Coded data generation or conversion – Analog to or from digital conversion – Analog to digital conversion
Reexamination Certificate
1998-04-16
2001-09-18
Young, Brian (Department: 2819)
Coded data generation or conversion
Analog to or from digital conversion
Analog to digital conversion
C341S126000
Reexamination Certificate
active
06292126
ABSTRACT:
BACKGROUND OF THE INVENTION
A. Field of Invention
The present invention generally pertains to analog to digital converters, and more particularly pertains to quantizers.
B. Description of the Background
Communications systems frequently use high frequency carrier signals that are modulated by information for transmission from a transmitter to a receiver. On the receiver end, the modulated signal is demodulated to retrieve the information that has been transmitted over the transmission medium. The transmission medium may, for example, constitute free space, electrical wires such as coaxial cables and telephone wires, fiber optics, acoustical transmission media, etc. Various types of waves can be used to transmit these modulated signals, including electromagnetic waves, magnetic waves, acoustical waves, etc.
Significant noise interference may occur as a result of the transmission medium. For example, atmospheric interference may cause noise of various types in the transmission of electromagnetic waves over free space. In wire transmission systems, such as coaxial cable or twisted pair networks, active components may cause nonlinear effects that can result in significant noise and/or distortion of the transmitted signal. Additionally, in transmitting signals such as quadrature amplitude modulation signals (QAM), a slight error in the phase of the carriers at the demodulator will not only result in the possible loss of the signal, but may also lead to interference between the two channels. Similar difficulties arise when the local frequency is in error. In addition, unequal attenuation of the upper and lower sidebands during transmission also leads to cross-talk or co-channel interference. Other types of noise can also occur, such as attenuation distortion, envelope delay distortion, incidental phase disturbances, including phase jitter, phase wobble, and phase hits that cause the transmitted signal to experience acceleration, slowdown, or even sudden discontinuities. Other sources of noise and distortion may include impulse noise and background noise, thermal noise, harmonic distortion and intermodulation distortion.
The information that is transmitted by modulating a carrier may comprise digital signals, including multilevel digital signals. Difficulties frequently arise in the ability to correctly quantize an analog signal to retrieve the correct digital signal at the receiver. In order to correctly quantize a multilevel signal, decision thresholds must be established to distinguish between different levels of the digital signal.
A typical approach for establishing the appropriate level of a decision threshold in a multilevel signal is to set the decision threshold at a midway point between the levels of the transmitted signal level. This is based on the assumption that noise and other distortion will equally affect the signal regardless of the amplitude or phase of the signal. Assuming that the received signals are equally as likely to be affected by the noise or distortion, on average the smallest possible error will occur by placing the threshold halfway between the two levels of the transmitted signal.
However, when nonlinear distortions occur and nonlinear noise affects the received signal, the optimum level for setting the decision threshold does not fall halfway between the levels of the transmitted signal. For example, amplitude-dependent noise will cause a signal at a higher level to cross over the decision threshold that is placed halfway between two levels more often than a lower amplitude signal, which will not have as many excursions across the decision threshold. In other words, since the larger amplitude signal will be more greatly affected by noise, the larger amplitude signal will have more frequent occurrences when it crosses over the threshold than a lower amplitude signal because the noise is more severe at a higher amplitude state.
It is therefore desirable to provide quantizer which adaptively places the decision threshold at a level that minimizes the overall error rate.
SUMMARY OF THE INVENTION
The present invention overcomes the disadvantages and limitations of the prior art by providing a quantizer that adaptively determines an optimum decision threshold by locating a decision threshold level such that it is equally likely that a disturbance will cause a wrong decision. The present invention provides a way of adaptively determining the location of the decision threshold to minimize the overall error rate by determining observed signal strengths in a system.
The present invention may therefore comprise a method of quantizing an analog signal using an optimum decision threshold comprising the steps of, slicing the analog signal into a series of discrete signals having indices that are representative of the amplitude of the analog signal, generating histograms from the indices indicative of the number of discrete signals for each of the indices, generating optimum reconstruction levels by using the histograms to determine minimum means squared errors, and generating the optimum decision thresholds by using the optimum reconstruction levels to determine the average of the optimum reconstruction levels.
The present invention may further comprise a quantizer that adaptively determines a value of an optimum decision threshold for quantizing an analog signal to minimize the overall quantizing error rate by observing signal strengths of the analog signal comprising, a histrogram generator that generates histograms that indicate the signal strengths of the analog signal, an optimum reconstruction level generator that uses the histograms to generate optimum reconstruction levels by determining minimum means squared errors, and an optimum decision threshold generator that uses the optimum reconstruction levels to generate optimum decision thresholds by determining the average of the optimum reconstruction levels.
The present invention therefore is designed to obtain the same error rate for different level signals independently of whether noise or other distortions nonproportionately affect the different levels. Additionally, the present invention provides symmetric decision thresholds if the noise is linear, and nonsymmetric decision thresholds when the noise is nonlinear. The use of the present invention therefore does not degrade the performance of the quantizer when nonlinear effects are not present. Hence, there are no tradeoffs in using the present invention. Also, significant decreases in the error rate can be achieved using the present invention. The nonlinear nature of noise and other distortions that can affect a transmitted signal can significantly affect the error rate of a quantized signal. An adaptive modification of the location of the decision threshold can provide significant decreases in the error rate. For example, a small percentage change in the level of the decision threshold may reduce the error rate by several hundred percent. Accordingly, the present invention can provide a substantial benefit in that it provides the ability to correctly quantize an analog signal with a high degree of accuracy.
REFERENCES:
patent: 3931596 (1976-01-01), Gersho et al.
patent: 4381428 (1983-04-01), Kolesar et al.
patent: 4386366 (1983-05-01), Mori
patent: 4470146 (1984-09-01), Yatsuzuka et al.
patent: 5295155 (1994-03-01), Gersbach et al.
patent: 5337373 (1994-08-01), Marandici et al.
patent: WO 88/10544 (1998-06-01), None
“Digital Coding of Waveforms, Principles and Applications to Speech and Video”, by N.S. Jayant—Peter Noll, Prentice-Hall, Inc., pp. 129-135.
1991 Intl. Symposium On Ckts. And Sys., vol. 1/5,Signal Image and Video Process., Jun. 11-14 1991, pp. 340-343, “On The Qauntization Error of Max Quantizer”, Teiji Ohata.
Chelehmal Majid
Prodan Rich S.
Cable Television Laboratories
Cochran William W.
Cochran & Collins LLP
Young Brian
LandOfFree
Quantizer that uses optimum decision thresholds does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Quantizer that uses optimum decision thresholds, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Quantizer that uses optimum decision thresholds will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2509619