Open AccessUsing CME and FCME Algorithms for the Task of Detecting Human Steps
Author(s) -
Yevhen Vistyzenko
Publication year - 2021
Publication title -
traektoriâ nauki
Language(s) - English
Resource type - Journals
ISSN - 2413-9009
DOI - 10.22178/pos.73-2
Subject(s) - algorithm , normalization (sociology) , computer science , false positive paradox , detector , database normalization , signal processing , detection theory , statistical power , amplitude , signal (programming language) , data mining , pattern recognition (psychology) , artificial intelligence , digital signal processing , mathematics , statistics , telecommunications , sociology , anthropology , computer hardware , physics , quantum mechanics , programming language
Currently, global trends in the development of perimeter security technology lead to the general miniaturization of devices and systems and the increase of their autonomy. The primary trend in developing these systems is the maximum processing and classification of signals by built-in tools. The article presents the study results on the possibility of using blind algorithms to calculate the level of operation of the threshold detector CME and FCME in detecting human steps in seismic perimeter security detectors using low-power computing modules. The algorithms were tested on actual data recorded during the experiments. As a result of applying algorithms for the seismic signal envelope, the human steps of the probability of false positives for the CME and FCME algorithms were 23% and 10%, respectively. Neutralizing the signal trend allowed obtaining values of false alarms of 16% for the CME algorithm and 7% for the FCME algorithm, and normalization of the signal amplitude within one analysis interval allowed obtaining the probability of false alarms at 0% for both algorithms. The obtained results give complete information when choosing the type of algorithm depending on the input data. In detecting seismic data by an autonomous sensor, it is most appropriate to use the SME algorithm with pre-normalization of the signal amplitude due to less computational complexity.