Intelligent intrusion detection for optical fiber perimeter security system based on an improved high efficiency feature extraction technique
Zhenshi Sun,Zheng Guo
DOI: https://doi.org/10.1088/1361-6501/ad1b9f
IF: 2.398
2024-01-06
Measurement Science and Technology
Abstract:The automated analysis of optical fiber vibration sensing data has been highly demanded in engineering applications. Therefore, intrusion analysis, which aims at detecting, recognizing, and classifying intrusions, holds great importance for optical fiber vibration sensing. In this work, an intelligent intrusion detection scheme employing an improved high-efficiency feature extraction technique and utilizing a dual Mach-Zehnder interferometer (DMZI)-based optical fiber perimeter security system is proposed. So, the DMZI-based perimeter security system in practical settings can be successfully established. Specifically, time-frequency feature vectors with nine features are firstly constructed using a maximal overlap discrete wavelet transformation approach and a zero crossing rate method. Then, the feature vectors are classified into corresponding categories using a radial basis function neural network. The effectiveness of the proposed scheme has been validated using six types of human intrusions, such as knocking, climbing, waggling, cutting, crashing and kicking the fence. The results show that the given intrusions can be accurately and rapidly recognized by the proposed scheme. The average recognition rate of 95.0% is achieved, and the average processing time for each sample data is only 0.033 s, which is significantly lower than the sampling interval (0.3 s) in our experiment. It is believed that the proposed scheme holds promising potential in the field of optical fiber perimeter security systems.
engineering, multidisciplinary,instruments & instrumentation