Measurement, calibration and pre-processing of signals for single-ended subscriber line identification

Abstract

A measurement set-up is proposed to measure the one- port scattering parameter of a subscriber line at the Central Office. The measurements are performed in the time domain using periodic multi-sine signals. A pre-processing algorithm will deliver the impulse response of the one-port scattering parameter. It will de-noise and de-alias the impulse response and search for the first significant reflection. It will be shown that the processed impulse response of the one-port scattering parameter reveals the most important signal features, which would otherwise remain undetectable using classic time domain reflectometry. I. INTRODUCTION It is known that a number of channel impairments complicate the deployment of Discrete Subscriber Line services, i.e. the line attenuation, bridged taps (stubs), different wire gauges are used in one loop, crosstalk noise and RF-ingress etc. These channel impairments can cause a communication link to malfunction. Nowadays, commercial available instrumentation is based on double -ended line testing. It requires two technicians at both line extremities in order to qua ntify the loop power transfer function and noise power spectral density. The determination of the subscriber line in this way is expensive and requires the cooperation of the customer itself. Recently, the focus shifted to the single-ended line testing. The idea is to perform measurements at the Central Office only and using advanced pre-processing algorithms combined with artificial intelligence techniques to come to a reasonable estimate of the channel capacity in bits/s of the subscriber line. There exist sufficiently accurate cable models (4)(5) to describe a twisted pair line, so a white box approach to predict the channel capacity is appropriate because in that way it can handle a priori knowledge and explain every feature detected in the excitation response. Due to the diversity and complexity of the measurement and the detection of features in the response of the subscriber line on a stimulus, the approach will be multidisciplinary: it includes a one -port measurement, after which the data needs pre- processing before being fed to the loop topology classification (7) and the loop identification part using the physical cable models (2). This paper will treat the measurement set-up used to qualify any subscriber loop and the further pre-processing of the data. Excitation signals will be injected in the subscriber loop and the frequency response function of the one-port scattering parameter will be determined. The obtained data will be pre -processed in order to derive a valid impulse response, which is fu rther used by the loop topology classification. It will be demonstrated that the processed impulse response of a subscriber line contains well- pronounced reflections despite its high attenuation and high dispersion.