Pulse or digital communications – Receivers – Interference or noise reduction
Reexamination Certificate
1999-02-12
2002-05-28
Deppe, Betsy L. (Department: 2734)
Pulse or digital communications
Receivers
Interference or noise reduction
C375S232000, C375S260000
Reexamination Certificate
active
06396886
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the use of a time-domain equalizer (TEQ) algorithm in a discrete multi-tone transceiver (DMT) and, more particularly, to a time-domain equalizer algorithm which operates as both a channel shortening filter and a noise whitening filter.
2. Related Art
The fast, efficient and error-free transmission of digital information from one point to another has become increasingly important. Many communications systems exist which permit digital information to be transmitted over various types of communication channels, such as wireless channels, fiber-optic channels, and wire line channels.
The present invention will be described in the context of a wire line communications channel, such as a telephone line which utilizes a twisted pair of copper wires. It is noted that the use of the present invention is not limited to wire line systems as those skilled in the art will appreciate from the discussion hereinbelow.
A modem is typically used to transmit and receive digital data over a telephone line. Modems employ a modulator to transmit the digital data over the telephone line and a demodulator to receive digital data from the telephone line. One common modulation technique is known as digital multi-tone modulation which requires a discrete multi-tone transmitter and a discrete multi-tone receiver at each modem in a communication system. Often, those skilled in the art refer to such modems as employing a DMT physical layer modulation technique.
Reference is now made to
FIG. 1
which is a block diagram of a conventional DMT communications system
1
. The system
1
includes a DMT transmitter
10
, a transmission channel
20
, and a DMT receiver
30
. The DMT transmitter
10
includes a symbol generator
12
, an inverse fast fourier transform (IFFT) modulator
14
, and a cyclic prefix generator
16
. The DMT transmitter
10
receives an input bit stream b(n) which is fed into the symbol generator
12
. The symbol generator
12
produces a signal X(k) which is fed into the IFFT modulator
14
. X(k) is a complex signal (i.e., a signal understood by those skilled in the art to comprise both a real and an imaginary component) formed by mapping pairs of bits of the input bit stream b(n) into a complex data space such that the complex signal X(k) has a length of N/2 samples. Symbol generator
12
also augments the signal X(k) with a complex conjugate to obtain a conjugate symmetric signal of N samples.
The IFFT modulator
14
performs an N-point inverse fast fourier transform on the conjugate complex signal X(k) to obtain the sampled real signal x(n). Since X(k) is a symmetric signal, the output of the IFFT modulator
16
is a real signal x(n). The real signal x(n) may be thought of as the summation of a plurality of cosine functions each having a finite length and a different frequency, phase, and amplitude, where these frequencies are multiples of a fundamental frequency. Since each of the cosine functions has a finite duration, x(n) is a varying amplitude discrete signal having a finite duration spanning N samples.
For the purpose of simplifying equations which will be discussed below, the transmission channel
20
is modeled as including a D/A converter
22
, transmit filter (not shown), a receive filter (not shown), and an A/D converter
26
on either end of a wire loop
24
. Those skilled in the art will appreciate that a practical system will employ the D/A converter
22
(and the transmit filter) in the DMT transmitter
10
and will employ the A/D converter
26
(and the receive filter) in the DMT receiver
30
.
Those skilled in the art will appreciate that the frequency spectrum of x(n) may be thought of as a plurality of orthogonal (SIN X)/(X) functions, each centered at a respective one of the frequencies of the cosine functions of x(n).
x(n) is transmitted over the channel
20
to the DMT receiver
30
. Since the transmission channel
20
has a non-ideal impulse response h(n), the received signal y(n) will not exactly match x(n). Instead, y(n) will be a function of the convolution of x(n) and h(n). Typically, h(n) will look substantially like the curve shown in FIG.
2
. The non-ideal characteristic of h(n) introduces an amount of interference (specifically intersymbol and interchannel interference) which should be compensated for in both the DMT transmitter
10
and the DMT receiver
30
.
A common technique in compensating for the non-ideal impulse response of the transmission channel
20
is to introduce a so-called guard band at the beginning of each finite duration signal x(n) to produce x′(n). The cyclic prefix generator
16
performs this function. The guard band is typically formed of the last V samples of x(n) for each DMT symbol. If the length of the impulse response h(n) of the transmission channel
20
is less than or equal to V+1, then the guard band of length V will be sufficient to eliminate the interference cause by the impulse response h(n). The guard band is commonly referred to in the art as a “cyclic prefix” (CP).
Unfortunately, the impulse response h(n) of a typical transmission channel
20
may be excessively long, requiring cyclic prefix lengths which substantially reduce the rate at which digital bits are transmitted across the transmission channel
20
. The DMT receiver
30
, therefore, employs signal processing techniques which effectively shorten the impulse response h(n) of the transmission channel
20
, thereby permitting a corresponding reduction in the length of the cyclic prefix required at the DMT transmitter
10
.
The DMT receiver
30
includes a time-domain equalizer (TEQ)
32
, a window circuit
34
, a fast fourier transform (FFT) demodulator
36
, and a bit generator
38
. The time-domain equalizer
32
is a finite impulse response (FIR) filter designed to compensate for the non-ideal impulse response h(n) of the transmission channel
20
. In particular, the time-domain equalizer
32
employs a finite number of coefficients (T) which are calculated to compensate for the non-ideal impulse response of the transmission channel
20
. As will be discussed in more detail below, the time domain equalizer
32
operates on the impulse response h(n) of the channel
20
such that the combined impulse response h
eff
(n) of the channel
20
and the time domain equalizer
32
has maximum energy within a limited band of samples. This may be thought of as “shortening” the effective impulse response of the channel
20
. The output of the time domain equalizer is z′(n).
A window circuit
34
is employed to remove the cyclic prefix from z′(n) to obtain z(n). The signal z(n) is input into the FFT demodulator
36
(which is understood to include a frequency domain equalizer function) to produce the complex symmetric signal X(k). After the complex conjugate portion of the signal X(k) is removed, the bit generator
38
maps the complex signal X(k) into an output bit stream b(n), which theoretically matches the input bit stream b(n).
Several algorithms exist for calculating the T coefficients of the time-domain equalizer
32
. One such algorithm is referred to as the least squares based pole zero cancellation algorithm (hereinafter “PROCESS #1”), which is discussed in detail in P. J. Melsa, R. Y. Younce and C. E. Rohrs, “Impulse Response Shortening for Discrete Multitone Transceivers,”
IEEE Trans. On Comm
. Vol. 44, No. 12, pp. 1662-71, December 1996, the entire disclosure of which is hereby incorporated by reference. Another such algorithm is referred to as the optimal shortening algorithm (hereinafter “PROCESS #2”), which is also discussed in detail in the above referenced IEEE publication. Still another algorithm is referred to as the eigenvector approach using the power method (hereinafter “PROCESS #3”), which is discussed in detail in M. Nafie and A. Gatherer, “Time-Domain Equalizer Training for ADSL,”
Proc. ICC
, pp. 1085-1089 (1997), the entire disclosure of which is hereby incorporated by reference.
Although the above techniques for calculati
Deppe Betsy L.
NEC USA Inc.
Ostrolenk Faber Gerb & Soffen, LLP
LandOfFree
DMT time-domain equalizer algorithm does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with DMT time-domain equalizer algorithm, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and DMT time-domain equalizer algorithm will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2844727