Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
2000-01-21
2003-09-16
Chin, Stephen (Department: 2634)
Pulse or digital communications
Equalizers
Automatic
C375S226000, C375S269000, C375S273000, C375S279000, C375S280000, C375S283000, C375S325000, C375S324000
Reexamination Certificate
active
06621863
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an equalizer for use in a receiver receiving a signal modulated by a method such as quadrature phase-shift keying (QPSK), multi-level quadrature amplitude modulation (m-QAM), or multi-level vestigial sideband (m-VSB) modulation.
QPSK, m-QAM, and m-VSB are digital modulation methods employed in various types of communication systems, including digital telephone and television systems. QPSK modulates a pair of carrier signals having identical frequencies and mutually orthogonal phase, generating a combined carrier signal with a phase that shifts in steps of ninety degrees (90°). In m-QAM, each of the two orthogonal carrier signals is modulated on multiple discrete amplitude levels, generating a combined carrier that varies in both amplitude and phase. In m-VSB, a single carrier signal is modulated at multiple discrete levels.
The modulated carrier signal is transmitted over a channel that distorts the amplitude and phase of the signal. The function of an equalizer is to correct the distortion, so that accurate data can be recovered from the demodulated signal. The distortion is removed by filtering the demodulated signal, using a digital filter. The coefficients in the digital filter are updated according to an error signal representing the residual error present in the filtered signal, so as to minimize the residual error. The performance of the equalizer depends strongly on the way in which the error signal is obtained.
A first type of error signal, referred to below as an amplitude error signal, is equal to the difference between the filtered signal and the recovered data values. This error signal is ideal when the recovered data values are correct, but is less useful when the data values are unreliable. Unreliable data values occur when there is large channel distortion, when the received signal is incompletely demodulated, and in the early stages of convergence, when the filter coefficients are still far from their optimal values.
A second type of error signal, referred to below as a squared envelope error signal, is obtained by comparing the squared absolute value of the received signal with a dispersion factor derived from the expected value of the squared absolute value. This error signal is useful during the initial stages of convergence, when the expected amplitude of the filtered signal is known, at least in a statistical sense, even though the recovered data values are unreliable. This error signal is also useful under the above-mentioned conditions of large channel distortion and incomplete demodulation. Under more favorable conditions, however, once the filter coefficients have converged and the amplitude error is small, the squared envelope error is virtually useless for updating the filter coefficients.
A third type of error signal is obtained by adding the above two error signals together, after multiplication by appropriate fixed constants. Since the squared envelope error is more useful when the amplitude error is large than when the amplitude error is small, the squared envelope error is also multiplied by the absolute value of the amplitude error. As a result, during initial convergence and under other conditions that produce large amplitude errors, the enhanced squared envelope component of the error signal guides the filter coefficients toward their correct values. As the filter coefficients approach convergence, the squared envelope component diminishes in weight, and the amplitude error component operates to complete the convergence process and maintain the converged state.
This type of combined error signal is not entirely satisfactory, however, for the following reasons.
In the early convergence stages, or under other adverse conditions, although the squared envelope component of the error signal is enhanced, the amplitude error component is also large, and its reliability is low. Incorrect amplitude error values, caused by incorrect data values, tend to slow the convergence of the filter coefficients.
After the filter coefficients have converged, under favorable conditions, although the squared envelope error component of the error signal is reduced, it does not disappear, and its continued presence tends to perturb the filter coefficients. In particular, the equalizer becomes overly sensitive to channel noise, and the residual error after convergence is far higher than the theoretical limit.
SUMMARY OF THE INVENTION
An object of the present invention is to provide an equalizing apparatus that combines rapid convergence with low residual error.
Another object is to obtain an appropriate error signal under a variety of conditions.
Another object is to provide an equalizing apparatus that is compatible with different modulation methods.
In the invented equalizing apparatus, a digital filter filters an input signal to correct channel distortion. A data decision unit obtains a data signal from the filtered signal, the data signal having values in accordance with a signal modulation method. An error signal generator receives the filtered signal and the data signal and generates an error signal for use in updating filter coefficients in the digital filter. The error signal generator comprises:
an amplitude error signal generator generating an amplitude error signal from the filtered signal and the data signal;
a squared envelope error signal generator generating a squared envelope error signal from the filtered signal;
a weighting coefficient calculation unit calculating a weighting coefficient from an absolute value of the amplitude error signal;
an amplitude error signal weighting unit weighting the amplitude error signal according to a first fixed constant and the calculated weighting coefficient;
a squared envelope error signal weighting unit weighting the squared envelope error signal according to a second fixed constant and the calculated weighting coefficient; and
an adder adding the weighted amplitude error signal and the weighted squared envelope error signal to obtain the error signal output by said error signal generator.
By weighting both the amplitude error signal and the squared envelope error signal, the error signal generator is able to generate an error signal in which the amplitude error component is reduced or absent when the amplitude error is large, and in which the squared envelope error component is completely absent when the amplitude error is small. This error signal leads to comparatively fast convergence of the filter coefficients under conditions producing large amplitude error, and to small residual error under conditions producing small amplitude error.
The invented equalizing apparatus may also include a parameter selector selecting parameters such as the first and second fixed constants according to the signal modulation method, enabling the apparatus to be used with a variety of modulation methods.
REFERENCES:
patent: 5230077 (1993-07-01), Raith
patent: 5311546 (1994-05-01), Paik et al.
patent: 5517530 (1996-05-01), Gardner
patent: 5524126 (1996-06-01), Clewer et al.
patent: 5777692 (1998-07-01), Ghosh
patent: 6327302 (2001-12-01), Shen
Benvenis et al., Blind Equalizers, IEEE Transactions on Communications, vol. Com-32, No. 8, 871-883 (1984).
Deshpande, Fast Recovery Equalization Techniques for DTV Signals, IEEE Transactions on Communications, vol. 43, No. 4, 370-377 (1997).
Birch Stewart Kolasch & Birch, LLP.
Chin Stephen
Mitsubishi Denki & Kabushiki Kaisha
Munoz Guillermo
LandOfFree
Equalizing apparatus combining amplitude error with squared... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Equalizing apparatus combining amplitude error with squared..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Equalizing apparatus combining amplitude error with squared... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3085388