Filter arrangement implementing a least squares method utilizing rotation matrices
AssigneeU.S. Philips Corporation
InventorMaurice Bellanger
Life of this patent
Estimating the term…
Abstract
A filter for implementing a least squares method receives samples x(n+1) and produces error signals e(n+1) based on reference signals y(n+1). Filtered samples y(n+1) are available at an output 28. The filtered samples are calculated based on: angle functions .theta..sub.i (i=1 to n) that form part of rotation matrices; a gain vector G.sub.N (n+1) that is based on the rotation matrices; and the error signals. Because the filtered samples are available the filter is suitable for use in an equalizer in which the reference signal is formed by from the filtered signals, as well as in other applications such as echo cancelers.
Claims10
Filter arrangement implementing a least squares method comprising
- input means for receiving samples x(n+1) ,
- evaluation means for evaluating angle functions .theta..sub.i =(i=1 to N), which angle functions form part of at least one of a plurality of rotation matrices,
- evaluation means for evaluating a gain vector G.sub.N (n+1) based on the rotation matrices,
- processor means for processing reference samples Y(n+1),
- evaluation means for evaluating an error signal e(n+1) in cooperation with the processor means,
- means for evaluating filtered samples y(n+1) recurring as a function of the angles .theta..sub.i, of the gain vector, and of the error signal, whereby the filtered samples are accessible,
- wherein n, i, and N are integers.
Digital transmission system comprising a transmitter connected to a receiver by a transmission channel, this receiver comprising a filter arrangement implementing a least squares method comprising
- input means for receiving samples x(n+1),
- evaluation means for evaluating angle functions .theta..sub.i =(i=1 to N), which angle functions form part of at least one of plurality of rotation matrices,
- evaluation means for evaluating a gain vector G.sub.N (n+1) based on the rotation matrices,
- processor means for processing reference samples y(n+1),
- evaluation means for evaluating an error signal e(n+1) in cooperation with the processor means,
- means for evaluating filtered samples y(n+1) recurring as a function of the angles .theta..sub.i, of the gain vector, and of the error signal, whereby the filtered samples are accessible, and
- a decision element for deriving the reference samples from the filtered samples,
- where n, i, and N are integers.
A digital filter for implementing a least squares calculating comprising
- input means for receiving signals x(n+1);
- input means for receiving reference signals y(n+1);
- output means for supplying error signals e(n+1);
- processing means for calculating
- a plurality of matrices at least one of which includes angle functions .theta..sub.i (i=1 to N);
- a gain vector G.sub.N (n+1) based on the rotation matrices;
- the error signals e(n+1), based on the reference signals;
- filtered signals y(n+1) as a function of the angles .theta..sub.i, the gain vector G.sub.N (n+1), and of the error signals e(n+1);
- output means for supplying the filtered signals y(n+1).
Dependencies are inferred from the claim text.