequency spatiotemporal feature construction module: used to construct a daily frequency spatiotemporal feature matrix based on the signaling trajectory data and the base station dynamic weights; Monthly frequency spatiotemporal feature construction module: used to construct a monthly frequency spatiotemporal feature matrix based on the signaling trajectory data; Edge system calculation and identification module: used to calculate the number edge index based on the daily frequency spatiotemporal feature matrix and the monthly frequency spatiotemporal feature matrix, and to accurately identify edge numbers based on the number edge index.

Images