The Denoising And Curve Analysis Research In The Signal Of COTDR

With the rapid development of Optical Fiber Communication, Network maintenance is essential. As an important tool of optical fiber maintenance, Optical Time Domain Reflectometer (OTDR) can analyze the detection signal, effectively reflect the breakpoint information, and determine the location and type of the breakpoint. On this basis, Coherent Optical Time Domain Reflectometer (COTDR) is developed, and it receive the returned signal by the way of coherent detection, which greatly enhance the dynamic range of the detection signal and make COTDR can detect long-distance and multi-span fiber links such as submarine cable system. However, as the impact of long-distance optical fiber transmission and EDFA repeater, the detection signals recovered by COTDR are with strong noise, and the detection curve can not effectively reflect the state of optical information. Therefore, denoising and curve analysis about COTDR detection signals become very important.In this article, OTDR detection signals are introduced and modeled firstly. On this basis, an improved wavelet threshold denoising algorithm is proposed, which is better able to enhance snr of the signal and dynamic range of the curve than other denoising algorithms; Secondly, COTDR detection signals are analyzed under strong noise interference and the combination algorithm containing the time-domain average denoising and the above improved wavelet threshold denoising is proposed to deal with the real data; Finally, according to the particularity of COTDR detection signal, the article finish appropriate curve analysis, including breakpoint location and slope analysis. The above analysises are all proved through simulation. This major work includes:(1) The working principle of OTDR and the detection curve characteristics are researched, two OTDR signals modeled, and an improved wavelet threshold denoising algorithm is proposed to deal with the signals under strong noise.(2) Multi-span COTDR detection signals are researched, the time-domain average denoising and the improved wavelet threshold denoising are proposed to joint processing the real data under strong noise and the effect is good.(3) Finish appropriate curve analysis after COTDR detection signals denoising, using wavelet modulus maxima to finish breakpoint location of the detection curve, and using the least square method to finish slope analysis of the curve.