Generalized noise cancellation in a communication channel

• US 5,621,768 A
• Filed: 11/29/1994
• Issued: 04/15/1997
• Est. Priority Date: 11/29/1994
• Status: Expired due to Term

• Current Assignee: TTI Inventions A LLC (Intellectual Ventures LLC)
• Original Assignee: Bell Communications Research, Inc. (Telefonaktiebolaget LM Ericsson)
• Inventors: Lechleider, Joseph W.
• Primary Examiner(s): Chin, Stephen
• Assistant Examiner(s): PHAN, HAI
• Application Number:
  US08/346,446
• Time in Patent Office: 868 Days
• Field of Search: 375/346, 375/349, 375/350, 375/232, 381/94, 381/71, 367/901, 379/410, 379/411, 348/607, 455/296, 455/303, 455/306, 455/307
• US Class Current: 375/346
• CPC Class Codes: H04B 1/123 using adaptive balancing or compensation means adaptive filter circuits and algorithms H03H

Abstract

A generalized noise canceler for canceling noise in a channel having noise composed of correlated noise components. The noise canceler is realized with a finite number of cascade networks, each network being composed of an inner product filter is series with another filter wherein both filters have characteristics determined from a eigenvector equations expressed in terms of the channel operator and noise operator.

Claims

1. A noise canceler for processing a channel output signal composed of a signal plus noise to produce a filtered channel output signal, the noise being a linear combination of first and second noise components, the canceler comprising

a parallel arrangement of a plurality of cascade networks, each cascade network being responsive to the signal and the noise to produce an output signal, wherein each cascade network includes:

an inner product filter circuit having a filter characteristic determined from a first eigenvector equation corresponding to minimization of noise in the first noise component, in series with a second filter having a characteristic determined from a second eigenvector equation related to the first eigenvector equation, and

means for combining the outputs of the second filters to produce the filtered channel output signal.

2. The noise canceler as recited in claim 1 wherein the channel is expressed by a channel operator C and the noise is expressed by a noise operator N, and wherein the first eigenvector equation is given by N^-1 C ψ

_i =λ

₁ θ

_i and the second eigenvector equation is given by C^T N^-1 Cψ

_i =λ

₁ ψ

_i, where ψ

_i is an eigenvector representing the transmitted signal shape, λ

₁ is an eigenvalue representing the signal-to-noise ratio, and θ

_i is a vector set representing the filter characteristic.

3. A noise canceler for processing a channel output signal composed of a signal plus noise to produce a filtered channel output signal, the noise being a linear combination of first and second noise components, the canceler comprising

a parallel arrangement of a plurality of cascade networks, each cascade network being responsive to the signal and the noise and including:

a first filter having a filter characteristic determined from a first eigenvector equation corresponding to minimization of the noise in the first noise component;

a sampler responsive to the first filter; and

a second filter, coupled to the sampler, the second filter having a characteristic determined from a second eigenvector equation related to the first eigenvector equation, and

means for combining the outputs of the second filters to produce the filtered channel output signal.

4. The noise canceler as recited in claim 3 wherein the channel is expressed by a channel operator C and the noise is expressed by a noise operator N, and wherein the first eigenvector equation is given by N^-1 Cψ

_i =λ

₁ θ

_i and the second eigenvector equation is given by C^T N^-1 C ψ

_i ═

λ

₁ ψ

_i, where ψ

_i is an eigenvector representing the transmitted signal shape, λ

₁ is an eigenvalue representing the signal-to-noise ratio, and θ

_i is a vector set representing the filter characteristic.

5. The noise canceler recited in claim 4 wherein the first filter in the k^th cascade network has an impulse response given by θ

_k, and the second filter in the k^th cascade network has an impulse response given by ψ

_k.

6. A method for canceling noise in a channel output signal composed of a signal plus noise to produce a filtered channel output signal, the noise being a linear combination of first and second noise components, the method comprising the steps of

processing the signal plus noise with a plurality of cascade networks, each cascade network being responsive to the signal and the noise, wherein the processing in each cascade network includes the steps of:

processing the signal and the noise with an inner product filter circuit having a filter characteristic determined from a first eigenvector equation corresponding to minimization of noise in the first noise component, and

processing the output of the inner product filter with a second filter having a characteristic determined from a second eigenvector equation related to the first eigenvector equation, and

combining the outputs of the second filters to produce the filtered channel output signal.

7. The method as recited in claim 6 wherein the channel is expressed by a channel operator C and the noise is expressed by a noise operator N, and wherein the first eigenvector equation is given by N^-1 Cψ

_i =λ

₁ θ

_i and the second eigenvector equation is given by C^T N^-1 Cψ

_i =λ

₁ ψ

_i, where ψ

_i is an eigenvector representing the transmitted signal shape, λ

₁ is an eigenvalue representing the signal-to-noise ratio, and θ

_i is a vector set representing the filter characteristic.

8. A method for canceling noise in a channel output signal composed of a signal plus noise to produce a filtered channel output signal, the noise being a linear combination of first and second noise components, the method comprising the steps of

processing the signal plus noise with a plurality of cascade networks, each cascade network being responsive to the signal and the noise, wherein the processing in each cascade network includes the steps of:

processing the signal and the noise with a first filter circuit having a filter characteristic determined from a first eigenvector equation corresponding to minimization of noise in the first noise component,

sampling the output of the first filter at a predetermined rate to produce a sampled value, and

processing the sampled value with a second filter having a characteristic determined from a second eigenvector equation related to the first eigenvector equation, and

combining the outputs of the second filters to produce the filtered channel output signal.

9. The method as recited in claim 8 wherein the channel is expressed by a channel operator C and the noise is expressed by a noise operator N, and wherein the first eigenvector equation is given by N^-1 Cψ

_i =λ

₁ θ

_i and the second eigenvector equation is given by C^T N^-1 Cψ

_i =λ

_i ψ

_i, where ψ

_i is an eigenvector representing the transmitted signal shape, λ

₁ is an eigenvalue representing the signal-to-noise ratio, and θ

_i is a vector set representing the filter characteristic.

10. The method as recited in claim 9 wherein the first filter in the k^th cascade network has an impulse response given by θ

_k, and the second filter in the k^th cascade network has an impulse response given by ψ

_k.