The key technology and main difficulty for optical fiber intrusion pre-warning systems\n(OFIPS) is the extraction of harmful-intrusion signals. After being processed by a phase-sensitive\noptical time-domain reflectometer (-OTDR), vibration signals can be preliminarily extracted.\nGenerally, these include noises and intrusions. Here, intrusions can be divided into harmful and\nharmless intrusions. With respect to the close study of signal characteristics, an effective extraction\nmethod of harmful intrusion is proposed in the paper. Firstly, in the part of the background\nreconstruction, all intrusion signals are first detected by a constant false alarm rate (CFAR). We then\nreconstruct the backgrounds by extracting two-part information of alarm points, time and amplitude.\nThis ensures that the detection background consists of intrusion signals. Secondly, in the part of the\ntwo-dimensional Kolmogorov-Smirnov (K-S) test, in order to extract harmful ones from all extracted\nintrusions, we design a separation method. It is based on the signal characteristics of harmful\nintrusion, which are shorter time interval and higher amplitude. In the actual OFIPS, the detection\nmethod is used in some typical scenes, which includes a lot of harmless intrusions, for example\nconstruction sites and busy roads. Results show that we can effectively extract harmful intrusions.
Loading....