Description:TECHNICAL FIELD
[0001] The present disclosure relates, in general, to surveillance and security systems, and more specifically, relates to an abnormal pattern detection system for perimetric surveillance data.

BACKGROUND
[0002] Almost all public or private organization spends a good amount of revenue on the security of their campus. The level of security layers also depends on the nature of the organization. Security layers can be described as different types of sensors based on their range, effectiveness and other characteristics used for different security purposes. These sensors are deployed across the perimeter of the campus for monitoring activities. The main challenge for security personals is the on-time assessment of threats and neutralizing threats at the earliest. This particular task becomes more difficult because of the data diversification due to seasonality and demographic variance.
[0003] Nowadays intelligent sensors are being used that provide essential and meaningful data. The amalgamation of various types of sensory data like radars, under-ground sensor, camera and so on, require an effective and efficient system that may prioritize and extract the relevant information from the voluminous sensor data.
[0004] Filtering algorithms should be incorporated to filter out non-critical alerts from the sensors. The filtering technique should be reliable enough to isolate the actual threat from the entire data. Traditional systems that lacked a data-driven statistical approach do not perform well due to the unavailability of correlated data with respect to terrain types, seasons, populations around the campus, and the like. The other disadvantage of the traditional approach is that it cannot be generalized to all campuses of any organization across the country, which means the system may learn the pattern or behaviour from its own data. All the campuses of an organisation within the country differ geographically and demographically. The behaviour and pattern of the local movements depend on the surrounding types, weather conditions, and livelihood of locals.
[0005] Anomaly detection based on the statistical analysis of multi-sensor data across the perimeter or border for any malicious activity differs from other anomaly detection fields. Trend and pattern analysis needs to be done to capture any anomalous event on the data received either from imagery or non-imagery sensors. Some systems use only specific sensors while others consider amalgamated input from multiple types of sensors for anomaly detection as per requirement. Below prior arts were found those proposed systems, which work for intrusion detection or any other types of anomaly detection for perimeter defence.
[0006] An example of such anomaly detection is recited in a patent US7274387B2. The invention proposes a method that helps to detect intruders, terror attacks or any other malicious activity across the perimeter of a restricted or sensitive area that can range from military bases to city water reservoirs or an oil refinery to an entire airfield containing enormous and endless potential targets. The system locates, identifies, and performs assessments on the detected objects to calculate the potential threat before providing its decision as an anomaly or intruder. The technique proposed in this invention is an intrusion detection system that uses image data from the camera (both visible and infrared) and applies optical image processing techniques to detect any anomaly or intruder. The method performs automatic motion detection on the image data that has a high probability of detection and a low false alarm rate. The method detects the change in the frames to detect any mischievous activity which is next assessed and the threshold is calculated by a threat assessment function to detect an anomaly or intruder. However. the proposed system is limited to only visual sensors for anomaly or intrusion detection and does not consider other sensor data like radar or any underground sensor.
[0007] The method proposed in the invention US7274387B2 uses motion detection techniques for intrusion or anomaly detection based on imagery data. The method suffers drawbacks caused by object speed and frame rate. Also, it suffers certain drawbacks on variable depth shots due to old techniques of motion detection. Moreover, the system proposed in this invention focuses only on imagery data from visual sensors for anomaly detection and does not consider other types of sensors. The method used for image processing shows low performance in detecting anomalies or intrusions when data is flooded at a very high rate. This approach does not perform proper area optimization of the perimeter and thus it is not suitable for densely populated areas like houses or roads outside or nearby the perimeter or border. In such circumstances, the system may mislead the operator.
[0008] Yet another example is recited in a patent CA2662444 titled “A method and system for determining a threat against a border” provided a solution for finding out an anomaly or threat from objects crossing or trying to cross the border. These objects can be categorized as employees, work-seekers, activists, terrorists, and many more with different motives. The above solution has segmented the entire border into border elements so that there is uniformity with respect to terrain features, infrastructure, and weather conditions by using high-resolution topographical, vector maps and satellite pictures of the border area. The drawback of the system proposed in the invention is it does not use any outlier detection techniques and lacks a proper statistical model. Using appropriate outlier detection techniques can automatically increase the accuracy of the system in terms of anomaly or intrusion detection as a large volume of data is received at a very high velocity from multiple sensors deployed along the perimeter or border
[0009] Therefore, it is desired to overcome the drawbacks, shortcomings, and limitations associated with existing solutions, and develop a system that increases the overall operational efficiency.

OBJECTS OF THE PRESENT DISCLOSURE
[0010] An object of the present disclosure relates, in general, to surveillance and security systems, and more specifically, relates to an abnormal pattern detection system for perimetric surveillance data.
[0011] Another object of the present disclosure is to provide a system that develops an analytic engine with the help of various exploratory data analysis techniques on the time-series data. As a result, the data that is prepared by the system describes the trends, which help to form clustered data.
[0012] Another object of the present disclosure is to provide a system that can segregate probable alerts from data coming from multiple sensors. The overall system is decision-making and performs efficiently in terms of handling large amounts of data at a very fast rate and filtering out threats.
[0013] Another object of the present disclosure is to provide a system that divides the areas into high traffic and low traffic zones to make resource allocation and reallocation tasks easier.
[0014] Yet another object of the present disclosure is to provide a system that reduces high consumption costs and expensive utilization of resources and increases the overall operational efficiency.

SUMMARY
[0015] The present disclosure relates, in general, to surveillance and security systems, and more specifically, relates to an abnormal pattern detection system for perimetric surveillance data. The main objective of the present disclosure is to overcome the drawback, limitations, and shortcomings of the system and solution, by providing a system that detects an abnormal pattern in context to any type of suspicious activities across the perimeter based on the data of various sensors deployed along the perimeter.
[0016] The system includes one or more sensors that are deployed along the perimeter of a target area, the one or more sensors adapted to capture a set of data pertaining to surveillance data of the perimeter in the target area. An edge layer having a plurality of edge nodes coupled to the one or more sensors, each edge node adapted to receive the set of data from the one or more sensors. An abnormal pattern detection system coupled to the edge layer, the abnormal pattern detection system is configured to collect the set of data based on detections of one or more sensors to determine trends and patterns, parse the set of data collected from the one or more sensors and filter the unwanted data from the received set of data, thereby reducing high consumption cost and expensive utilization of resources and increases the overall operational efficiency. The abnormal pattern detection system performs all the required data transformations for obtaining meaningful insights of the hidden data patterns for advanced data analytics. A decision support system coupled to the abnormal pattern detection system, the decision support system is adapted to receive resultant analysis on the perimetric surveillance data to draw appropriate decisions on any malicious activity across the perimeter. Thereby, the system can segregate probable alerts from data coming from multiple sensors. Thus, the overall system is itself decision-making and performs efficiently in terms of handling a large amount of data at a very fast rate and filtering out the threats.
[0017] The abnormal pattern detection system can include an on-board display and configuration unit (ODCU) sub-system, the ODCU configured to receive, by an entity creator component of the ODCU, the geo-positional information along with terrain profile of the area to automatically create static entities which are virtual boundaries and sectors that lie exterior to the perimeter at different proximities and display, by a display component of the ODCU, the static entities which are virtual boundaries and sectors.
[0018] The abnormal pattern detection system can include a data storage server sub-system that is configured to store, by an entity storage component of the data storage server sub-system, entity configuration received from the entity creator component of the ODCU. The historical storage component of the data storage server sub-system can store the data transformed by the DWS at a periodic rate, where the data is used as historical evidence for further analysis of data patterns. The data storage server sub-system can store an updated resultant model at a model storage component of the data storage server sub-system. The system enables faster storage of real-time data that may be further used for historical data analysis to gain deep insights and provide high-quality information on the hidden data trends and patterns for better decision-making.
[0019] Further, the abnormal pattern detection system can include data wrangling server (DWS) sub-system, the DWS sub-system configured to collect, by a data collection component of the DWS, the raw surveillance data based on detections of the one or more sensors so that the set of data is examined to determine trends and patterns, filter, by a data cleansing component of the DWS, collected data that corrects the corrupted, inaccurate, or irrelevant parts of the set of data that can distort the results of analysis, and restructures the disorganized data and transform, by a data transformation component of DWS, the filtered set of data for advanced data analytics.
[0020] The abnormal pattern detection system can include an adaptive learning engine (ALE) sub-system configured to receive, by an information retrieval component of the ALE, date range and area polygons from the user, fetches the corresponding historical evidence from the data storage server, analyse, by a data analyzer component of the ALE, extensive data analysis and learn, by a pattern learning component of the ALE, the patterns of respective areas using clustering process.
[0021] Moreover, the abnormal pattern detection system can include an analytic engine (AE) sub-system that receives real-time transformed data from the DWS sub-system for advanced data analysis. The AE sub-system is configured to detect, by an abnormal pattern detector component of the AE sub-system, the abnormalities in the real-time data that aid in efficient decision-making and calculate, by a threat score calculator and prioritizer component, the threat score based on the weights, area type and priority level and assign weightage to the surveillance data. The system develops an analytic engine with the help of various exploratory data analysis techniques on the time-series data. As a result, the data that is prepared by the system describes the trends, which help to form clustered data.
[0022] Various objects, features, aspects, and advantages of the inventive subject matter will become more apparent from the following detailed description of preferred embodiments, along with the accompanying drawing figures in which like numerals represent like components.

BRIEF DESCRIPTION OF THE DRAWINGS
[0023] The following drawings form part of the present specification and are included to further illustrate aspects of the present disclosure. The disclosure may be better understood by reference to the drawings in combination with the detailed description of the specific embodiments presented herein.
[0024] FIG. 1 illustrates an exemplary representation of system of abnormal pattern detection, in accordance with an embodiment of the present disclosure.
[0025] FIG. 2 illustrates an exemplary block diagram of abnormal pattern detection system, in accordance with an embodiment of the present disclosure.
[0026] FIG. 3 illustrates an exemplary block diagram of on-board display and configuration unit sub-system, in accordance with an embodiment of the present disclosure.
[0027] FIG. 4 illustrates an exemplary block diagram of data storage server sub-system, in accordance with an embodiment of the present disclosure.
[0028] FIG. 5 illustrates an exemplary block diagram of data wrangling server sub-system, in accordance with an embodiment of the present disclosure.
[0029] FIG. 6 illustrates an exemplary block diagram of adaptive learning engine sub-system, in accordance with an embodiment of the present disclosure.
[0030] FIG. 7 illustrates an exemplary flow chart of clustering algorithm for pattern learning, in accordance with an embodiment of the present disclosure.
[0031] FIG. 8 illustrates an exemplary block diagram of analytic engine (AE) sub-system, in accordance with an embodiment of the present disclosure.
[0032] FIG. 9 illustrates an exemplary flow chart of a method of abnormal pattern detection, in accordance with an embodiment of the present disclosure.

DETAILED DESCRIPTION
[0033] The following is a detailed description of embodiments of the disclosure depicted in the accompanying drawings. The embodiments are in such detail as to clearly communicate the disclosure. If the specification states a component or feature “may”, “can”, “could”, or “might” be included or have a characteristic, that particular component or feature is not required to be included or have the characteristic.
[0034] As used in the description herein and throughout the claims that follow, the meaning of “a,” “an,” and “the” includes plural reference unless the context clearly dictates otherwise. Also, as used in the description herein, the meaning of “in” includes “in” and “on” unless the context clearly dictates otherwise.
[0035] The present disclosure relates, in general, to surveillance and security systems, and more specifically, relates to an abnormal pattern detection system for perimetric surveillance data. The proposed system disclosed in the present disclosure overcomes the drawbacks, shortcomings, and limitations associated with the conventional system by providing a system that detects an abnormal pattern in context to any type of suspicious activities across the perimeter based on the data of various sensors deployed along the perimeter. The method creates static entities which are virtual boundaries and zones/sectors that lie exterior to the actual perimeter at different proximities. The proposed method has developed an adaptive learning engine that memorizes the pattern along the perimeter using a clustering algorithm on the static entities. The real-time data analytic engine detects any type of unusual and abnormal pattern across the perimeter that can pose a threat to the entire security system of the area. Likewise, resources can be allocated or reallocated to take protective measures.
[0036] The present disclosure provides the user more control in terms of the data selection in a given period and entity selection to customize the adaptive learning process of the abnormal pattern detection system to learn the data patterns along the perimeter or border. The pattern can be updated month-wise or season-wise or any specific period and with respect to area-entity selection. The proposed pattern learning method is developed using a clustering algorithm. The clusters are formed considering adjacent time clusters and neighbouring area/sector clusters. The inclusion of adjacent time and area clusters makes the method more robust to define the pattern along the perimeter or border very precisely. The present disclosure can be described in enabling detail in the following examples, which may represent more than one embodiment of the present disclosure.
[0037] The advantages achieved by the system of the present disclosure can be clear from the embodiments provided herein. The system develops the analytic engine with the help of various exploratory data analysis techniques on the time-series data. As a result, the data that is prepared by the system itself describes the trends, which help to form clustered data. The system can segregate probable alerts from data coming from multiple sensors. The overall system is decision-making and performs efficiently in terms of handling large amounts of data at a very fast rate and filtering out threats. Further, the system can divide the areas into high traffic and low traffic zones to make resource allocation and reallocation tasks easier. The description of terms and features related to the present disclosure shall be clear from the embodiments that are illustrated and described; however, the invention is not limited to these embodiments only. Numerous modifications, changes, variations, substitutions, and equivalents of the embodiments are possible within the scope of the present disclosure. Additionally, the invention can include other embodiments that are within the scope of the claims but are not described in detail with respect to the following description.
[0038] FIG. 1 illustrates an exemplary representation of a system of abnormal pattern detection, in accordance with an embodiment of the present disclosure.
[0039] Referring to FIG. 1, a system 100 of abnormal pattern detection in a border or perimetric surveillance data is disclosed. One or more sensors (102-1 to 102-4 (which are collectively referred to as sensors 102, herein)) are deployed along the perimeter or border of any area. The sensors 102 include radars, cameras, electric fencing, and an underground sensor deployed along any area's perimeter and border. The sensors 102 adapted to capture the set of data pertaining to surveillance data of the perimeter in the target area. The data captured from the sensors 102 are received by an edge layer 104 that can include multiple edge nodes (106-1 to 106-4 (which are collectively referred to as edge node 106, herein)), the edge node 106 coupled to each sensor 102. Each edge node 106 is adapted to receive the set of data from the sensors 102. The edge nodes 106 are responsible for the collection, data forwarding and connection to the servers.
[0040] In an embodiment, an abnormal pattern detection system 110 is connected to the edge layer 104 via a network 108 that provides maximum agility, performance, and operational efficiency. The system 100 can include multiple components that perform various functions to detect any abnormality in data patterns along the perimeter or border based on the deployed sensor detections.
[0041] The abnormal pattern detection system 110 is configured to collect the set of data based on detections of the sensors 102 to determine trends and patterns. The abnormal pattern detection system 110 parses the raw data collected from the sensors 102 and filters out the unwanted data from a deluge of unclean sensor data. This helps in reducing high consumption costs and expensive utilization of resources and increases the overall operational efficiency. It performs all the required data transformations that help in obtaining meaningful insights into the hidden data patterns for advanced data analytics. The system 110 also contains components that help in faster storage of real-time data that may be further used for historical data analysis to gain deep insights and provide high-quality information on the hidden data trends and patterns for better decision-making.
[0042] The system 110 performs advanced data analysis using clustering algorithms to detect any abnormality in the patterns of surveillance data along the border or perimeter that can pose a potential threat. Using adaptive learning, the system 110 learns the hidden data trends and patterns in the recorded data and gains a deeper understanding of the valuable information that easily helps to detect any abnormality in the real-time data. A decision support system 112 coupled to the abnormal pattern detection system 110. Any abnormality or threat detected by system 110 is further transmitted to the decision support system 112 to assist it to take an appropriate decision on any kind of malicious activity across the boundary and take quick actions accordingly.
[0043] In an embodiment, the abnormal pattern detection system 110 shown in FIG. 2 can include an on-board display and configuration unit (ODCU) sub-system 202, the ODCU configured to receive the geo-positional information along with the terrain profile of the area to automatically create static entities, which are virtual boundaries and sectors and display the static entities which are virtual boundaries and sectors. The abnormal pattern detection system 110 can include a data storage server sub-system 206 to store entity configuration.
[0044] The abnormal pattern detection system 110 can include data wrangling server (DWS) sub-system 204, the DWS sub-system configured to collect the raw surveillance data based on detections of the sensors 102 so that the set of data is examined to determine trends and patterns. The DWS sub-system 204 filters the collected data that corrects the corrupted, inaccurate, or irrelevant parts of the set of data that can distort the results of analysis, and restructures the disorganized data and transform the filtered set of data for advanced data analytics. Further, the data storage server sub-system 206 is configured to store the data transformed by the DWS at a periodic rate, where the data is used as historical evidence for further analysis of data patterns.
[0045] The abnormal pattern detection system 110 can include an adaptive learning engine (ALE) sub-system 208 configured to receive date range and area polygons from user, and fetches the corresponding historical evidence from the data storage server 206. The ALE sub-system 208 analyse extensive data analysis and learns the patterns of respective areas using a clustering process. The abnormal pattern detection system 110 can include an analytic engine (AE) sub-system 210 that receives real-time transformed data from the DWS sub-system for advanced data analysis. The AE sub-system 210 is configured to detect the abnormalities in the real-time data that aids in efficient decision-making and calculate the threat score based on the weights and prioritizes and assigns weightage to the surveillance data.
[0046] The anomaly detection algorithm on time-series data that has been used is unique of its kind because the algorithm identifies the patterns along the perimeter considering the adjacent time and area clusters. The anomaly or the outlier, which are the same are extracted using elongated dimensions with respect to time and proximity of sectors. The analytic engine uses desired and selected historical evidence of incoming sensory data with inherited demographical characteristics of area type to declare the probability of threat severity. The solution formulates three exterior virtual boundaries and sectors along the perimeter. This helps to associate perimetric surveillance data to areas along the perimeter and prioritize them based on the historical evidence and the terrain profile of its associated area type. This assists the quick response team to take appropriate actions by measuring the impact of the alerts generated, based on the proximity, area type and priority level and time required for the quick action.
[0047] FIG. 2 illustrates an exemplary block diagram of the abnormal pattern detection system, in accordance with an embodiment of the present disclosure.
[0048] Referring to FIG. 2, the abnormal pattern detection system 110 detects any abnormality in perimetric surveillance data by performing extensive data analysis. The system 100 helps to find out any potential threat along the perimeter and aids in better decision-making. The system 110 can include ODCU sub-system 202, the DWS sub-system 204, the data storage server sub-system 206, ALE sub-system 208, and the AE sub-system 210.
[0049] The abnormal pattern detection system 110 provides a graphical user interface to the operator using the ODCU 202. The abnormal pattern detection system 110 allows reliable data storage using its data storage server sub-system 206. The abnormal pattern detection system 110 collects, pre-processes, and transforms the raw data of the deployed sensors 102 using the DWS sub-system 204. The input from the user may be given to the ALE sub-system 208. The real-time transformed data is sent to the AE sub-system 210 by the DWS sub-system 204 for advanced data analysis.
[0050] The abnormal pattern detection system 110 coupled to the edge layer 104 via a network 108, the abnormal pattern detection system configured to collect the set of data based on detections of the one or more sensors to determine trends and patterns, parse the set of data collected from the one or more sensors and filter the unwanted data from the received set of data and perform all the required data transformations for obtaining meaningful insights of the hidden data patterns for advanced data analytics.
[0051] FIG. 3 illustrates an exemplary block diagram of the on-board display and configuration unit sub-system, in accordance with an embodiment of the present disclosure.
[0052] In an exemplary embodiment, the ODCU 202 can include Windows 10 operating system, with 16 GB RAM, a display component 302, an entity creator component 304, a data selector component 306, and a threat notifier component 308.
[0053] The ODCU sub-system 202 is configured to receive, by an entity creator component 304 of the ODCU 202, the geo-positional information along with the terrain profile of the area to automatically create static entities which are virtual boundaries and sectors and display, by a display component 302 of the ODCU 202, the static entities which are virtual boundaries and sectors.
[0054] The entity creator component 304 of ODCU 202 may receive the geo-positional information along with the terrain profile of the area given as input by the operator. Based on this information, the entity creator component 304 may automatically create static entities which are virtual boundaries and sectors and the same may be displayed to the operator by the display component 302 of the ODCU. The entities created may be saved in the data storage server sub-system 206 by the entity creator component 304.
[0055] FIG. 4 illustrates an exemplary block diagram of data storage server sub-system, in accordance with an embodiment of the present disclosure. The data storage server sub-system 206 (MongoDB in Ubuntu operating System) can include a historical data storage component 402, an entity storage component 404, and a model storage component 406. The entity storage component 404 may save the entity configuration details received from the entity creator component 304 of ODCU 202.
[0056] The data storage server sub-system 206 configured to store, by an entity storage component 404 of the data storage server sub-system, entity configuration received from the entity creator component 304 of the ODCU 202. The data storage server sub-system 206 is configured to store in a historical storage component 402 of data storage server sub-system 206 the data transformed by the DWS at a periodic rate, where the data is used as historical evidence for further analysis of data patterns. The data storage server sub-system 206 is configured to store an updated resultant model at a model storage component 406 of the data storage server sub-system 206.
[0057] FIG. 5 illustrates an exemplary block diagram of data wrangling server sub-system, in accordance with an embodiment of the present disclosure. The abnormal pattern detection system 110 collects, pre-processes, and transforms the raw data of the deployed sensors using the DWS sub-system 204. In an exemplary embodiment, the DWS 204 can include Ubuntu operating system with 32 GB RAM, a data collection component 502, a data cleansing component 504, and a data transformation component 506.
[0058] The data collection component 502 of DWS 204 collects the raw surveillance data based on sensor detections so that data can be understood and examined to find trends and patterns. The collected data is further cleansed by the data cleansing component 504 of DWS that detects and corrects the corrupted, inaccurate, or irrelevant parts of the data that can distort the results of analysis, and restructures the disorganized data to fit in with the analytical model and standardizes the data format to improve quality and consistency. The cleansed data is transformed by the data transformation component 506 of DWS 204 for advanced data analytics that helps in obtaining deeper insights into the hidden data patterns and efficient adaptive learning. The data transformed by the data wrangling server 204 may be stored at a periodic rate in the historical storage component 402 of data storage server sub-system 206. The data may be used as historical evidence for further analysis of data patterns.
[0059] FIG. 6 illustrates an exemplary block diagram of adaptive learning engine sub-system, in accordance with an embodiment of the present disclosure. As shown in FIG. 6, the proposed system in the innovation provides the operator the functionality to select the date range and the static entities for which the system can learn the hidden data patterns adaptively using the ODCU 202. The input from the user may be given to the ALE sub-system 208.
[0060] The ALE sub-system 208 reveals the deeper intelligence of the system as it makes the system capable and flexible enough to easily adapt itself to new information and can gain meaningful insight almost instantaneously. In an exemplary embodiment, the ALE sub-system 208 can include Ubuntu operating system with 64 GB RAM, an information retrieval component 602, data analyzer component 604 and a pattern learning component 606.
[0061] The information retrieval component 602 of the ALE 208 receives date range and area polygons from the user and fetches the corresponding historical evidence from the data storage server 206. After receiving the desired data, the component forwards the same to data analyzer component 604 of ALE. The data analyzer component 604 after performing an extensive data analysis, sends its results to the pattern learning component 606, which learns the patterns of respective areas using the clustering method shown in FIG. 7.
[0062] The existing clustering algorithms do not provide the desired result. To overcome the limitations of the existing clustering algorithm, the customized clustering algorithm is proposed. In the clustering method, the standard deviation (σ) of each cluster is calculated using the points of the two adjacent time clusters and area clusters and the radius of circle epsilon is calculated by multiplying the standard deviation(σ) with Γ. Here Γ is the multiplication factor of standard deviation (In our case Γ is 1.8 and the minimum number of points is 3 which is obtained based on various experiments and result comparisons on data sets).
[0063] The updated resultant model of the clustering process is saved into the model storage component 406 of the data storage server 206. The real-time transformed data is sent to the AE sub-system 210 by the DWS sub-system 204 for advanced data analysis.
[0064] FIG. 8 illustrates an exemplary block diagram of analytic engine (AE) sub-system, in accordance with an embodiment of the present disclosure.
[0065] In an exemplary embodiment, the AE sub-system 210 can include Ubuntu operating system with 32 GB RAM, an abnormal pattern detector component 802 and a threat score calculator and prioritizer component 804.
[0066] The AE 210 receives model from model storage component 406 of data storage server 206 for extensive data analysis and detection of abnormal data patterns. The abnormal pattern detector component 802 detects the abnormalities in the real-time sensory data that aids in efficient decision-making.
[0067] The AE sub-system 210 receives real-time transformed data from the DWS sub-system for advanced data analysis. The AE sub-system 210 is configured to detect, by the abnormal pattern detector component 802 of the AE sub-system, the abnormalities in the real-time data that aid in efficient decision-making and calculate, by a threat score calculator and prioritizer component 804, the threat score based on the weights, area type, priority level and assigns weightage to the surveillance data.
[0068] The threat (w) is calculated using the measured frequency c of a specific time-period and the pattern threshold frequency is t.
[0069] The threat score calculator and prioritizer component 804 calculates the threat score based on the weights and prior of the area type and prioritizes and assigns weightage to the surveillance data.

[0070] Where a is type of the physical characteristics of that area. The assigned scaling factor of a are as follows: -
a= 0.5 – For areas surrounded by houses or roads
a= 1.0 – For areas containing agricultural land
a= 1.5 – For areas containing barren land
[0071] Threat score interpretation matrix shown in table 1 below.
Serial No. Probability score Result
1. < 0.3 Ignore
2. > 0.3 and < 0.7 Moderate
3. > 0.7 Severe

Table 1: Threat score interpretation matrix
[0072] If the calculated threat score is greater than 0.7 then this condition may be treated as potential threat. The resultant analysis on perimetric surveillance data of the AE is further transmitted to the decision support system, to draw appropriate decision on any kind of malicious activity across the boundary and take quick actions accordingly. The analysis results are further displayed to the user on the ODCU.
[0073] Thus, the present invention overcomes the drawbacks, shortcomings, and limitations associated with existing solutions, and provides the analytic engine with the help of various exploratory data analysis techniques on the time-series data. As a result, the data that is prepared by the system itself describes the trends which help to form clustered data. The system can segregate probable alerts from data coming from multiple sensors. The overall system is itself decision making and perform efficiently in terms of handling large amount of data at a very fast rate and filtering out the threats. Further, the system can divide the areas into high traffic and low traffic zones to make resource allocation and reallocation task easier.
[0074] FIG. 9 illustrates an exemplary flow chart of a method of abnormal pattern detection, in accordance with an embodiment of the present disclosure.
[0075] Referring to FIG. 9, the method 900 of abnormal pattern detection, the method includes at block 902 one or more sensors deployed along the perimeter of a target area, the one or more sensors adapted to capture a set of data pertaining to surveillance data of the perimeter in the target area.
[0076] At block 904, each edge node receives the set of data from the one or more sensors, an edge layer having a plurality of edge nodes coupled to the one or more sensors. At block 906, the abnormal pattern detection system can collect the set of data based on detections of the one or more sensors to determine trends and patterns, the abnormal pattern detection system coupled to the edge layer.
[0077] At block 908, the abnormal pattern detection system parses the set of data collected from the one or more sensors and filter the unwanted data from the received set of data. At block 910, the abnormal pattern detection system can perform all the required data transformations for obtaining meaningful insights of the hidden data patterns for advanced data analytics. At block 912, a decision support system coupled to the abnormal pattern detection system can receive resultant analysis on the perimetric surveillance data to draw appropriate decisions on any malicious activity across the perimeter.
[0078] It will be apparent to those skilled in the art that the system 100 of the disclosure may be provided using some or all of the mentioned features and components without departing from the scope of the present disclosure. While various embodiments of the present disclosure have been illustrated and described herein, it will be clear that the disclosure is not limited to these embodiments only. Numerous modifications, changes, variations, substitutions, and equivalents will be apparent to those skilled in the art, without departing from the spirit and scope of the disclosure, as described in the claims.

ADVANTAGES OF THE PRESENT INVENTION
[0079] The present invention provides a system that develops an analytic engine with the help of various exploratory data analysis techniques on the time-series data. As a result, the data that is prepared by the system itself describes the trends which help to form clustered data.
[0080] The present invention provides a system that can segregate probable alerts from data coming from multiple sensors. The overall system is itself decision making and perform efficiently in terms of handling large amount of data at a very fast rate and filtering out the threats.
[0081] The present invention provides a system that is able to divide the areas into high traffic and low traffic zones to make resource allocation and reallocation task easier.
, Claims:1. A system (100) of abnormal pattern detection, the system comprising:
one or more sensors (102) that are deployed along the perimeter of a target area, the one or more sensors adapted to capture a set of data pertaining to surveillance data of the perimeter in the target area;
an edge layer (104) having a plurality of edge nodes (106) coupled to the one or more sensors, each edge node adapted to receive the set of data from the one or more sensors;
an abnormal pattern detection system (110) coupled to the edge layer, the abnormal pattern detection system configured to:
collect the set of data based on detections of the one or more sensors to determine trends and patterns;
parse the set of data collected from the one or more sensors and filter the unwanted data from the received set of data;
perform all the required data transformations for obtaining meaningful insights of the hidden data patterns for advanced data analytics; and
a decision support system (112) coupled to the abnormal pattern detection system, the decision support system is adapted to receive resultant analysis on the surveillance data to draw appropriate decisions on any malicious activity across the perimeter.
2. The system as claimed in claim 1, wherein the abnormal pattern detection system (110) comprises an on-board display and configuration unit (ODCU) sub-system (202), the ODCU (202) configured to:
receive, by an entity creator component (304) of the ODCU, the geo-positional information along with terrain profile of the target area to automatically create static entities which are virtual boundaries and sectors that lie exterior to the perimeter at different proximities; and
display, by a display component (302) of the ODCU, the static entities which are virtual boundaries and sectors.
3. The system as claimed in claim 1, wherein the abnormal pattern detection system (110) comprises a data storage server sub-system (206), the data storage server sub-system configured to:
store, by an entity storage component (404) of the data storage server sub-system, entity configuration received from the entity creator component of the ODCU.
4. The system as claimed in claim 1, wherein the abnormal pattern detection system (110) comprises data wrangling server (DWS) sub-system (204), the DWS sub-system configured to:
collect, by a data collection component (502) of the DWS, the raw surveillance data based on detections of the one or more sensors so that the set of data is examined to determine trends and patterns;
filter, by a data cleansing component (504) of the DWS, the collected data that corrects the corrupted, inaccurate, or irrelevant parts of the set of data that distort the results of analysis, and restructures the disorganized data; and
transform, by a data transformation component (506) of the DWS, the filtered set of data for advanced data analytics.
5. The system as claimed in claim 1, wherein the data storage server sub-system (206) is configured to store in a historical storage component (402) of the data storage server sub-system the data transformed by the DWS at a periodic rate, wherein the transformed data is used as historical evidence for further analysis of data patterns.
6. The system as claimed in claim 1, wherein the data storage server sub-system (206) configured to store an updated resultant model of a clustering process at a model storage component (406) of the data storage server sub-system.
7. The system as claimed in claim 1, wherein the abnormal pattern detection system (110) comprises adaptive learning engine (ALE) sub-system (208) configured to:
receive, by an information retrieval component (602) of the ALE, date range and area polygons from user, and fetches the corresponding historical evidences from the data storage server;
analyse, by a data analyzer component (604) of the ALE, extensive data analysis; and
learns, by a pattern learning component (606) of the ALE, the patterns of respective areas using the clustering process.
8. The system as claimed in claim 1, wherein the abnormal pattern detection system (110) comprises an analytic engine (AE) sub-system (210) that receives real-time transformed data from the DWS sub-system for advanced data analysis.
9. The system as claimed in claim 1, wherein the AE sub-system is configured to:
detect, by an abnormal pattern detector component (802) of the AE sub-system, the abnormalities in the real-time data that aids in efficient decision-making; and
calculate, by a threat score calculator and prioritizer component (804), the threat score based on the weights, area type and priority level and assigns weightage to the surveillance data.
10. A method of abnormal pattern detection, the method comprising:
deploying one or more sensors along the perimeter of a target area, the one or more sensors adapted to capture a set of data pertaining to surveillance data of the perimeter in the target area;
receiving by each edge node the set of data from the one or more sensors, an edge layer having a plurality of edge nodes coupled to the one or more sensors;
collecting, at an abnormal pattern detection system, the set of data based on detections of the one or more sensors to determine trends and patterns, the abnormal pattern detection system coupled to the edge layer;
parsing, at the abnormal pattern detection system, the set of data collected from the one or more sensors and filter the unwanted data from the received set of data;
performing, at the abnormal pattern detection system, all the required data transformations for obtaining meaningful insights of the hidden data patterns for advanced data analytics; and
receiving, at a decision support system coupled to the abnormal pattern detection system, resultant analysis on the perimetric surveillance data to draw appropriate decision on any malicious activity across the perimeter.