Wavelet-based Echo Identification in Time Domain Reflectograms

The automatic interpretation of reflectometry traces to identify twisted-pair loops can be considered a pattern recognition problem. Thus, any loop make-up algorithm must firstly extract features from the reflectogram before starting a classification process. This paper focuses on feature extraction through the echo identification on time domain reflectometry (TDR) measurements, which are analyzed by a multi-scale edge detection wavelet approach that identifies all singularities in a reflectogram and relate them to theirs respective echoes. From echoes, it is provided an approximation for an auto-regressive TDR trace model, furnishing features for classification.