Telephonic communications – Echo cancellation or suppression – Using digital signal processing
Reexamination Certificate
1999-01-06
2004-06-01
Harvey, Minsun Oh (Department: 2644)
Telephonic communications
Echo cancellation or suppression
Using digital signal processing
C379S406090, C708S322000, C381S071110, C381S071120, C370S465000
Reexamination Certificate
active
06744886
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to the field of signal processing and in particular to adaptive filtering, and specifically to an adaptive filter utilizing a new fast converging algorithm that is well suited for network echo cancellation applications in a telephone network.
BACKGROUND OF INVENTION
Adaptive filters operate on a supplied signal in a prescribe manner such that a desired output signal is generated. Typically, adaptive filters generate a transfer function according to an algorithm that includes updating of the transfer function characteristic in response to an error signal. In this manner, the filter characteristic is optimized to produce a desired result.
When used in an echo canceller, an adaptive filter is used to generate an echo path estimate that is updated in response to an error signal. Echoes commonly occur because of imperfect coupling of incoming signals at the 4-to-2 wire junctions in communications systems. The echoes typically result because the impedance of the 2-wire facility is imperfectly balanced in the 4-to-2 wire junction, causing the incoming signal to be partially reflected over an outgoing path to the source of incoming signals.
Adaptive echo cancellers have been employed to mitigate the echoes by adjusting the transfer function (impulse response) characteristic of an adaptive filter to generate an estimate of the reflective signal or echo and, then, subtracting it from the outgoing signal. The filter impulse response characteristic and, hence, the echo estimate is updated in response to continuously updated samples of the outgoing signal for more closely approximating the echo to be cancelled.
Additionally, double talk detectors (DTD) are generally used to disable adaptation during double talk conditions, that is when both the near end and far end party to a telephone conversation taking place across a telephone line speak simultaneously. Because the double talk detector cannot disable adaptation at the precise instant the double talk occurs, a number of samples occur in the delay period between the commencement of actual double talk and the double talk detector shutting down adaptation. Samples taken during this delay period can and often do perturb the echo path estimate considerably. Also, characteristic changes in the system due to environmental or other causes can also perturb the filtering.
Thus, although prior art arrangements of adaptive filters perform satisfactorily in some applications, often it is impossible to simultaneously achieve both sufficiently fast response and the ability to resist perturbations caused by samples occurring prior to the cessation of adaptation.
SUMMARY OF THE INVENTION
The present invention overcomes shortcomings in the prior art. Specifically, the present invention is based upon a recognition that fast converging algorithms such as normalized least mean squares (NLMS), proportional normalized least mean squares (PNLMS) and a version which combines the two, specifically PNLMS++, are insufficient of themselves to resist perturbations in the echo path estimate resulting from disruptive samples collected in the echo canceller prior to the cessation of adaptation by the double talk detector. In addition to the fast conversion algorithms just discussed, there is another, known an affine projection algorithm (APA). In the present invention, a proportional affine projection algorithm (PAPA) is presented which combines APA and PNLMS++. This algorithm converges very fast, and unlike PNLMS++, is not as dependent on the assumption of a sparse echo response.
REFERENCES:
patent: 5272695 (1993-12-01), Makino et al.
patent: 5428562 (1995-06-01), Gay
patent: 5675644 (1997-10-01), Sih
patent: 5951626 (1999-09-01), Duttweiler
patent: 6108412 (2000-08-01), Liu et al.
patent: 6137881 (2000-10-01), Oh et al.
patent: 6198819 (2001-03-01), Farrell et al.
patent: 6246760 (2001-06-01), Makino et al.
patent: 6269117 (2001-07-01), Peng
Duttweiler, D. C. “Proportionate Normalized Least-Mean-Squares Adaptation in Echo Cancelers,” IEEE Transactions on Speech and Audio Processing, vol. 8, No. 5, Sep. 2000, pp. 508-518.*
Douglas, S. C. “A Family of Normalized LMS Algorithm,” IEEE Signal Processing Letters, Vol 1, No. 3, Mar. 1997, pp. 49-5.*
Liu et al. “On the Use of a Modified Fast Affine Projection Algorithm in Subband for Acoustic Echo Cancelation,” IEEE, Digita Signal Processing Workshop Proceedings, 1996, pp. 354-357.*
Tanaka et al. “Fast Projection Algorithm and its Step Size Control,” ICASSP International Conference. Acoustics, Speech, a Signal Processing, vol. 2, May 1995, pp. 945-948.*
Douglas et al. “Normalized Data Nonlinearities for LMS Adaptation,” IEEE Transactions on Signal Processing, vol. 42, No. 6 Jun. 1994, pp. 1352-1365.*
Oh et al. “A fast affine projection algorithm for an acoustic echo canceller using a fixed-point DSP processor,” ICASSP IEEE Conf. Acoustics, Speech, and Signal Processing, Apr. 1997, pp. 4121-4124.
Benesty Jacob
Gaensler Tomas Fritz
Gay Steven Leslie
Sondhi Man Mohan
Harvey Minsun Oh
Lucent Technologies - Inc.
Tran Con P.
LandOfFree
Adaptive filter utilizing proportional affine projection... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Adaptive filter utilizing proportional affine projection..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Adaptive filter utilizing proportional affine projection... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3293645