Telephonic communications – Echo cancellation or suppression
Reexamination Certificate
2000-12-21
2004-10-12
Isen, Forester W. (Department: 2644)
Telephonic communications
Echo cancellation or suppression
C379S406020, C379S406060, C379S406080, C379S406090, C379S406110, C379S406140
Reexamination Certificate
active
06804350
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the field of echo cancellers. Echo cancellers simulate the hybrid response that is reflected back to a caller in order to cancel the reflected echo signal.
2. Related Art
Communication networks such as the Internet using Voice-over Internet Protocol, digital wireless or cellular networks, and packet switched networks are only a few of the many digital communication networks currently available to the public. Increased competition within these markets focuses technology towards achieving better quality of service, specifically in the area of echo cancellation.
Digital communication networks are affected by echo signals that are reflected back to a caller generating a voice signal. These reflected echoes can reduce the quality of service to unacceptable levels. A primary source of echo signals comes from the hybrid response from a hybrid circuit. This hybrid circuit is necessary to convert signals that are transmitted through a 4-wire public switched telephone network (hereinafter referred to as “PSTN”) to 2-wire local connections to end users. For best performance, a 4-wire connection throughout a network from the PSTN to the end user would eliminate the hybrid response. However, since most end users such as homes and businesses are wired with 2-wire local connections from the local exchange in order to reduce costs, this hybrid response must still be addressed.
As a voice signal from a first end user passes through the hybrid circuit, energy from the voice signal is leaked back to the first end user in the form of a hybrid response. This hybrid output is the echo signal. Thus, a voice signal will be reflected back in the form of an echo on top of a voice signal from a second end user. This echo significantly degrades voice quality of the signal received by the first end user.
Some form of echo cancellation is needed in order to attenuate the echo signal produced through the hybrid circuit. Where there are significant delays between end users, as discussed previously, the higher the attenuation of the echo signal is needed. Echo cancellers subtract an algorithmically generated synthetic echo signal from the incoming voice signal in order to reduce or eliminate echo. Currently, echo cancellers present in different types of systems using VoIP technology usually employ plain or normalized Least-Mean- Squares (hereinafter referred to as “LMS”) algorithms for estimating a replica of the echo and properly subtracting it.
FIG. 1
illustrates a generalized echo canceller in the prior art that uses the LMS algorithm (plain or normalized LMS algorithms) to come up with the synthetic echo, or the canceling signal. The input x(n) represents the input to the adaptive filter and is typically speech. This speech input is the same input into the hybrid circuit that produces the hybrid response
110
. The adaptive filter
120
essentially tries to model the hybrid response
110
. This is done by applying the LMS algorithm to the input signal x(n) to estimate the reflected echo signal. This estimated or synthetic signal is subtracted from the hybrid response to reduce its degrading effects.
However, the convergence of the LMS algorithm is dependent on the input signal's spectral characteristics. The LMS and normalized LMS algorithms are very sensitive to correlated signals such as voice, having a faster convergence rate for uncorrelated signals such as white noise (flat spectrum), and a slower convergence rate for non-white signals or correlated signals. Unfortunately, since voice signals are correlated, echo cancellers using plain or normalized LMS algorithms have slow convergence rates for estimating the echo signal. Thus, there is a need to improve the performance of echo cancellers in digital communication networks.
Variations of the LMS algorithm such as LMS-lattice and LMS-Newton can be used for improving convergence but they have significantly higher computational complexity at higher additional cost. This fact has prevented the use of these alternatives to the conventional LMS algorithm in commercial systems. Thus, there is a need for a cost-effective way of improving the performance of echo cancellers.
Voice-over Internet Protocol Networks are poised to enter the commercial market. VoIP Gateways act as a bridge between the local PSTN and the digital network. The VoIP gateway converts analog voice signals from a first end user into digitized data packets for transmission over a communication network such as the internet to a second end user. VoIP technology reassembles the packets and reconstructs the analog signal at the second end user. The cost can be dramatically reduced for long distance telephone charges such as those in the international long distance market.
However, massive delays between the routers through the Internet communication network require significant levels of echo cancellation. The present echo cancellers that use LMS or normalized LMS algorithms are incapable of significantly reducing the hybrid response within a short period of time. As a result, the quality of service for VoIP calls is reduced. Thus, it would be advantageous to improve the quality of service in VoIP telecommunications.
SUMMARY OF THE INVENTION
It is an object of the present invention to improve the performance of echo cancellers in digital communication networks. It is another object of the present invention to improve the quality of service in digital communication networks such as VoIP telecommunications.
Specifically, the present invention discloses a method and system for improving performance of an echo canceller with one additional filtering operation. By using information from internal variables of Code Excited Linear Prediction (CELP) based codecs in a Voice-over Internet Protocol, the rate of convergence of an echo canceller is significantly improved with minimal additional complexity.
In one embodiment of the present invention, an error signal associated with a voice signal transmitted over a digital communication network, such as the Voice-over Internet Protocol network, is filtered by a transversal filter using Linear Predictive Coefficients (LPC) coefficients associated with the voice signal. The LPC parameters provide the filtering parameters or filter transfer functions for a Filtered-X Least Mean Squares algorithm. Additionally, an adaptive filter uses the Filtered-X Least Mean Squares algorithm and the pre-filtered input to create a synthetic echo signal. This synthetic echo signal is subtracted from the echo signal to attenuate the resulting echo level.
These and other objects and advantages of the present invention will no doubt become obvious to those of ordinary skill in the art after having read the following detailed description of the preferred embodiments which are illustrated in the various drawing figures.
REFERENCES:
patent: 5651091 (1997-07-01), Chen
patent: 5912966 (1999-06-01), Ho
patent: 6209013 (2001-03-01), Gonikberg et al.
patent: 6434233 (2002-08-01), Bjarnason et al.
Cisco Technology Inc.
Isen Forester W.
Singh Ramnandan
LandOfFree
Method and apparatus for improving echo cancellation in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for improving echo cancellation in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for improving echo cancellation in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3274614