Discriminant approach to detect anomalies using Markov sequences

A.V. Skatkov, A.A. Bryukhovetskiy, D.V. Moiseev, Yu.E. Shishkin

 1Sevastopol State University, RF, Sevastopol, Universitetskaya St., 33

Email: dmitriymoiseev@mail.ru

2Institute of Natural and Technical Systems, RF, Sevastopol, Lenin St., 28

DOI: 10.33075/2220-5861-2019-4-43-49

UDC 681.3


     The work describes a discriminant approach and a program system for detecting anomalies of processes occurring in the ecosystem of the water area based on the Markov model. The conditions and results of a statistical simulation experiment are presented in order to compare the reliability of the analytical and simulation models and determine the control and warning boundaries for making decisions on the presence of anomalies. To solve the problem of detecting process anomalies according to the monitoring of the water area, an adaptive method based on a discrete wavelet decomposition of observation data and a statistical detection algorithm is proposed. To adapt the wavelet transform to the task of identifying process anomalies according to monitoring data, it is proposed to use the sliding window method.

     The essence of the discriminant approach is to compare two matrices of transition probabilities constructed by the analytical and simulation models. Using the matrix of the analytical model, we can determine the matrix of final probabilities. This will be a vector depending on the number of modeling steps. According to the discriminant approach, matrices of transition probabilities are compared, the difference between them is estimated according to the element maximum difference criteria and if it is modulo larger than the border set by the expert, then the matrices are considered different, this is the critical controlled quantity decision threshold. The values ​​of the environmental parameters observed in such transitional states are designated as the control and warning boundaries of decision making.

     The development of an intelligent technology for detecting anomalies in the state of ecosystems in the Sevastopol aquatorium based on the application of proposed approach will lead to an increase in the validity, reliability, reliability and efficiency of decision support processes for assessing the ecological state of the environment.

Keywords: simulation modeling, anomalies, water area monitoring, anomaly detection, Markov model, wavelet decomposition.

To quote, follow the DOI link and use the Actions-Cite option or copy:
[IEEE] A.V. Skatkov, A.A. Bryukhovetskiy, D.V. Moiseev, and I. E. Shishkin, “DISCRIMINANT APPROACH TO DETECT ANOMALIES USING MARKOV SEQUENCES,” Monitoring systems of environment, vol. 4, pp. 43–49, Dec. 2019.

Full text in PDF(RUS)


    1. Ecological bases of management of natural and technical systems / Under the editorship of M.P. Fedorov. SPb.: Publishing of the Polytechnic University, 2007. 506 p.
    2. Main sources of marine pollution in the Sevastopol region / E.I. Ovsiany, A.S. Romanov, Р.Y. Minkovskay, et al // Environmental safety of coastal and shelf zones and integrated use of shelf resources. Sevastopol: ECOSI-Hydrophysics, 2001. P. 138–152.
    3. Skatkov A.V., Bruhovetsky A.A., Moiseev D.V. Intelligent monitoring system for solving large-scale scientific problems in cloud computing environments // Information and control systems. 2017. № 2 (87). P. 19–25.
    4. E.N. Maschenko, U.P. Nunkina. Approach to data virtualization of multi-dimensional object monitoring systems // Monitoring systems of environment – 2016 / Abstracts of the International scientific and technical conference. Sevastopol, October 24-27 2016 г.: INTS. 2016. P. 70.
    5. U.V. Doronina, V.O. Rybovay. Development of a method for restructuring environmental monitoring information systems based on the analysis of their implementation options // Monitoring systems of environment – 2016 / Abstracts of the International scientific and technical conference. Sevastopol, October 24-27 2016: INTS. 2016. С. 64.
    6. Bruhovetsky A.A., Skatkov A.V., Shishkin Y.E. Modeling of anomaly detection processes in complex structured monitoring data // Monitoring systems of environment. Sevastopol: INTS. 2017. Issue 9 (29). P. 45–49.
    7. Shishkin Y.E., Skatkov A.V. Information technologies for detecting anomalies in monitoring observations: monograph. Simferopol: ET «ARIAL». 2019. 368 p.
    8. Truhaev R.I. Models of decision-making in conditions of uncertainty. М.: «Nauka». 1981. 258 p.
    9. Sheluhin O.I., Pankrushin A.P. Estimation of reliability of detection of network traffic anomalies by discrete wavelet analysis methods // T-Comm: Telecommunications and transport. 2013. I. 7. № 10. P. 110–115.