Pulse or digital communications – Equalizers
Reexamination Certificate
1999-04-30
2003-01-28
Corrielus, Jean (Department: 2631)
Pulse or digital communications
Equalizers
C708S323000
Reexamination Certificate
active
06512789
ABSTRACT:
BACKGROUND
1. Field of the Invention
This invention relates to communication systems, transceivers, and modems and to processing of digital communication signals.
2. Description of Related Art
Emerging standards for digital communications widely use multi-carrier modulation, such as Discrete Multi-Tone (DMT) modulation, to provide high data transmission rates. Such standards include “high bit rate digital subscriber loop” (HDSL), “very fast asymmetric digital subscriber loop” (VDSL), “asymmetric digital subscriber loop” (ADSL) which is ITU, ANSI, and ETSI standard G.992.1 (or G.dmt), and light rate ADSL which is an ITU standard G.992.2 (or G.lite). Transceivers or modems that implement any of these or similar standards are sometime referred to herein as xDSL transceivers. The data communication protocols for xDSL transceivers are specified for copper wire (telephone lines) as the transportation medium between regional central offices (CO) and homes. These long copper wires cause non-uniform distortion across the spectrum of broadband communication signals. Accordingly, quadrature amplitude modulation (QAM) as used in V.32 and V.34 modem standards or pulse amplitude modulation (PAM) as used in V.90 and ISDN are not suitable modulation techniques for broadband signals on long copper wires. Instead, xDSL transceivers use DMT modulation and break a broad bandwidth channel into multiple sub-channels. With a large number of sub-channels, xDSL transceivers deliver high bit rates, when compared to ISDN or voice band modems.
Typically, each sub-channel has a bandwidth of about 4 kHz, and the G.992.1 and G.992.2 standards implement 256 and 128 sub-channels respectively. QAM independently modulates data for transmission in each sub-channel rather than modulating the full bandwidth as a whole. An advantage of DMT modulation is that the attenuation and distortion within each sub-channel are fairly constant, hence, a receiver can independently equalize each sub-channel. Moreover, one can estimate each sub-channel's transport capacity and assign an appropriate load to each sub-channel by selecting the number of modulated bits ‘b’ and the transmit power, ‘g’ for that sub-channel. The appropriate choices of bit loading and transmit power per channel enable a DMT system to efficiently utilize the channel's capacity and reduce the receiver's burden of distortion mitigation.
After loading of the N sub-channels, an inverse discrete Fourier transform (IDFT) bundles the sub-channels into a time domain signal, called a DMT symbol that a series of digital samples can represent. A redundant guard signal or cyclic extension of a fix length can be appended to each DMT symbol to help the receiver reduce intersymbol interference (ISI) between DMT symbols. To transmit the signal through the copper wire, a Digital-to-Analog Converter (DAC) converts the digital samples to an analog signal, and an amplifier (or line driver) boosts the power of the analog signal transmitted.
The receiver samples a received waveform and passes the digitized samples through a time domain equalizer (TEQ), typically an FIR filter. The TEQ reduces the channel's impulse response and at least partially corrects for distortion across the broadband channel. The main channel induced distortion in the received signal is typically due to the long impulse response of the channel. Generally, the channel's impulse response stretches over several samples (e.g., if an impulse signal is input to the channel, the output signal from the channel takes more than one sample period to settle to zero). Inter-symbol interference (ISI) occurs when the residue from the preceding DMT symbol overlaps the following DMT symbol. ISI can corrupt a received DMT symbol. Typically, the TEQ attempts to reduce the impulse response time as much as possible. With a guard signal, the TEQ has adequately reduced the impulse response time to prevent ISI between DMT symbols when the impulse response time in the output signal from the TEQ is less than the period of the guard signal. Hence, at the output from the TEQ, only the guard signal is ISI distorted and not the following DMT symbol. The receiver can discard the guard signal and use a discrete Fourier transform (DFT) to decompose the time domain DMT symbol into independent constituents (sub-channels).
A frequency domain equalizer (FEQ) and slicing (or quantization) extract the separate content (i.e., a QAM symbol) of each sub-channel. Thereafter, the N QAM symbols are decoded into their corresponding bits, and a parallel to serial converter assembles the bits to provide the information that was transmitted.
To optimize xDSL transceiver performance, efficient methods for equalizing received signals to remove intersymbol interference and channel distortion are sought.
SUMMARY
In accordance with an embodiment of the invention, a training process for a filter such as a time domain equalizer of an xDSL transceiver uses a novel method for spectral estimation of a channel. The spectral estimation determines the taps of the filter using an overdetermined set of equations based on the auto-correlation estimates of the received signal. A weighting function such as a sigmoidal function is applied to the AC coefficients, before use in the overdetermined equations, to change the relative weighting of the AC coefficients. Upon solving the equations for the taps of the filter using a fitting criterion such as the least square error criterion, the filter significantly reduces the impulse response of the channel.
In accordance with another embodiment of the invention, a receiver samples and averages values of a signal over a channel. The received signal corresponds to repeated copies of a known signal. The receiver then determines an auto-correlation sequence from the averaged sample values and corresponding values of the known signal and applies a weighting function to coefficients from the auto-correlation (AC) sequence. The weighting function can be a sigmoidal function of the correlation length or another function that provides weighting values for multiplication by corresponding AC coefficients. Applying the sigmoidal weighting function reduces the magnitudes of weighted AC coefficients at large correlation lengths relative to the weighted AC coefficients at short lengths. Solving an overdetermined system of equations containing the weighted AC coefficients identifies the taps of a digital filter for reducing the impulse response of the channel. A time-domain equalizer for a xDSL transceiver can use the digital filter to reduce impulse response and distortion from a broadband communication channel.
Another embodiment of the invention is a communication system that trains a time domain equalizer using the training processes.
REFERENCES:
patent: 4494214 (1985-01-01), Bernard et al.
patent: 5404322 (1995-04-01), Gehring
Corrielus Jean
PC-Tel, Inc.
Skjerven Morrill LLP
LandOfFree
Partial equalization for digital communication systems does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Partial equalization for digital communication systems, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Partial equalization for digital communication systems will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3066882