Pulse or digital communications – Equalizers
Reexamination Certificate
1998-12-22
2002-09-24
Chin, Stephen (Department: 2734)
Pulse or digital communications
Equalizers
C375S230000, C375S231000, C375S260000, C375S347000
Reexamination Certificate
active
06456654
ABSTRACT:
This invention relates to a method of operating a receiver of a system which communicates multicarrier modulation (MCM) symbols via a communications channel, and to such a receiver. In particular, the invention is concerned with aspects of the receiver and its operation which relate to time domain equalization and frame alignment (also referred to as frame synchronization). Most commonly, MCM is implemented by discrete multitone (DMT) modulation, and this is assumed to be the case in the description below, but the invention can also be applied to other implementations of MCM.
BACKGROUND OF THE INVENTION
Communications systems using DMT modulation, referred to below as DMT systems, are becoming of increasing importance, especially for ADSL (Asymmetrical Digital Subscriber Line) communications on telephone subscriber lines. For example, American National Standards Institute (ANSI) draft standard T1.413-1998, entitled “T1.413 Issue
2
” and also known as “Network and Customer Installation Interfaces—Asymmetrical Digital Subscriber Line (ADSL) Metallic Interface”, referred to below for convenience simply as T1.413, provides detailed information on DMT systems.
In such a DMT system, ADSL data frames, for example at a frame rate of 4 kHz, are encoded and modulated each into a respective DMT symbol using an inverse discrete Fourier transform (IDFT), the DMT symbols being communicated in a superframe structure comprising 68 ADSL data frames or DMT symbols and a DMT synchronization symbol. As there is one DMT symbol for each ADSL data frame, the terms “frame” and “DMT symbol” are to some extent synonymous.
It is well known that inter-symbol interference (ISI) and inter-carrier interference (ICI) can seriously degrade performance of a DMT system. Accordingly, it is known to provide a time domain equalizer (TDEQ) at the receiving end of the communications path or channel in order to shorten the channel impulse response (CIR) to be less than or equal to the length of a cyclic prefix which is added to the communicated information. For example for a block of N=512 samples supplied from the IDFT to a subsequent digital-to-analog converter (DAC) for supplying a resulting analog signal to the communications channel, the last
32
of these samples can be additionally added in sequence before the start of the block as a cyclic prefix. If the CIR is reduced by the TDEQ to be no greater than the duration of the cyclic prefix, then ISI from the preceding symbol is avoided by discarding the cyclic prefix at the receiver, and the effect of the cyclic prefix on the subsequent samples of the block is easily accommodated. A frequency domain equalizer (FDEQ) is also provided at the receiving end, after the IDFT, in order to complete the equalization of the channel.
In an initialization process for subsequent communication of information using a DMT system, it is necessary among other things to determine coefficients for the taps of the TDEQ, referred to as training the TDEQ. It is known to do this using a minimum mean square error (MMSE) method, for example as is known from Al-Dhahir and J. Cioffi, “A Low-Complexity Pole-Zero MMSE Equalizer for ML Receivers”, Proceedings of the 32nd Annual Allerton Conference on Communication, Control, and Computing”, Sep. 28-30, 1994, pages 623-632. In this method, the TDEQ is trained in order to produce an overall impulse response, due to the channel and the TDEQ, that matches, in a MMSE sense, a model consisting of a cascade of a pure delay &Dgr; and an ideal target impulse response (TIR) of length p+1 samples, where p is the length in samples of the cyclic prefix and &Dgr; is the overall delay due to the channel and TDEQ. This method is dependent upon an appropriate CIR estimate and value of &Dgr;.
It is known to estimate the CIR during the initialization process by transmitting a probing signal with a period equal to the DMT symbol period; due to this periodicity neither a cyclic prefix nor the TDEQ at the receiver is required or used. A comb of tones with the same frequency spacing as the DMT tones is transmitted to enable the receiver to determine amplitudes and phases associated with all of the DMT tones, i.e. the channel frequency response for the DMT tones, and to derive the estimated CIR by performing an IDFT of the channel frequency response.
However, the channel frequency response and CIR as estimated in this manner are affected by the frame alignment between the transmitter and receiver during this channel estimation process. As the frame alignment affects the channel delay that is seen by the receiver, estimated CIRs are produced with different time shifts and, due to the periodic nature of the probing signal, as described below these can result in the estimated CIR being wrapped-around relative to the frame alignment. If a wrapped-around estimated CIR is used for TDEQ training, then the performance of the TDEQ may be significantly degraded.
As the TDEQ training is carried out in real time during the initialization process, it is also desirable to reduce the computational complexity of this training. To this end it may be desirable to reduce the number of equalizer taps used in the TDEQ, but in this case the degradation of the TDEQ performance discussed above is increased.
After training, the TDEQ is introduced into the communications channel where it presents an additional delay which affects the frame alignment. As the receiver must use the correct frame alignment in order to detect the MCM symbols communicated in normal operation of the system, the correct frame alignment must be recovered on introduction of the TDEQ. It is desirable t
6
avoid a lengthy search by the receiver for the correct frame alignment after training of the TDEQ.
An object of this invention, therefore, is to provide an improved method of operating a receiver of a communications system using MCM, such as a DMT system, and an improved receiver for such a system.
SUMMARY OF THE INVENTION
One aspect of this invention provides a method of operating a receiver of a system which communicates multicarrier modulation (MCM) symbols via a communications channel, comprising the steps of: in an initialization process: estimating a channel impulse response (CIR) of the channel using a predetermined periodic signal received via the channel; circularly advancing the estimated CIR relative to a frame alignment of the MCM symbols to eliminate wrap-around of the estimated CIR relative to the frame alignment; and determining parameters for time domain equalization, of MCM symbols received via the channel, using the advanced estimated CIR; and, in subsequent communications of MCM symbols via the channel: equalizing received MCM symbols in accordance with the determined time domain equalization parameters; and retarding frame alignment for the equalized received MCM symbols to compensate for the advance of the estimated CIR in the initialization process.
The circular advancement of the estimated CIR to eliminate of wrap-around relative to the frame alignment avoids determination of inappropriate parameters for the time domain equalization, and the retarding of the frame alignment for equalized received symbols avoids a need for searching for a new frame alignment.
Preferably the step of circularly advancing the estimated CIR relative to a frame alignment of the MCM symbols comprises circularly advancing the estimated CIR to remove leading substantially zero parts of the estimated CIR. This reduces the computational complexity required for the time domain equalization.
In a preferred embodiment of the invention, the step of circularly advancing the estimated CIR relative to a frame alignment of the MCM symbols comprises determining a first position relative to the frame alignment of a maximum energy part of the estimated CIR, determining a second position in advance of the first position at which an energy of the estimated CIR is below a threshold, and circularly advancing the estimated CIR by an amount D to move the second position to a frame alignment boundary. The step of retarding the f
Bourget Christian
Ginesi Alberto
Pakravan Mohammad
Chin Stephen
Liu Shuwang
Nortel Networks Limited
LandOfFree
Frame alignment and time domain equalization for... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Frame alignment and time domain equalization for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Frame alignment and time domain equalization for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2842399