Research On Events Analysis Algorithms In Optical Time Domain Reflectometry

Optical Time Domain Reflectometry(OTDR) is a practical technology based on fiber backscattering measurement, which detects backwards signal with type like Rayleigh, Raman, Brillouin Scattering or Fresnel Reflection, when optical signal transmits, to obtain characteristics of fiber links, is widely used for online monitoring and fault localization of Passive Optical Network(PON). With increasing complexity of PON, it is extremely important for us to develop events analysis algorithms of OTDR data with high accuracy and efficiency, in order to reduce time consumption and manpower expense for network maintenance.Traditional algorithms for events localization of OTDR data includes the two point method combining with least mean square method, GT, and WT. The two point method combing with least mean square method is a relatively mature example among the early researches of events analysis. It has advantages of simple principles and easy programmable realization. However, with its sensitivity to SNR, this method has low accuracy for distinguishing different events.Be aimed at hereat, Gabor Transform, which has analysis ability in timefrequency field, is capable of improving precision for detecting events. However, with fixed window function, its spatial resolution is limited. By contrast, Wavelet Transform has multi-resolution analysis by with discrete dyadic wavelet function, which seems to more promising. However, with high complexity, it is not unsatisfied for real PON monitoring system.Therefore, based on profound researches on performance of different algorithms for events detection in OTDR data, we proposed the template matching method combining STFT for analyzing OTDR events. Considering tradeoff between computing complexity and algorithm performance, this method has relatively high accuracy and noise tolerability.On the other hand, traditional analysis of OTDR events is mostly based on a point to point fiber link, while the real PON has P2 MP structure. Therefore, besides higher requirements for event localization, it also calls for reasonable fault alarm algorithms to automatically report the geographical location of the fault. In this thesis, the fault alarm algorithm is proposed, which periodically operates referenced and measured trace and set priority for different fault types, then finish the alarm process aimed at fault like breakpoints, crack, connector loose, slice loss and so on.Finally, based on a hardware testing platform that made by our lab, we designed a software system for OTDR measurements, using Code::Blocks IDE. It contains major capabilities like realizing the C/C++ algorithms, graphic user interfaces with wxWidgets, trace display and management, fault alarm information. The system has user-friendly interfaces, reasonable workflow arrangements, robust and practical functionality, which is capable of satisfying needs for PON online monitoring system.