Back to results
US 8,077,765 B1Granted
Mixed minimum mean square error/zero-forcing FIR filter adaptation
AssigneeMarvell International Ltd.
InventorPanu Chaichanavong
Life of this patent
Estimating the term…
Abstract
A system and method of adapting a FIR filter with a mixed minimum-mean-square-error/zero-forcing adaptation is disclosed. A channel response module attempts to approximate a noiseless component of the channel response. The output of the channel response module is utilized to adapt a FIR filter module. In some embodiments, a combination of the output of the channel module and the noiseless channel output is utilized to adapt the FIR filter. In some embodiments, a second FIR filter module is utilized to process the noiseless channel output, which is then compared to the target response to generate an error signal, which may be used to adapt both the first and second FIR filter modules.
Claims36
A system for adapting an equalizer, the system comprising:
- a channel module that receives input data and outputs a channel output comprising a noiseless component and a noise component;
- a filter module that filters the channel output to obtain a filter sample;
- a detector module that processes the filter sample to estimate the input data;
- a target response module that processes the estimated input data to obtain a target response output, wherein the target response output comprises a desired response of the filter module;
- and
- a channel response module that processes the estimated input data to obtain a noiseless channel output, wherein the noiseless channel output comprises an approximation of the noiseless component of the channel output,
- wherein the filter module is adapted based on the channel output, the filter sample, the target response output, and the noiseless channel output,
- wherein the filter module is adapted based on an error signal comprising a difference between the filter sample and the target response output, and
- wherein the filter module is adapted based on a product of i) the error signal and ii) a combination of the channel output and the noiseless channel output.
6 more elements
A system for adapting an equalizer, the system comprising:
- a channel module that receives input data, and outputs a channel output comprising a noiseless component and a noise component;
- a filter module that filters the channel output to obtain a filter sample;
- a detector module that processes the filter sample to estimate the input data;
- a target response module that processes the estimated input data to obtain a target response output, wherein the target response output comprises a desired response of the filter module;
- and
- a channel response module that processes the estimated input data to obtain a noiseless channel output, wherein the noiseless channel output comprises an approximation of the noiseless component of the channel output,
- wherein the filter module is adapted based on the channel output, the filter sample, the target response output, and the noiseless channel output, and
- wherein the noiseless channel output is based on the equation:
- r′k(ak−n . . . ak+m)=r′previous(ak−n . . . ak+m)+γ(rk−r′previous(ak−n . . . ak+m)),
- where (ak−n . . . ak+m) comprises a data set comprised of the estimated input data at times k−n through k+m, r′k(ak−n . . . ak+m) comprises the noiseless channel output corresponding to (ak−n . . . ak+m) at time k, r′previous(ak−n . . . ak+m) comprises the noiseless channel output corresponding to (ak−n . . . ak+m) immediately preceding time k, rk comprises the channel output at time k and γ comprises a constant.
7 more elements
A method for adapting an equalizer, the method comprising:
- processing input data to obtain a channel output comprising a noiseless component and a noise component;
- filtering the channel output with a filter module to obtain a filter sample;
- processing the filter sample to estimate the input data;
- processing the estimated input data to obtain a target response output that comprises a desired response of the filter module;
- processing the estimated input data to obtain a noiseless channel output that comprises an approximation of the noiseless component of the channel output;
- and
- adapting the filter module based on the channel output, the filter sample, the target response output, and the noiseless channel output,
- wherein the filter module is adapted based on an error signal comprising a difference between the filter sample and the target response output, and
- wherein the filter module is adapted based on a product of i) the error signal and ii) a combination of the channel output and the noiseless channel output.
6 more elements
A method for adapting an equalizer, the method comprising:
- processing input data to obtain a channel output comprising a noiseless component and a noise component;
- filtering the channel output with a filter module to obtain a filter sample;
- processing the filter sample to estimate the input data;
- processing the estimated input data to obtain a target response output that comprises a desired response of the filter module;
- processing the estimated input data to obtain a noiseless channel output that comprises an approximation of the noiseless component of the channel output;
- and
- adapting the filter module based on the channel output, the filter sample, the target response output, and the noiseless channel output,
- wherein the noiseless channel output is based on the equation:
- r′k(ak−n . . . ak+m)=r′previous(ak−n . . . ak+m)+γ(rk−r′previous(ak−n . . . ak+m)),
- where (ak−n . . . ak+m) comprises a data set comprised of the estimated input data at times k−n through k+m, r′k(ak−n . . . ak+m) comprises the noiseless channel output corresponding to (ak−n . . . ak+m) at time k, r′previous(ak−n . . . ak+m) comprises the noiseless channel output corresponding to (ak−n . . . ak+m) immediately preceding time k, rk comprises the channel output at time k and γ comprises a constant.
7 more elements
Dependencies are inferred from the claim text.