P. Boets

Estimation of the transfer function of a subscriber loop by means of a one-port scattering parameter measurement at the central office

In order to qualify a subscriber loop for xDSL transmission, the channel capacity has to be estimated, which depends on the transfer function of the network. A method is provided to estimate the transfer function of the subscriber loop only measuring the one-port scattering parameter at the central office. We consider three types of networks according to their topology: a single line, a homogeneous network with a bridged tap, and a cascade of two line sections. For each type of network a parametric model is derived of its one-port scattering parameter and transfer function based on the physical line model VUB0. The model for the scattering parameter is used to identify the network based on the corresponding measurements by means of a maximum-likelihood estimator. The estimated parameters are substituted in the transfer function model, which is needed for the capacity estimation. The proposed models and estimators are validated by measurements and simulations. For the measurements, which were performed with a network analyzer, three types of twisted-pair cables were used: British Telecom (BT), Deutsch Telekom (FT), and Belgacom.

Efficient hybrid optimization of fixed-point cascaded IIR filter coefficients

In this paper we present a hybrid optimization scheme for the determination of fixed-point coefficients of IIR filters. Tests show that the developed algorithm is very time efficient and capable of achieving a high accuracy in both magnitude and phase response.

Binder Identification by Means of Phantom Measurements

This paper presents a technique to identify if two twisted pairs (TPs) share the same binder or not, based on single-ended line measurements. If the TPs share the binder only over a certain distance, this length is determined. We first show the challenges of achieving binder identification under normal differential operation. Next, we propose an alternative by measuring two TPs simultaneously in phantom mode. The proposed approach is validated by laboratory measurements.

Calibration of a metallic access network analyzer

A calibration approach is proposed to compensate for the systematic errors when the input reflection coefficient of a metallic access network cable is obtained using single port measurements from an access network analyzer. The analyzer is a versatile instrument and uses optimal cable matching and internal balancing to achieve maximum measurement resolution. The analyzer uses test-leads connected to a fixed bridge structure, but with an adaptable balance and match impedance, and two variable synchronous acquisition channels so that the calibration can be split up into the non-parametric determination of the fixed part and a parametric determination of the variable part. The outcome of the non-parametric calibration is nine frequency dependent coefficients, and shows as an important side product the autobalance procedure. The parametric calibration will result in two 35 item sets of coefficients of rational functions in the S-domain. The parametric calibration depends on frequency domain system identification where the order of the numerator and denominator polynomials of the rational functions are not fixed.