Pulse or digital communications – Synchronizers – Phase displacement – slip or jitter correction
Reexamination Certificate
2000-12-18
2004-08-17
Corrielus, Jean B. (Department: 2631)
Pulse or digital communications
Synchronizers
Phase displacement, slip or jitter correction
C370S516000, C375S355000
Reexamination Certificate
active
06778622
ABSTRACT:
TECHNICAL FIELD
This invention relates generally to the field of demodulating discrete multitone modulated signals, and in particular to estimating timing error in samples taken of a received discrete multitone modulated (DMT) signal.
BACKGROUND OF THE INVENTION
An almost universal goal of communications systems is to transfer information from one location to another. In many cases, the information takes the form of digital data, which is typically used to modulate a carrier to facilitate transmission.
For example, the Internet is made up of interconnected repositories of data stored all over the world. The repositories transmit the data to the computers of requesting users, where the data is displayed or used in some other way.
In an oil well drilling application, which is of particular interest to the assignee of this patent application, data regarding surrounding earth formations may be collected by wireline instruments or logging-while-drilling instruments. The instruments may analyze the collected data and transmit the collected data and the results of the analysis to the surface where further analysis is performed.
Discrete multi-tone modulation (“DMT”) is a popular technique for modulating a carrier with digital information. DMT employs a group of carriers, each of which carries a portion of the data to be transferred. Each carrier is digitally modulated with its portion of the data, using, for example, Quadrature Phase Shift Keying (QPSK) or Quadrature Amplitude Modulation (QAM), and is then transmitted over a communications medium, such as a cable or an optical medium.
QPSK represents a digital symbol, i.e. group of bits to be transmitted, by phase shifting a carrier a different amount for each symbol. For example, in 8-QPSK the modulator can apply eight different phase shifts to the carrier to represent eight different three-bit symbols. QAM represents a digital symbol by applying different phase shifts and amplitude adjustments for each symbol. For example, in 256-QAM the modulator can apply 256 different combinations of phase shifts and amplitude adjustments to represent 256 different eight bit symbols. If 256 carriers are modulated using 256-QAM, a total of 256 eight bit symbols can be transmitted at the same time.
One step in demodulating a DMT signal is to sample the DMT signal, which is made up of a number of modulated carriers, at a sampling rate sufficient to preserve the information in the signal. The sampled signal is typically digitized and transformed into the frequency domain where it can be processed to recover the data that was used to create the DMT signal. One step in the processing, known as “slicing,” attempts to correlate a sample with one of the possible phase modulations, in the case of QPSK, or amplitude and phase modulations, in the case of QAM.
Errors in sample timing may put all of the received data out of phase in the frequency domain. These phase errors may cause errors in the data because, as discussed above, the modulation techniques used in DMT convey information in the phase of the carriers.
A variety of techniques have been devised to correct the phase error. One common technique is to use a pilot tone to estimate the phase error. Such a pilot tone is subject to other types of system noise which may render erroneous an estimate of phase error based on an analysis of the pilot tone.
A portion of the communication bandwidth being used to transmit the DMT signal may be allocated to allow transmission of multiple pilot tones in order to average out the effect of the pilot tone noise. Such a technique will decrease the noise with the square root of the number of pilot tones being used.
Another known technique uses the data being transmitted to estimate the phase error. This technique uses the data as a block to estimate the phase error. For example, the technique may slice the data, compute slicing errors, and use the block of computed slicing errors to estimate the phase error. It may then use the estimated phase error to correct the phase of all carriers and slice the data again. This is a computationally intense algorithm which may be sensitive to slicing errors.
From the foregoing it will be apparent that there is still a need for a way to estimate the sample timing error that is computationally efficient, accurate and robust against slicing errors, without the need to allocate communication channel capacity to one or more pilot tones.
SUMMARY OF THE INVENTION
The deficiencies in the prior art are solved in the present invention which, in a preferred embodiment, provides a method and apparatus for estimating sample timing error. The method is computationally efficient, requiring only a few arithmetic operations. It is accurate, because it uses successively more accurate estimates of sample timing error. It is robust against slicing errors because it rejects slicing errors in its computations. It avoids the need to allocate communication channel capacity to one or more pilot tones because it performs its function without the use of a pilot tone or using only one pilot tone.
In one aspect of the invention, samples are taken of a signal which includes two or more measured tones, each of which has a frequency, a phase, and a magnitude. Each measured tone represents a modulated carrier in the signal. Each carrier has an associated constellation of ideal tones each of which represents an ideal modulation of its associated carrier. Before the method begins, an initial average phase error estimate is established. In one embodiment, this is accomplished by computing the average phase error from a pilot tone. In another embodiment, the average phase error is initialized to zero.
The method involves performing the following processing on each of a subset of the measured tones in the signal. The method begins by selecting an unprocessed measured tone. Next, the method adjusts the phase of the selected measured tone using an average phase error. Then, the method estimates a phase error for the selected measured tone, which includes slicing the selected measured tone. Finally, the method updates the average phase error with the estimated phase error for that tone.
In a preferred embodiment, estimating the phase error for the selected measured tone includes: (1) slicing the selected measured tone to associate it with an ideal tone; (2) computing a slicing residual for the measured tone; (3) computing an estimated phase error for the selected measured tone using the slicing residual; and (4) normalizing the estimated phase error.
In a preferred embodiment, slicing the selected measured tone includes selecting one of the ideal tones associated with the carrier associated with the selected measured tone that best approximates the selected measured tone.
In a preferred embodiment, computing the slicing residual includes subtracting a phasor representing the measured tone from a phasor representing the selected ideal tone.
In a preferred embodiment, computing the estimated phase error for the selected measured tone includes using the following equation:
P
k
≈
imag
(
⟨
E
k
,
V
k
⟩
)
w
k
where
P
k
=phase error for measured tone k;
E
k
=slicing residual for measured tone k (complex number);
V
k
=ideal tone for measured tone k (complex number);
<E
k
,V
k
>=inner product of E
k
and V
k
;
imag(A)=imaginary part of A; and
w
k
=weighting factor for measured tone k.
In various embodiments the weighting factor w
k
can be: (1) the squared magnitude of the ideal tone, (2) the variance of the magnitudes of all of the ideal tones in the constellation associated with the carrier associated with the measured tone, or (3) the standard deviation of the magnitude of all of the ideal tones in the constellation associated with the carrier associated with the measured tone multiplied by the magnitude of the ideal tone. Other weighting factors can also be used.
In a preferred embodiment, normalizing the estimated phase error includes applying the following equation:
NP
k
=
P
Corrielus Jean B.
Echols Brigitte L.
Jansson Pehr B.
Nava Robin
Schlumberger Technology Corporation
LandOfFree
Estimating timing error in samples of a discrete multitone... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Estimating timing error in samples of a discrete multitone..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Estimating timing error in samples of a discrete multitone... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3326958