Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
2000-09-20
2004-12-07
Chin, Stephen (Department: 2634)
Pulse or digital communications
Equalizers
Automatic
C375S231000, C375S233000, C375S229000
Reexamination Certificate
active
06829296
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
This invention generally relates to filters, and more specifically to a spectrally flat time-domain equalizer (TEQ) filter, which increases the data rate of a communications system.
2. Background Art and Technical Problems
In modern electronic circuits, many different kinds of filters are used in a variety of applications. Filters are typically used to remove one or more components of a signal so that a “clean” signal is obtained. For example, filters are used in a number of digital communication applications, such as equalization, echo cancellation, band selection, and the like. For example, a Digital Subscriber Line (DSL) system, such as an Asymmetric Digital Subscriber Line (ADSL) system, uses filters to process and provide specific attributes to the signals, which are transmitted over communications channels. More specifically, a Discrete Multitone (DMT) modulation transmission method can be used in ADSL. The DMT transmission divides the channel into several independent sub-channels making it easy to transmit data on each sub-channel. It is known in the art that a channel refers both to the physical channel and the mathematical representation of the channel (e.g., the channel impulse response). The overall data rate of the channel is the sum of the data rates over all these sub-channels. In this way, instead of transmitting data over one wideband channel, data is transmitted over the narrower sub-channels.
The transmission of successive DMT symbols (or packet of data) may allow inter-symbol interference (ISI) to appear due to the dispersive nature of the channel. One way to reduce or ideally eliminate ISI is to equalize the overall channel, for example, by appending a filter at the receiver end and making the overall channel impulse response at the receiver end very close or equal to a Dirac impulse. However, such a method is not efficient for the type of channels encountered in ADSL applications. For example, ADSL channels are dispersive; yielding a long impulse response, which implies an ISI corrupted signal together with a complex equalization structure. As ISI becomes more severe, the equalizer complexity rises rapidly and the computational cost increases to an unacceptable level. This implies that to keep a reasonable performance, a more complex, expensive, and powerful receiver must be adopted.
Another way to reduce or eliminate ISI is to insert between any two adjacent information symbols (such as DMT symbols) a time interval during which some non-information carrying data is transmitted. For example, a predefined data sequence may be transmitted. In any case, if this time interval (also known as guard period) is at least as long as the channel memory, the effects of ISI can be substantially isolated from one information symbol to the next information symbol, and processing may be performed in an information symbol by information symbol basis. If the transmitted information symbol is repeated, then the guard period is commonly referred to as the cyclic prefix because the repetition is essentially a cyclic extension of the information symbol. However, the channels used in ADSL applications, for example, have a long impulse response, which makes the guard period a waste of the available bandwidth.
A more practical solution combines the above two methods by both equalizing the channel and appending each information symbol with a guard period. This equalization method attempts to reduce the channel at the receiver end (i.e., the effective channel) to CP+1 taps, where CP is the cyclic prefix and taps are the coefficients of a filter. Accordingly, the channel may be coupled with a filter (e.g., a TEQ filter) to effectively shorten the channel to CP+1 significant taps, where significant taps are coefficient values that exhibit a significant or much higher value as compared to the value of the remaining coefficient values. By applying this method, ISI may theoretically be eliminated while maintaining a small overhead per transmitted information symbol (equal to the value of the CP).
Another concern in communications systems is noise. For instance, noise may be added to the analog signal while travelling over the channel (e.g., white Gaussian noise) by the process of digitizing the analog signal (e.g., quantization noise), and by the digital processing applied to the digitized signal. Those of skill in the art recognize that different kinds of noise will affect the signal quality in different ways. Furthermore, residual ISI usually has a flat spectrum and may contaminate the performance of the TEQ filter, which may decrease the ratio of the signal to noise power (SNR). A decrease in the SNR also reduces the data rate of the system, which is highly undesirable. As such, designing a TEQ filter that merely shortens the effective channel to CP+1 significant taps may also degrade the data rate, if the frequency response of the TEQ filter results in a significant attenuation of signal frequencies not originally attenuated. If the TEQ filter has a relatively flat frequency response (e.g., few ripples), then most useful or “good” frequencies will not be further attenuated. Information or data lost before the TEQ filter processing will remain lost, but substantially no additional losses will occur due to the TEQ filter processing.
Communications systems are often characterized by a transmitter side and a receiver side that communicate via a channel.
FIG. 1
illustrates a communications system
101
having a transmitter
103
, a channel
105
, and a receiver
107
. Transmitter
103
transmits data in the form of one or more information symbols (e.g., DMT symbols) across channel
105
to receiver
107
. Such information symbols have a cyclic prefix appended at or attached to the beginning of each information symbol transmitted. As such, each information symbol is preceded by its cyclic prefix and transmitted over channel
105
. If channel
105
is longer than the cyclic prefix, then ISI may result at receiver
107
of communications system
101
. In order to avoid such ISI, channel
105
may communicate with a time domain equalizer (TEQ) filter (not shown) to shorten the channel to a desired length.
FIG. 2
illustrates an exemplary communications system
201
having a TEQ filter
217
. Communications system
201
includes a transmitter side and a receiver side, where a channel
211
provides a medium of communication between the two. On the transmitter side, communications system
201
includes a signal processor
203
, an inverse fast Fourier transform (IFFT) engine
205
, a parallel to serial converter
207
, and a transmitter filter
209
. Channel
211
may be any medium commonly used in communications systems. For example, in ADSL transmission, channel
211
may be a twisted pair of wires. On the receiver side, communications system
201
includes a receiver filter
213
, a fast Fourier transform (FFT) engine
219
, and a frequency domain equalizer (FEQ)
221
.
Communications system
201
receives data
223
and outputs filtered data
225
. Signal processor
203
processes data
223
and communicates the results to IFFT engine
205
, and IFFT engine
205
produces time-domain data. A cyclic prefix (CP) is appended to each time-domain information symbol by an add CP means
206
, and the modified information symbol communicated to parallel to serial converter
207
. Parallel to serial converter
207
then converts the modified information symbol into a serial signal. The serial signal is communicated to transmitter filter
209
(e.g., a digital to analog conversion means), which transforms the serial signal into a continuous time signal.
Channel
211
provides a medium for transmitting the continuous time signal to receiver filter
213
, which may perform some analog and digital filtering to the incoming signal. Receiver filter
213
digitizes the continuous time signal into a digital signal. TEQ filter
217
filters the digital signal in order to reduce and ideally cancel the ISI. The cyclic prefix is removed from each
Petsalis Evangelos
Troulis Markos G.
Zhang Xuming
Chin Stephen
Mindspeed Technologies Inc.
Vartanian Harry
LandOfFree
Spectrally flat time domain equalizer and methods does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Spectrally flat time domain equalizer and methods, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Spectrally flat time domain equalizer and methods will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3317335