Pulse or digital communications – Bandwidth reduction or expansion
Reexamination Certificate
1999-07-26
2004-11-30
Fan, Chieh M. (Department: 2634)
Pulse or digital communications
Bandwidth reduction or expansion
C375S295000
Reexamination Certificate
active
06826227
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a data transmission method using loss-free compression techniques, in order to optimize the use of available transmission channels.
2. Description of the Prior Art
Various seismic data transmission systems are used to connect local acquisition units to a central station, either directly or via intermediate stations provided with complex local unit concentration or control functions. Links can be provided by cables, radio links, one or more relays, or by combining cable links and radio links as described, for example, in French Patents 2,720,518, 2,696,839, 2,608,780, 2,599,533, 2,538,561, 2,511,772 or 2,627,652 filed by the assignee.
French Patent 2,608,780, filed by the assignee, notably discloses use of seismic acquisition boxes provided with two transmission channels, one with a relatively high transmission rate, the other with a passband that can be relatively narrow according to the local availability of transmission frequencies, more readily available within the scope of the radio communication emission regulations in force. The seismic data collected during successive cycles are stored in a mass storage in each box and intermittently transferred to a central control and recording station. In order to enable the operator in the central station to check that data acquisition by each acquisition box is normal, partial data transmission is performed, which is suitable with a transmission channel with a relatively narrow passband.
French Patents 2,692,384 and 2,757,641 and French Patent Application 97/09,547, all filed by the assignee, describe the use of seismic acquisition boxes provided with a specialized processor for signal processing, that perform many geophone and acquisition system element controls, seismic trace preprocessing previously performed in the central station after transmission, and apply compression algorithms to the seismic data to be transmitted. All these preprocessing operations considerably decrease the volume of data to be transmitted to the central station.
The current trend, notably within the scope of 3D seismic exploration methods, is to spread out over a zone to be explored, onshore, offshore or in coastal areas, often over several kilometers, hundreds or even thousands of seismic receivers. The volume of data to be collected and transmitted continually increases. In order to prevent transmission problems from acting as a break on the evolution of seismic systems, the trend is to use data compression processes selected to be compatible with geophysicists' specific demands.
Compression of seismic data can provide both an appreciable saving in space for the mass storage modules in the local acquisition boxes and/or the local control and concentration stations, and a considerable saving in transmission time.
The known data compression methods can be classified into two families, a) compression methods without information loss and b) methods leading to an information loss so the restored data loses a large part of its precision.
In geophysics, it is essential that the losses due to compression remain as low as possible because the most pertinent information often has a very low amplitude and can be isolated from the background noise only by digital processing on several traces. Possible precision losses are tolerable only in certain particular cases. For example, the information transmitted is only used for controlling the working order of the equipment and for displaying the “pattern” of the sampled traces.
Examples of methods of the first family are the methods which eliminate the redundance of the data or the methods referred to as dictionary methods where each word is replaced by its index in a reference table, which are all the more interesting since the files to be compressed contain many redundancies, the methods referred to as RLE (Run-Length Encoding) methods, statistical type coding methods where data is replaced by a code having the same significance but occupying less room and arithmetic coding type methods which represent a variable number of data by a constant number of bits.
A known compression technique referred to as LPC (Linear Predictive Coding) method is well suited for compression of acoustic or seismic waves. It essentially replaces a signal sample s(t) by a prediction made from p previous samples, assuming that the signal is stationary. Instead of transmitting a sample s(t), its prediction Ŝ(t) is transmitted, i.e. the prediction coefficients and the residues e(t), i.e. the difference between the real value and the prediction made at the time t, which allows finding the value s(t)=Ŝ(t)+e(t) after decompression. If the prediction is good, the residues are few and they occupy less room than the initial values s(t). The coefficients used for calculating Ŝ(t) are generally small in number and occupy little room in relation to s(t).
However, with this type of coding, it is conventional to store the successive trace samples in buffer memories of definite size and to calculate prediction coefficients that are used for all the samples of a single group. This type of convention is ill-suited to signals of wide dynamic range such as the signals emitted by impulsive sources (dynamite charges for example) and coding errors appear because the predictive coefficients selected are not really representative.
SUMMARY OF THE INVENTION
The method according to the invention performs loss-free compression of signals of wide dynamic range, such as seismic signals, for transmission.
The method according to the invention finds applications notably in the field of seismic prospecting where transfer of considerable data to a central station, such as a recording truck, is necessary. Signals are picked up by a large number of receivers such as geophones coupled with a geologic formation to be studied, in response to impact pulses emitted by a seismic source and reflected by the subsoil discontinuities. The signals picked up are collected by local acquisition units, sometimes spread out over a distance of kilometers, which collect the signals received by one or more receivers, to digitize the signals, to apply complex preprocessings thereto and to store the signals in a local memory before real-time or off-line transmission thereof to a collection station through a transmission channel such as a cable, an optical fiber, a radio channel, etc.
The method comprises sampling and digitizing the signals, and using a predictive coding technique.
The method comprises determining an amplitude range depending on at least one threshold value (S), that contains most of the prediction residues, coding the prediction residues by binary words having a number of bits sufficient for coding of the range and coding the prediction residues whose amplitude is outside the range, by binary words having a fixed number of bits greater than n, and transmitting only the prediction residues (the prediction coefficients are not transmitted but recalculated).
According to an embodiment, the method comprises determining an estimation factor (Ŝ) of an optimum threshold value corresponding to a mean value obtained from the position of the respective most significant bits of N prediction residues determined successively.
According to another embodiment, the method comprises determining amplitude ranges depending on distinct threshold values.
The method can also comprise separate coding of least significant and most significant bits of the signal samples.
According to a preferred embodiment, the method comprises determining prediction residues for each new signal sample.
The method according to the invention is advantageous with its two stages taken separately or preferably in combination
a) determining the number of bits at most necessary for coding most of the errors allows obtaining a very high compression rate, even if the particular processing applied for coding the remaining atypical errors (few in number when the coding range is well selected) is taken in
Duval Laurent
Tran Van Bui
Fan Chieh M.
Institut Francais du Pe'trole
LandOfFree
Method intended for loss-free compression of signals of wide... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method intended for loss-free compression of signals of wide..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method intended for loss-free compression of signals of wide... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3314612