Comparison of clustering methods DBSCAN and modified WrapDBSCAN to find abnormal user movements in the mobile UBA system

One of the urgent problems in the existing systems of behavior analysis is the extraction of signs of anomalous activity of user activity from large arrays of input data.The problem solved in this study is based on the impossibility of searching for anomalous activity of users by their movements, due to the high variability of the input data. The aim of the study is to develop a modified density clustering method for application in a mobile system of behavioral analysis using machine learning methods and algorithms to find deviations in user behavior based on their movements. This article provides a comparative analysis of the density clustering methods used in the developed software package for searching for anomalies in the behavioral biometric characteristics of system users. Smoothing interpolation of the input data is performed. The results of searching for anomalies by the modified method of spatial clustering with different input parameters are described and the results are compared with the basic method. Thanks to the use of the developed method of spatial clustering, an increase in the quality of the analysis of anomalous activity in the activities of users on their movements has been achieved. Finding deviations in the collected data will ensure a timely response of the system administrator to deviations from the user's behavioral profile

1. Dhiman C., Vishwakarma D. K. A review of state.of.the.art techniques for abnormal human activity recognition. Engineering Applications of Artificial Intelligence. 2019;(77):21-45.

2. Shaw E. D. et al. Behavioral risk indicators of malicious insider theft of intellectual property: Misreading the writing on the wall .White Paper, Symantec, Mountain View. 2011.

3. Carmagnola F., Cena F. User identification for cross-system personalisation. Information Sciences. 2009;(179):16-32.

4. Shashanka M., Shen M. Y., Wang J. User and entity behavior analytics for enterprise security. 2016 IEEE International Conference on Big Data (Big Data). – IEEE. 2016:1867-1874.

5. Ivutin A. N., Savenkov P. A., Veselova A. V. Neural network for analysis of additional authentication behavioral biometrie characteristics. 2018 7th Mediterranean Conference on Embedded Computing (MECO). – IEEE. 2018(8):1-3.

6. Mon T. L. L. Analysis of Trajectory Cleaning Based on DBSCAN and CB-SMOT Clustering Algorithms. International Journal of Advanced Research in Technology and Innovation. 2020(2):35-41.

7. Savenkov P. A., Ivutin A. N. Methods of Machine Learning in System Abnormal Behavior Detection. International Conference on Swarm Intelligence. Springer, Cham. 2020:495-505.

8. Huang F. et al. Research on the parallelization of the DBSCAN clustering algorithm for spatial data mining based on the Spark platform. Remote Sensing. 2017(9):1301.

9. Akutota T., Choudhury S. Big data security challenges: An overview and application of user behavior analytics. Int. Res. J. Eng. Technol. 2017(4):1544-1548.

10. Touma M. et al. Framework for behavioral analytics in anomaly identification. Ground/Air Multisensor Interoperability, Integration, and Networking for Persistent ISR VIII. – International Society for Optics and Photonics. 2017(10190):101900H.