Pulse or digital communications – Transmitters – Antinoise or distortion
Reexamination Certificate
1999-09-09
2001-03-06
Chin, Stephen (Department: 2734)
Pulse or digital communications
Transmitters
Antinoise or distortion
C375S377000, C332S107000
Reexamination Certificate
active
06198778
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the field of multicarrier modulation transmissions (DMT: Discrete Multitone Modulator) and, more specifically, to the transmission of signals coded by orthogonal multiplexing in the frequency range (COFDM: Coded Orthogonal Frequency Division Multiplexing).
2. Discussion of the Related Art
The transmission, in multicarrier modulation, of signals coded by orthogonal frequency division multiplexing is a relatively recent technique which is advantageous in many applications with respect to the other conventionally used techniques.
A first application relates to wire communication telephone systems. By using a DMT/COFDM transmission over an asymmetric digital subscriber line (ADSL), very high rate channels are available, in which both voice and digital signals, for example, compressed video signals, can travel. As an example, the rate of such a transmission can be 10 Mbits/s in one direction and 640 kbits/s in the other direction. As a comparison, a modem operating according to standard V34bis which is one of the fastest modems currently used provides a rate of 33.6 kbits/s in both directions.
A second application relates to the audio-digital diffusion from a satellite or from a terrestrial transmitter to a mobile system, for example, a vehicle. The DMT/COFDM transmission here enables, for example, a transmission of information at a rate of about 1.7 Mbits/s.
A third application relates to the terrestrial diffusion of digital television signals (DTTB) for which {fraction (1/9)} of the capacity of the available frequencies is currently lost due to the necessary overlapping areas between the different transmitters. The implementation of a DMT/COFDM transmission allows a given channel to be transmitted on the same frequency whatever the transmitter.
A DMT signal is formed by superposition of
n
carriers modulated independently from one another. The carriers are modulated, for example, by a quadrature amplitude modulation (QAM) or a frequency shift keying modulation (FSK).
FIG. 1
very schematically shows a conventional example of circuit performing multicarrier modulation.
A stream of data arrives in series on a series-to-parallel (S/P) converter
1
, the outputs of which are sent onto a circuit
2
for generating the DMT symbols by means of an inverse fast Fourier transform (IFFT). From a functional point of view, circuit
2
is formed of a QAM or FSK modulator using several carriers f
1
, f
2
, . . . fn, which provides the modulated carriers to an adder (&Sgr;) 4 superposing the successive samples of these carriers to generate the DMT symbols. Generally, each carrier is associated with a data packet, that is, the data stream is sent in parallel to modulator (MOD)
3
by grouping the data bits into packets of the same size. For example, each carrier is associated with a 3-bit group transmitted in 8-QAM modulation to reproduce the eight possible states of the combination of the 3 data bits. In this example, if 256 carriers are used (n=256), a DMT symbol includes 768 bits, and the transmission rate is 758/T bits/s, where T represents the duration of a DMT symbol.
FIG. 2
illustrates an example of DMT symbol corresponding to the superposition of a sample of all the modulated carriers.
A problem which arises in multicarrier modulation is that adding the samples of modulated carriers can result, randomly, in peaks p of very high amplitude when several modulated carriers superpose in phase. These peaks have significant consequences upon the analog portion of the circuit, especially upon the complexity and the feasibility of digital-to-analog (on the transmit side) and analog-to-digital (on the receive side) converters, the peaks being conventionally likely to reach some twenty volts from peak to peak.
To solve this problem, the digital DMT signal is generally clipped so that its maximum amplitude never exceeds a predetermined value Aclip. Value Aclip is generally chosen according to a threshold of probability of occurrence of a peak for a given application. Indeed, the application of an inverse fast Fourier transform leads to an amplitude distribution having the shape of a Gauss curve schematized on the right-hand side of FIG.
2
. Value Aclip is thus generally chosen according to the analog transmission circuit and to the digital-to-analog converter used to respect a certain error rate, that is, a certain probability of clipping of the symbols, in order to minimize signal losses. For example, for a DMT transmission applied to an asymmetrical digital subscriber line, the standards establish a probability which is lower than 10
−7
.
The clipping of the digital DMT signal introduces a clipping noise which is detrimental to the performance (signal-to-noise ratio) of the communication system. Especially, the signal-to-noise ratio conditions the possible rate since it conditions the space to be provided, in the phase diagram of each carrier, between two reception points. The better the signal-to-noise ratio, the higher the number of bit combinations contained in a sample of a carrier can be by increasing the length of the bit packets associated with each carrier.
A first conventional solution to improve the signal-to-noise ratio is illustrated in FIG.
3
. This solution consists of providing, at the output of IFFT circuit
2
, a detector (DETECT)
5
that detects the presence of a clipped DMT and controls a circuit (CODE)
6
of coding back of the concerned data, interposed between series-to-parallel converter
1
and circuit
2
. The function of circuit
6
is to modify the coding of the data, according to a law known by the receiver of the sent symbols, to transmit back a DMT symbol, clipped during a first run, as an unclipped symbol. Indeed, a coding modification causes a modification of the phases of the modulated carriers and the probability that a DMT symbol exhibits peaks for two different codings is extremely low.
A disadvantage of such a solution is that it requires storage, upstream of series-to-parallel converter
1
, of the data to be transmitted to enable their retransmission if a clipped DMT symbol is detected after coding. Another disadvantage of this solution is that the circuit must transmit, in addition to at least one second DMT symbol if the first one has been clipped, a code for indicating to the receiver the number of runs of a same DMT symbol sent. Further, this solution requires a faster IFFT circuit so that the second possible transformation occurs before the modulation of the following symbol and/or an additional memory so as to keep the symbols upstream of the circuit.
This first solution is described in a paper entitled “A method to reduce the probability of clipping in DMT-based transceivers” by D. Mestdagh and P. Spruyt, in IEEE Transactions on Communications, October 1996, volume 14, number 10, pages 1234-1238.
A second solution consists of modifying the coding, at the input of circuit
2
, for the bit combinations which are likely to produce the highest peaks. Indeed, it has been acknowledged that for all combinations, there is a peak amplitude level for the case where the carriers become in phase. Such a solution is described, for example, in a paper entitled “Block coding scheme for reduction of peak to mean envelope power ratio of multicarrier transmission schemes” by A. E. Johns, T. A. Wilkinson and S. K. Barton, in Electronics Letters, December 1994, volume 30, number 25, pages 2098 and 2099, and in a paper entitled “Simple coding scheme to reduce peak factor in QPSK multicarrier modulation”, by S. J. Shepherd et al., in Electronics Letters, July 1995, volume 31, no14, pages 1131 and 1132.
If this second solution does not require a double transmission of the same DMT symbol, it still requires transmission of additional bits associated with the coding. Further, this solution requires a high data processing speed to perform the additional coding, and thus a decrease of the energy per bit for the same general transmission power, which results in a degradation o
Chin Stephen
Galanthay Theodore E.
Morris James H.
Park Albert
SGS-Thomson Microelectronics S.A.
LandOfFree
Method and device for setup of clipping noise of a... 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 and device for setup of clipping noise of a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for setup of clipping noise of a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2448442