Echo cancellation in two-wire full-duplex data transmission with estimation of far-end data components

Claim 2

An arrangement for processing samples of an incoming signal which is corrupted by echoes of a transmitted signal representing a train of transmitted near-end data symbols, said incoming signal representing a train of transmitted far-end data symbols and said samples having respective far-end data components and echo components, said arrangement comprising echo cancelling means operative in response to a stream of said transmitted near-end data symbols and a stream of adaptation error signals for generating a replica of each of said echo components, and signal processing means for producing a stream of echo compensated samples as a function of the difference between each of said samples and the replica of its echo component and for forming decisions as to the values of said transmitted far-end data symbols in response to said echo compensated signals, and means for generating an individual one of said adaptation error signals as a function of said difference between an individual one of said samples and the replica of its echo component and as a function of an estimate signal, said estimate signal being substantially equal to the far-end data component of said individual one of said samples, said generating means including reference former means for forming said estimate signal as a function of individual ones of said decisions and as a function of at least ones of said stream of adaptation error signals.

Claim 3

The invention of claim 1 wherein said reference former means is comprised of estimating means for linearly combining said individual ones of the decisions to form said estimate signal.

Claim 4

The invention of claim 2 further comprising means for generating a plurality of coefficients and wherein said estimating means performs said linear combining by multiplying said individual ones of the decisions by respective ones of said coefficients and summing the resulting products.

Claim 5

The invention of claim 3 wherein said one of said adaptation error signals is substantially equal to an error signal multiplied by a predetermined parameter, said error signal being a function of said estimate signal and a function of said difference between said individual one of said samples and said replica of its echo component.

Claim 6

The invention of claim 4 wherein said reference former means is further comprised of update means for updating the value of each of said respective coefficients as a function of said individual ones of said decisions and as a function of said error signal.

Claim 7

The invention of claim 5 wherein in said update means said value of said each of said respective coefficients is updated in response to the product of said error signal with a predetermined parameter.

Claim 8

The invention of claim 5 wherein said update means updates the value of each of said respective coefficients in response to the product of said error signal with the one of said decisions with which said each respective coefficient was multiplied.

Claim 9

An arrangement for processing a stream of samples of an incoming signal which is corrupted by echoes of a transmitted signal representing a train of transmitted near-end data symbols, said stream of samples representing a train of transmitted far-end data symbols and said samples having respective far-end data components and echo components, said arrangement comprising echo cancelling means operative in response to a stream of said transmitted near-end data symbols and a stream of adaptation error signals for generating a stream of replicas of said echo components, and signal processing means for producing a stream of echo compensated samples as a function of the difference between each of said samples and the replica of its echo component and for forming decisions as to the values of the individual transmitted far-end data symbols in said train in response to said echo compensated signals, and means for generating said stream of adaptation error signals as a function of said difference between each one of said samples and the replica of its echo component and as a function of a stream of estimate signals, said estimate signals being substantially equal to the far-end data components of associated ones of said samples, said generating means including reference former means for forming said estimate signals as a function of said decisions and said stream of adaptation error signals.

Claim 10

An arrangement for processing a stream of samples of an incoming signal which is corrupted by echoes of a transmitted signal representing a train of transmitted near-end data symbols, said stream of samples representing a train of transmitted far-end data symbols and said samples having respective far-end data components and echo components, said arrangement comprising echo cancelling means operative in response to a stream of said transmitted near-end data symbols and a stream of adaptation error signals for generating a stream of replicas of said echo components, and signal processing means for producing a stream of echo compensated samples as a function of the difference between each of said samples and the replica of its echo component and for forming decisions as to the values of the individual transmitted far-end data symbols in said train in response to said echo compensated signals, and means for generating said stream of adaptation error signals as a function of said difference between each one of said samples and the replica of its echo component and as a function of a stream of estimate signals, said estimate signals being substantially equal to the far-end data components of associated ones of said samples, said generating means including reference former means for forming said estimate signals as a function of said decisions and said stream of adaptation error signals.

Claim 11

The invention of claim 8 wherein said reference former means includes estimating means for linearly combining individual ones of said decisions to form said estimate signals.

Claim 12

The invention of claim 9 further comprising means for generating a plurality of coefficients and wherein said estimating means includes means for performing said linear combining by multiplying said individual ones of said decisions by respective ones of said coefficients and summing the resulting products.

Claim 13

The invention of claim 10 wherein each of said adaptation error signals is substantially equal to an error signal multiplied by a predetermined parameter, said error signal being a function of a respective one of said estimate signals and a function of said difference between its associated sample and the replica of the echo component of said associated sample.

Claim 14

The invention of claim 11 wherein said reference former means is further comprised of update means operative in response to said decisions and in response to a stream of update error signals proportional to said stream of adaptation error signals for updating the values of said respective coefficients.

Claim 15

The invention of claim 11 wherein update means updates the value of each respective coefficient in response to the product of an individual one of the update error signals with the one of said decisions with which said each respective coefficient was multiplied.

Claim 16

In a transmission system of the type which includes a near-end terminal and a far-end terminal, the near-end terminal being adapted to transmit to the far-end terminal over a two-way communication line near-end data signals each representing a train of near-end data symbols and to receive over said line from said far-end terminal far-end data signals each representing a train of far-end data symbols, said system further being of the type in which data signals received from the far-end terminal are corrupted by echoes of transmitted near-end data signals, apparatus at said near-end terminal for removing said echoes from said received far-end data signals comprising sampling means for producing at at least the Nyquist rate a stream of samples of said corrupted received far-end data signals, each of said samples being comprised of a far-end data component and an echo component, echo cancelling means operative in response to said near-end data symbols and a stream of adaptation error signals for generating replicas of the echo components of respective ones of said samples, signal processing means for producing a stream of echo compensated samples as a function of the difference between each of said samples and its respective echo replica and for forming decisions as to the values of said far-end data symbols from said echo compensated signals, and means for producing said adaptation error signals as a function of said difference between each one of said samples and the respective replica of its echo component and as a function of a stream of estimate signals, said estimate signals being substantially equal to the far-end data components of associated ones of said samples, said generating means including reference former means for forming said estimate signals as a function of said decisions and said stream of adaptation error signals.

Claim 17

In a transmission system of the type which includes a near-end terminal and a far-end terminal, the near-end terminal being adapted to transmit to the far-end terminal over a two-way communication line near-end data signals each representing a train of near-end data symbols and to receive over said line from said far-end terminal far-end data signals each representing a train of far-end data symbols, said system further being of the type in which data signals received from the far-end terminal are corrupted by echoes of transmitted near-end data signals, apparatus at said near-end terminal for removing said echoes from said received far-end data signals comprising sampling means for producing at at least the Nyquist rate a stream of samples of said corrupted received far-end data signals, each of said samples being comprised of a far-end data component and an echo component, echo cancelling means operative in response to said near-end data symbols and a stream of adaptation error signals for generating replicas of the echo components of respective ones of said samples, signal processing means for producing a stream of echo compensated samples as a function of the difference between each of said samples and its respective echo replica and for forming decisions as to the values of said far-end data symbols from said echo compensated signals, and means for producing said adaptation error signals as a function of said difference between each one of said samples and the respective replica of its echo component and as a function of a stream of estimate signals, said estimate signals being substantially equal to the far-end data components of associated ones of said samples, said generating means including reference former means for forming said estimate signals as a function of said decisions and said stream of adaptation error signals.

Claim 18

The apparatus of claim 14 wherein said generating means is arranged to form each of said adaptation error signals by multiplying an error signal by a predetermined parameter, said error signal being a function of said difference between an individual one of said samples and a function of the replica of its echo component and the estimate signal associated with said individual one of said samples.

Claim 19

The apparatus of claim 15 wherein said reference former means includes means for forming each of said estimate signals by multiplying individual ones of said decisions by respective ones of a set of coefficients and summing the resulting products.

Claim 20

The apparatus of claim 16 wherein said reference former means is further comprised of update means for updating the value of each of said set of coefficients in response to the product of an individual one of a stream of update error signals with the one of said decisions with which said each set coefficient was multiplied.

Claim 21

In an arrangement for processing samples of an incoming signal to form decisions a n , n=0,1,2, . . . , as to the values of data symbols represented by said signal, said samples having respective data components and echo components, said echo components being derived from an outgoing signal representing a stream of near-end data symbols, a method of generating a replica of the echo component present in each of said samples, said method characterized by the steps of storing P coefficient sequences, forming in response to individual ones of said decisions an estimate X M-D of the data component of the (M-D) th one of said samples in accordance with ##EQU3## N being a selected number, W n K (i) being the value of the K th coefficient in the i th one of said P coefficient sequences and a n being the most recently formed one of said decisions, and generating said replica in response to the combination of said (M-D) th sample, and said estimate X M-D- , and in response to ones of said data symbols, generating an updated value W n+1 K (i) of said coefficient in accordance with ##EQU4## E.sub.(M-D-P) being equal to the combination of the (M-D-P) th one of said samples, the replica of its echo component, and the estimate of its data component and γ being a predetermined parameter, and storing said updated value for use in forming subsequent estimates of said data components.

Claim 22

In an arrangement for processing samples of an incoming signal to form decisions a n , n=0,1,2, . . . , as to the values of data symbols represented by said signal, said samples having respective data components and echo components, said echo components being derived from an outgoing signal representing a stream of near-end data symbols, a method of generating a replica of the echo component present in each of said samples, said method characterized by the steps of storing P coefficient sequences, forming in response to individual ones of said decisions an estimate X M-D of the data component of the (M-D) th one of said samples in accordance with ##EQU3## N being a selected number, W n K (i) being the value of the K th coefficient in the i th one of said P coefficient sequences and a n being the most recently formed one of said decisions, and generating said replica in response to the combination of said (M-D) th sample, and said estimate X M-D- , and in response to ones of said data symbols, generating an updated value W n+1 K (i) of said coefficient in accordance with ##EQU4## E.sub.(M-D-P) being equal to the combination of the (M-D-P) th one of said samples, the replica of its echo component, and the estimate of its data component and γ being a predetermined parameter, and storing said updated value for use in forming subsequent estimates of said data components.

Claim 25

The method of claim 19 wherein said estimate signal forming step includes the step of linearly combining said individual ones of said decisions.

Claim 26

The method of claim 20 wherein said linear combining step includes the steps of generating a plurality of coefficients, multiplying said individual ones of said decisions by respective ones of said coefficients and summing the resulting products.

Claim 27

The method of claim 21 wherein said adaptation error signal producing step includes the step of multiplying an error signal by a predetermined parameter, said error signal being a function of said estimate and a function of said difference between said said individual one of samples and said replica of its echo component.

Claim 28

The method of claim 22 wherein said estimate forming step further includes the step of updating the value of each of said respective coefficients as a function of said individual ones of said decisions and as a function of said error signal.

Claim 29

The method of claim 23 wherein said updating step includes the step of multiplying said error signal with a predetermined parameter.

Claim 30

The method of claim 23 wherein said updating step includes the step of multiplying said error signal with the one of said decisions with which said each respective coefficient was multiplied.