United States Patent
`
`[191
`
`l|||l|||||l||||l|lllll||l|l||||||l||||||||||||l||||||||||||||||llllllllllll
`USO05197061A
`nu Patent Number:
`
`5,197,061
`
` .
`
`Halbert-Lassalle et al.
`
`[45] Date of Patent: Mar. 23, 1993
`
`[54] DEVICE FOR THE TRANSMISSION OF
`DIGITAL DATA WITH AT LEAST TWO
`LEVELS or PROTECTION AND
`CORRESPONDING RECEPTION DEVICE
`
`[75]
`
`Inventors: Roselyne Halbert-Lassalle;
`Jean-Francois Helard; Bernard Le
`Floch, all of Rennes, France
`
`[73] Assignees: Eat Fnnuis; Teledmusion de
`Fnnce’ France
`
`[211 APPL N0-1 6719483
`
`[22] Filed:
`
`Mar. 19, 1991
`
`[30]
`Mar. 23,
`
`-
`-
`-
`-
`-
`Fouls“ Application Pnonty Data
`France ................................ 90
`
`Int. Cl.5 ................................................ 1104.] 9/00 .
`[51]
`[52] U.s. Cl. ......................................... 370/11; 370/19
`[58] Field of Search ..................... .. 370/11, 19, 21, 50,
`370/691, 70; 375/38; 371/41
`
`[56]
`
`References Cited
`U5‘ PATENT DOCUMENTS
`4,881,241 ll/1989 Pommier et al.
`................... .. 375/38
`Primary Examiner-Curtis Kuntz
`Assistant Examiner'—T. Ghebretinsae
`Attorney, Agent, or Firm—Locke Reynolds
`[57]
`ABSTRACT
`A device for the transmission of digital data with at least
`two levels of protection, of the type providing for the
`distribution of the data to be transmitted in the form of
`digital elements in the time-frequency space and the
`transmission of symbols each formed by a multiplex of
`N orthogonal carriers modulated by a set of the digital
`elements, and transmitted simultaneously,
`the device
`including channel encoding means comprising at least
`two types of modulation and/or at least two encoding
`efficiency levels. This enables to optimize the use of the
`transmission channel
`differentiated trans-
`mission techniques to portions of data of a same digital
`train as a function of the different levels of protection
`sought, against transmission errors.
`
`14 Claims, 6 Drawing Sheets
`
`52
`
`53
`
`54
`
`bl
`
`co/vvau/rs
`ENCODER
`n/-,0,
`
`rmE- FREO.
`/;vr5m_5,4y5
`
`SIGNAL
`BINARY
`swoon:
`
`
`
`
`
`
`
`
` DIS7PIfll770M
`
`UODULA TION
`
`COFDAI
`
`0/7
`
`
`
`
`
`TIME-FWEO.
`INTERLEAE
`
`SET
`A RTITIONING
`
`LATTICE
`ENCO ER
`R2’
`
`
`
`Aruba Networks et al. Exhibit 1007 Page 00001
`
`Aruba Networks et al. Exhibit 1007 Page 00001
`
`

`
`U.S. Patent
`
`Mar. 23, 1993
`
`Sheet 1 of 6
`
`5,197,061
`
`-—-—RArLé'/GH
`
`§
`
`”
`GZIUSS/AN
`
`’ ’
`
`SPECTRAL
`EFFICIENCY
`
`/?=2/3
`___.. =
`-
`”-i’41—v—i§—/-9- 8'P.S‘K
`4'P$K
`(Lattice)
`
`F/0.1
`
`3 G/‘I/s/Hz)
`/?= 3/4
`/6-0AM
`(Larf/'ce/
`
`
`
`Page 00002
`
`Page 00002
`
`

`
`aU
`
`115,
`
`mV,um‘NWnewamSNG
`
`3R..J®%Z__./ESkmmEm_fix_.:w.5N_.6M.0_
`
`_5%.____
`
`6.MsseéW._r__fscwm_«."SNecw_am
`\<_.Q__as‘W_3»__
`
`1m60
`0.,M...WW.©\u\M
`
`
`
`Page 00003
`
`
`

`
`U.S. Patent
`
`Mar. 23, 1993
`
`Sheet 3 of 5
`
`5,197,061
`
`hm.
`
`h.m.
`
`‘M%Mahm.Nxh.NOM.
`
`0mmw
`
`VM.QM.VONFm.
`
`
`km,qmzzwtudot.maouzm
`kkkE.t.:<\wkkIManSn_m
`.......two?»nht
`
`
`
`maoumqFm,Mm..3:mqouzm.
`
`MQQU-
`.3.3!mnouzm
`
`
`dot
`
` uncut.gEmm.an.Gm.rm.
`
`N4.~UN.u\
`
`QM.nonHW
`
`G9:
`
`WQQUEW
`
`Page 00004
`
`Page 00004
`
`

`
`S”U
`
`tHetaP.
`
`Mar. 23, 1993
`
`Sheet 4 of 5
`
`5,197,061
`
`_V%
`
`.::<..uR.
`
`..Q1&3
`
`MQ90§.N
`
`Ghkkno.3
`
`.§.6.E.w.=S
`
`
`
`M3.~Q3§QU
`
`kuflQO.u§.W
`
`N\I\t
`
`Nb.
`
`scfikbb
`
`tot1.3%!
`
`>«O\k\§hxkh<Q
`
`.§3%dthk3.Q.WkKc.Ws¢sk
`
`LU
`
`M.
`
`%§~§Dskit‘Q
`
`tmuouzm
`&INQ
`
`nuts‘4
`
`Page 00005
`
`Page 00005
`
`
`
`
`
`
`
`
`

`
`4...HetaPQMU
`
`Mar. 23, 1993
`
`Sheet 5 of 5
`
`5,197,061
`
`QNQOU8
`
`W3VUq.t.wbsnmuZ3
`
`mm
`
`vm
`
`.8
`
`>6\kumsotm
`
`QQKQU
`
`>\0sLVdgghfi
`
`xb0%
`
`UsRm4lEUk>§8
`
`knitYb
`
`.<ot§E«.m
`
`Nb
`
`Page 00006
`
`Page 00006
`
`
`
`
`
`

`
`f.He4...aPQMU
`
`Mar. 23, 1993
`
`Sheet 6 of 6
`
`5,197,061
`
`RNbwnk
`
`~->BtV.33!
`
`~o.<NQQ.u>\W
`
`
`
`.~.$<>\3&0
`
`
`
`Muzak:4863
`
`
`
`N..QMQOUEM
`
`qmzzscu
`
`NE
`
`Wb§85had.V
`
`‘L‘Q
`
` N.©\r.\
`
`kk
`
`tosumqmm
`
`M.
`
`K
`
`mgSmhm
`
`MK
`
`
`
`
`
`22..umqmm..§s<3<u
`
`wkmu
`
`zwmtimfi
`
`aP
`
`8
`
`70000e
`
`Page 00007
`
`
`
`
`

`
`1
`
`5,197,061
`
`2
`tion against transmission errors, irrespectively of the
`importance of the data elements.
`It often happens that there are major differences
`among the pieces of digital information designed to be
`transmitted in the same channel. Thus, for example, in
`the case of sound signals, it is known that it is possible to
`tolerate an error rate of about 1% for the least signifi-
`cant bits (LSBs) while the most significant bits (MSBs)
`often require an error rate of less than 10-5. In the same
`way, in an image signal, all the transmitted coefficients
`do not have the same importance, especially from a
`psychovisual point of view.
`It is clear that the error rate is related notably to the
`type of encoding used, all conditions of reception being
`moreover equal, and in particular to the error correc-
`tion methods and to the redundancies introduced. It can
`be seen therefore that the encoding efficiency, in terms
`of bit rate,
`is related to the encoding used. In other
`words, the more reliable the encoding, the lower is its
`bit rate.
`
`DEVICE FOR THE TRANSMISSION OF DIGITAL
`DATA WITH AT LEAST TWO LEVELS OF
`PROTECTION AND CORRESPONDING
`RECEPTION DEVICE
`
`BACKGROUND OF THE INVENTION
`
`1. Field of the Invention
`The field of the invention is that of the transmission of
`digital data, notably in disturbed channels. More specifi-
`cally, the invention relates to the transmission, in one
`and the same channel, of data requiring different levels
`of protection against transmission errors.
`The transmitted data may be, for example, sound data
`or audiovisual data (notably in television, visiophony
`etc.) and, more generally, any type of digital data on
`which it may be worthwhile, useful or at any rate not
`harmful to carry out a discrimination between the digi-
`tal elements using a criterion of the minimum protection
`level desired.
`1. Description of the Prior Art
`The technological background of the invention is the
`digital sound broadcasting system as described in the
`US. Pat. No. 4,881,241 dated 14th November 1990. The
`digital broadcasting system presented in these prior art
`patent applications is based on the joint use of a channel
`encoding device and a coding orthogonal frequency
`division multiplex (COFDM) system.
`The modulation method proper of this known system
`consists in providing for the distribution of constituent
`digital elements of the data signal in the frequency-time
`space f-t and in the simultaneous transmission of the sets
`of digital elements on N parallel broadcasting channels
`by means of a multiplex of orthogonal carrier frequen-
`cies. This type of modulation makes it possible to pre-
`vent two successive elements of the data train from
`being transmitted at the same frequency. This enables
`the absorption of the frequency fluctuating selectivity
`of the channel through the frequency dispersal, during
`the broadcasting, of the initially adjacent digital ele-
`ments.
`
`10
`
`20
`
`25
`
`30
`
`35
`
`45
`
`55
`
`65
`
`From the viewpoint of channel encoding alone, it is
`therefore clear that a channel encoding system that
`uniformly protects the flow of data and is based on the
`sensitivity to transmission errors of the most significant
`bits is sub-optimal in terms of spectral efficiency (the
`number of bits/s/Hz).
`The result thereof is high quality encoding for all the
`bits, and therefore an over-coding of the bits with low
`significance, leading to a loss in the transmission bit rate.
`There already exist known methods to match the
`channel encoding with the requirements of the source
`encoding. It has notably been proposed to use rate com-
`patible punctured convolutional (RCPC) codes which
`are associated, at reception, with a single Viterbi de-
`coder working in soft decision mode. This method,
`described by R. V. Cox, N. Seshadri and C-E. W. Sund-
`berg in “Combined Subband Source Coding And Con-
`volutional Channel Coding”, ITG Tagung: Digital
`Sprachverarbeitung, 26, Oct. 28, 1988, Bad Nauheim,
`achieves the periodic suppression, or puncturing, of
`certain bits of the source code when the maximum error
`rate permits it. However, this type of encoding remains
`related to a particular modulation,
`thus limiting the
`The known encoding process is aimed, for its part, at
`spectral efliciency that can be obtained. Thus, in the
`the processing of the samples coming from the demodu-
`case of an RCPC encoding used with the 4-PSK modu-
`lator to absorb the effect of amplitude variation of the
`lation, it is possible at most to achieve a spectral effi-
`received signal, due to the Rayleigh process. The en-
`ciency that is strictly below 2. Besides, it does not seem
`coding is advantageously a convolutive encoding, pos-
`to be possible to use this technique efficiently with mod-
`sibly concatenated with a Reed-Solomon type of encod-
`ulations where there are more than four phase states.
`mg.
`50
`The invention is aimed at overcoming these draw-
`' backs.
`In a known way, the encoded~digital elements are».
`furthermore interleaved, in time as well as in frequency,
`so as to maximize the statistical independence of the
`samples with respect to the Rayleigh process and to the
`selective character of the channel.
`This method is well adapted to the broadcasting of
`digital signals at a high bit rate (several megabits/s) in
`channels that are particularly hostile. This has been
`demonstrated by the fust embodiment of this method in
`digital sound radio broadcasting. In particular, it ena-
`bles the reception of digital data by mobile receivers
`moving about in an urban environment, i.e. in the pres-
`ence of parasitic noise and jamming, and under condi-
`tions of multiple propagation (Rayleigh process) gener-
`ating a phenomenon of fading.
`However, in its present form, this method is not used
`in an optimal way. The same charmel encoding is used
`for all the data to be transmitted, with the same protec-
`
`SUMMARY OF THE INVENTION
`
`More specifically, the invention is aimed at providing
`a COFDM type digital transmission device optimizing
`the efficiency of the transmission.
`Another aim of the invention is to provide a device
`such as this that enables the optimization of the use of
`the transmission channel through the assigning of differ-
`entiated transmission techniques to portions of data of
`one and the same digital train as a function of the differ-
`ent protection levels sought, against transmission errors.
`An additional aim of the invention is to provide a
`device such as this making use of the flexibility and
`independence between the carriers of the COFDM
`method.
`These aims, as well others that shall appear hereinaf-
`ter, are achieved by means of a device for the transmis-
`
`Page 00008
`
`Page 00008
`
`

`
`3
`sion of digital data with at least two levels of protection,
`of the type providing for the distribution of the data to
`be transmitted in the form of digital elements in the
`time-frequency space and the transmission of symbols
`each formed by a multiplex of N orthogonal carriers
`modulated by a set of said digital elements, and trans-
`mitted simultaneously, wherein said device includes
`charmel encoding means comprising at least two types
`of modulation and/or at least two encoding efficiency
`levels.
`Thus, it is possible to assign, to each type of data to be
`transmitted, as a function of the required level of pro-
`tection against errors, an adequate modulation and/or
`encoding efficiency.
`Advaritageously, said multiplex of N carriers is di-
`vided into at least two sets of carriers, with a different
`type of modulation and/or an encoding with different
`encoding efficiency being assigned to each of said sets.
`In this case, said sets of carriers are preferably inter-
`leaved along the frequency axis, in such a way that said
`sets of carriers benefit from the frequency independence
`related to the total bandwidth. Indeed, it is worthwhile
`to distribute the carriers over the greatest possible band-
`width, so as to provide for maximum robustness with
`respect to selective disturbances in frequency (notably
`fading phenomena).
`In another embodiment, the transmission device of
`the invention includes, for at least one of said carriers,
`means for selection between at least two types of modu-
`lation and/or between at least two encoding efficiency
`levels as a function of the transmission bit rate and of the
`disturbances affecting the channel.
`This provides for the optimal matching of the bit rate
`with the data to be transmitted.
`the
`Advantageously,
`in this second embodiment,
`transmission device includes means for the generation
`of assistance data making it possible to have knowledge,
`in the receivers and for each digital data train received,
`of the corresponding selected types of modulation and-
`/or encoding efficiency levels.
`These two embodiments may also be implemented
`simultaneously, it being possible for each set of carriers
`to use at least two types of modulation and/or two
`encoding efficiency levels, as a function of the data to
`be transmitted.
`Preferably, said types of modulation are phase and/or
`amplitude modulation.
`In another advantageous embodiment, the device of
`the invention includes, for at least one of said carriers,
`means for the optimal association of the encoded digital
`elements with the states of the constellation of the mod-
`ulation, according to the so-called technique of trellis or
`lattice—encoded modulations.
`To enable a coherent demodulation, the device ad-
`vantageously includes means for the insertion of a fre-
`quency synchronization pattern recurrent in time, mak-
`ing it possible to carry out a coherent demodulation in
`said receivers.
`Preferably, the transmission device of the invention
`includes at least two channel encoders using identical
`generating polynomials so as to enable the use, in the
`receivers, of a sa.me decoder for several data trains
`encoded by distinct encoders.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`Other features and advantages of the invention will
`appear from the following description of an embodi-
`
`l0
`
`15
`
`20
`
`25
`
`30
`
`35
`
`45
`
`50
`
`55
`
`65
`
`5,197,061
`
`4
`merit, given by way of a non-restrictive illustration, and
`from the appended drawings, of which:
`FIG. 1 shows curves of the ratio of the energy per
`useful bit to the spectral density of noise as a function of
`the spectral efficiency of different channel encoding
`modes, in the case of the Gaussian and Rayleigh chan-
`nels;
`‘
`FIG. 2 is a block diagram of a transmission device
`according to the invention;
`FIG. 3 is a block diagram of a overall transmission
`and reception chain according to the inventions, show-
`ing the encoding and decoding parts;
`FIG. 4 is an example of interleaving of the sets of
`carriers, in the case of three difference sources, from the
`viewpoint of the level of protection against the trans-
`mission errors;
`FIG. 5 shows a detailed block diagram of a transmis-
`sion device according to FIG. 2, in the case of an appli-
`cation to two protection levels;
`FIG. 6 shows the detailed block diagram of a recep-
`tion device corresponding to the transmission device of
`FIG. 5.
`FIG. 7 shows a detailed block diagram of a selection
`means for selecting one of several possible modulation
`means.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`The device of the invention enables the optimum
`resolution of the problem of transmission of different
`data sources requiring different protection levels. It is
`based on the use of the COFDM method. Indeed, each
`of the carriers of the OFDM multiplex is modulated
`independently, thus making it possible to apply different
`modulations to them.
`Thus, for example, it is possible to envisage the use,
`for the transmission of essential data, of a 4-PSK modu-
`lation, and for less significant data, of an 8-PSK or 16-
`PSK modulation. This latter modulation will be less
`robust than the former one, but each carrier will carry
`1.5 times (8-PSK) or twice (16-PSK) as much informa-
`tion, for equal encoding techniques, thus leading to an
`increase in the final bit rate, without modifying the
`error rate associated with the essential data.
`The overall bit rate D of binary information coming
`out of a source encoder to be transmitted on a multiplex
`of N carriers in a given band chaimel B, where B=N/ts,
`ts being the duration of an elementary symbol, can be
`written as:
`
`II
`D = _2 Di,
`(=1
`
`where n is the number of sources.
`If the different sources call for different protection
`levels with respect to the transmission errors, the bit
`rate values D; may be matched with each of the sources.
`It is notably possible, with the COFDM method, to
`adapt to this type of differentiated sources by acting on
`the efficiency R; of the code associated with the bit rate
`D; source, for example by using puncturing techniques.
`FIG. 1 shows two curves of the ratio of energy per
`useful bit to the spectral noise density (Eb/No), for a
`binary error rate of 10-4, as a function of the spectral
`efficiency (in bits/s/Hz) of the modulation, for several
`types of modulation (4—PSK, 8-PSK, 16-QAM),
`in
`Gaussian and Rayleigh channels. For a 4-PSK encod-
`
`Page 00009
`
`Page 00009
`
`

`
`5
`ing, it is possible to make the efficiency vary by i to 8/9,
`the spectral efficiency then varying by 0.5 bit/s/Hz to
`nearly 2 bits/s/Hz. At the same time, the error rate
`increases substantially, notably in the case of disturbed
`channels, of the selective Rayleigh channels type. Fur-
`thermore,
`the spectral efficiency remains below 2
`bits/s/Hz.
`It is therefore more worthwhile, from the viewpoint
`of power efficiency, to pass on to constellations of mod-
`ulation with a greater number of states associated with
`adequate methods of encoding according to the princi-
`ple of Ungerboeck lattice—encoded modulations (LEM).
`It is noted, for example, that it is better to use an 1-PSK
`modulation with an efficiency R=§ (with an LEM
`encoding) than a 4-PSK modulation with an efficiency
`R=8/9 (LEM encoding).
`The system of the invention also enables acting on the
`type of modulation of each carrier. This carrier will be
`characterized by the number of bits nb,-bome per modu-
`lation state. A carrier i will therefore have 2"“ states.
`To the bit rate D,-, there therefore corresponds, at the
`output of the encoder, a bit rate D,-/R; to be distributed
`over N; carriers modulated with 2"“ states, with the
`following relationships:
`
`N," = (D; - B)/R,‘ - Ilbi
`
`To obtain optimal results, it will be sought to adapt
`D.- and ts so that N; is an integer.
`If we apply the principle of lattice-encoded modula-
`tion described by Ungerboeck in “Channel Coding
`With Multilevel Phase Signal”, IEEE Transactions,
`Information Theory, Vol. l.T. 28th Jan. 1982, i.e. the
`optimum association of ng+l bit encoded words coming
`out of an encoder having an efficiency value of R,-=-
`n,-/(n,-+1) with the states of the constellation of 2'"'+1
`state modulation so as to maximize the distance between
`signals, we also having the following relationship:
`
`nb;= ni+ l
`
`or again R,-nb.-=n,~.
`The optimum association between encoded words
`and states of modulation by lattice encoding enables a
`major encoding gain, for equal spectral efficiency, as
`compared with a modulation system having a 2" state
`system without encoding.
`FIG. 2 shows a block diagram of a transmission de-
`vice with n data sources S1 and Sn according to the
`invention, with ii types of modulation and, hence n
`different encoding efficiency values R,~.
`After the operation 20,- (with i varying from 1 to n) for
`the encoding of each series of data having a bit rate D;
`with an efficiency R; and allocation 21,- optimized of a
`modulation state according to Ungerboeck’s method,
`we therefore obtain complex symbols C('7_,-,k, belonging
`to an alphabet having 2'"'+1 states. The symbols C(0~,k
`are then interleaved (22,-) in time and in frequency and
`then, according to the known COFDM method, they
`undergo a reverse Fourier transform 23 to give the
`signal to be transmitted.
`
`5,197,061
`
`6
`
`,
`
`+
`
`n
`
`5
`
`wfih:
`
`_
`Card (11)=Ni
`ilg-,1;(t)=gl<(t—_1ts) for Oét-Es
`gk(t)=e2"'/7“ for 0§t_S_ts 0 again
`
`W
`
`fk=fo+k/ts
`
`i: index of the alphabet of modulation
`k: temporal index of the symbols
`k: index of the carriers
`
`the complex carriers received after
`At reception,
`demodulation and discrete Fourier transform have the
`form:
`
`20
`
`25
`
`Yj.k“’=H;.zrC"’;.k+1V;.k
`
`where N,~,k(*7 represents a complex Gaussian noise and
`H,-,1, the response of the channel.
`Each decoding process, associated with the index i,
`then minimizes the following expression according the
`criterion of maximum a posteriori likelihood:
`
`22 ll Yj,k"7-Hj,IcC" j,k ll /20“ ',k
`
`30
`
`E
`
`where 0'2,-,1; is the variance of each component of com-
`plex Gaussian noise NM.
`The invention is not restricted to the use of several
`types of modulation. It is notably possible to use also the
`puncturing technique or any other technique to adapt
`the efficiency of the encoding with one or more types of
`modulation.
`
`FIG. 3 shows the general block diagram of a trans-
`mission and reception chain according to the invention,
`implementing several modulations, and the RCPC
`puncturing technique.
`This system achieves the differential encoding of five
`sources of data S1 to S5 calling for levels of protection
`against distinct and decreasing transmission errors.
`the fust three data sources S1, S2 and S3 are encoded
`according to a 4-PSK modulation 311, 312 and 313 with
`punctured codes having respective efficiency values
`Rl=§, R2=§ and R3=i in the encoders 301, 302 and
`303.
`The data source S4 is processed by a lattice encoder
`304 with an efficiency R=§ and an 8-PSK modulation
`314, and the data source S5 is processed by a lattice
`encoder 305 with an efficiency R5= 5/6 and a 64~QAM
`315 modulation (64-state quadrature amplitude modula-
`tion), both being processed according to a lattice modu-
`lation technique.
`Advantageously, the generating polynomials of the
`encoders 304 and 305 are identical so that the encoded
`data can be decoded at reception by only one decoder
`37 ifthis decoder is made in a way that can be suffi-
`ciently parametrized.
`According to the known COFDM encoding tech-
`niques, the different pieces of encoded data are sub-
`jected to a reverse fast Fourier transform (FFT'-1) 32,
`and then transmitted by the transmitted channel 33.
`At reception, the demodulation 34 may be either
`differential (for the PSK modulations) as in the radio
`broadcasting system described in the above-mentioned
`U.S. Pat. No. 4,881,241 or done coherently, as pres-
`
`45
`
`50
`
`S5
`
`60
`
`65
`
`Page 00010
`
`Page 00010
`
`

`
`7
`ented in the French patent application No. 90.01492
`dated Feb. 6th 1990 and filed on behalf of the same
`Applicant. It is clear, by contrast, that a QAM modula-
`tion can be demodulated only coherently.
`In the latter case, one method consists in the intro-
`duction, into the transmitted multiplex, of a frequency
`synchronization pattern that is recurrent in time, en-
`abling the decoders to recover a phase and/or ampli-
`tude reference.
`
`The reception part then includes a fast Fourier trans-
`form (FFF) 35 in which the reverse of the FFT—1
`operation 32 is performed, and then the decoding itself.
`The choice of identical encoding generator polyno-
`mials enables the number of decoders in the receiver to
`be limited.
`
`Thus, in the example given, the three sources S1, S2
`and S3 could be decoded by the Viterbi decoder 36.
`The two sources S4 and S5, processed by the two lattice
`encoders 304 and 305 having the same polynomials,
`could also be decoded by the same Ungerboeck decoder
`37.
`The COFDM system fully uses the two dimensions,
`namely the temporal and frequency dimensions, by its
`wideband character and by means of the time-frequency
`interleaving which, in being associated with the method
`of de-interleaving at reception, makes it possible to
`obtain, at the input of the decoder, the maximum statis-
`tical independence of the successive samples with re-
`spect to disturbances due to transmission.
`The method of the invention makes it possible to lose
`nothing in terms of frequency independence if we use an
`optimum frequency multiplexing of the different combs
`of the carriers associated with the different sources Di.
`For this purpose, the different sets of carriers are
`interleaved along the frequency axis. For example, in
`the case of three different sources, the multiplexing
`could of the kind shown in FIG. 4 for the three sets of
`carriers J1, J2, J3. In this case, each of the three sets of
`carriers benefits from the independence in frequency,
`related to the total bandwidth.
`Thus, the method of the invention remains optimal
`for each source Di in terms of power and spectral effi-
`ciency.
`The method described by Ungerboeck, defining the
`right codes and relying on the optimum association of
`the encoded words with the states of the constellation
`according to the criteria of maximization of distance
`between signals makes it possible to organize the perfor-
`mance characteristics independently for each of the
`sources Di.
`An example of an application with numerical values
`is given here below. It can be applied notably to the
`broadcasting of sequences of images distributed among
`two complementary trains of data elements bl and b2,
`as described in the joint patent application filed on the
`same date on behalf of the present Applicant.
`In this case, the modulation and encoding parameters
`are fixed. The devices described can nevertheless be
`adapted to a different choice of these parameters.
`A transmission channel identical to the one used in
`the sound broadcasting system already made it used.
`The available width of the transmission channel
`is
`B=N/ts=7 MHz. The width of the symbols Ts= 80 us
`(including the duration of the useful signal ts=64 p.s
`and a guard interval A= 16 us). The number of carriers
`of the multiplex N is then equal to 448.
`It is therefore proposed to use two different levels of
`protection with respect to transmission errors.
`
`l0
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`55
`
`65
`
`5,197,061
`
`8
`the first level, associated with the first data train bl,
`corresponds to the method used during the first imple-
`mentation of the COFDM encoding in the known radio
`broadcasting system. The following are its parameters:
`4—PSK modulation, demodulated coherently, giving a
`spectral efficiency
`nb1=2 binary elements per Hertz (eb/Hz)
`code efficiency Rl=§
`number of carriers of the associated OFDM multiplex
`equal to N1.
`The useful bit rate transmitted D1 is therefore equal
`to:
`
`D1=nb1XR1X(N'l/u')X(t.r/T.r)X(ts/Tr)=2><(§)><-
`(N1/tS)X(4/5)
`
`If we lay down Nl=224, namely half of the available
`carriers, we obtain a useful bit rate D1=2.8 Mbit/s.
`The second protection level, associated with the sec-
`ond data train, makes use of lattice-encoded modulation
`techniques (Ungerboeck techniques) in achieving a
`closer association of a lattice code with a modulation
`with a large number of states. The following are its
`parameters:
`8-PSK modulation coherently demodulated, giving a
`spectral efficiency of nb2=2 eb/Hz,
`efficiency of the code R2=§,
`the number of carriers is N2.
`The transmitted useful bit rate D2 benefiting from
`this second level of protection is equal to:
`
`D2=nb2><R2><(N2/Lt)><(ts/Ts)=3><(§)><(N2/ts)-
`><4.5
`
`If we fix N2=224, we get:
`
`D2=5.6 Mbits/§.
`
`The two data trains preferably include data of differ-
`ent degrees of importance, notably according to a psy-
`chovisual criterion. The method of the invention ena-
`bles the transmission of the most pertinent data, corre-
`sponding to the train bl, by means of a sufficiently
`robust encoding. The less significant data of the train B2
`is transmitted with less efficient protection against the
`transmission errors (this is not troublesome) and with a
`double useful bit rate D2.
`FIG. 5 shows a block diagram of transmission equip-
`ment corresponding to the above-described example.
`the pieces of data 50 coming from the source are
`separated into two binary trains bl and b2, with respec-
`tive bit rates D1, D2, by a distribution module 51.
`The first binary train bl is processed in a way similar
`to that applied during the first setting up of the
`COFDM system. A convolutive encoding 52 is there-
`fore done, with efficiency R1.-=1, then a time-frequency
`interleaving 53 is carried out followed by a signal binary
`encoding 54. Complex data C_,-,1, are then obtained.
`These pieces of data are processed for transmission in
`the COFDM modulation module 56.
`The second binary train b2 undergoes an Ungerboeck
`type convolutive encoding 57, or lattice encoding, with
`2’‘—‘ states (1: being the constraint length) and with
`efficiency R2=§, then an operation 58 is done for the
`association, with each triplet of bits coming from the
`lattice encoder 57, of a signal an of the constellation of
`PSK modulation according to the method described by
`Ungerboeck under the term “set partitioning” in the
`already mentioned document.
`
`Page 00011
`
`Page 00011
`
`

`
`9
`The signal a,. may be written:
`
`a,=Jk»/8+v/I6, mo . . . ,7}
`
`5 , 197,06 1
`
`in
`
`This signal an is then time and frequency interleaved 5
`(59) and then directed towards the COFDM modula-
`tion module 56.
`In a known way, this module 56 notably achieves a
`reverse fast Fourier transform on complex 512 word
`blocks and a digital-analog conversion.
`The resultant complex sample then modulates a car-
`rier in phase and in quadrature to produce the signal 60
`to be transmitted.
`FIG. 6 shows a block diagram of the complete recep-
`tion equipment corresponding to the above-described
`ii-iiiisiiiiiton The iooeivod sigiiai 50 is processed by the 15
`CQFDM demodulation module 51 which iiombiy
`carries out a channel filtering, a demodulation on two
`channels in quadrature with reference to its central
`frequency, a digitization and a processing operation by 20
`a processor of the signal which carries out a fast Fourier
`transform (FFT).
`A function 62 for the estimation of the carriers of the
`OFDM multiplex is used to make the projection 63 on
`the two axes of the complex plane, using frequency
`synchronization words, so as to carry out a coherent 25
`d¢1'n0dl1latl0n-
`The {W0 information trains b1 and b2 are then de-
`coded separately. The train bl undergoes a time-fre-
`Cluency Cl¢'l1lt°fl¢3Vl118 54 and l5 then de°0d°d by 3
`Viterbi decoder‘ 65. The second train b2 is also de-inter- 30
`l¢3V¢<l (56) 111 “me and frequfiflcyy 9-I1C_l d€C0<‘l¢d by 311
`Ungerboeck decoder 67. The data coming from the two
`decoders 65 and 67 are then assembled by a multiplexer
`68 5° 35 t° give the °°mP1°t° data signal 69'
`,
`In th? ‘ixaiilple d°S_°F“’°d» °°n°eming the br°ad°ast' 35
`mg of digital images, it is possible to make a second type
`°f r°.°°iV°r’ wifich is simpler and. include? only ihe pm"
`cfissléig °.gen.m°nb:°lated Kathe mfomianot? tram
`if
`3 °
`him. lmon
`.tw.°°i1 t 6 two ‘.“““S
`1 and
`‘S 40
`_ one judiciously, it is indeed possible to reconstruct
`3213195 by mums of the Kai.“ bl alone‘ These images
`clearly be of lower quality, but they will however
`be acceptable, notably for small-sized screens.
`.
`.
`.
`.
`.
`These receivers using only the tram bl, which is more
`robustly encoded, may especially be used under difficult
`conditions of reception, for example reception in mobile
`receivers in an urban environment.
`It is clear that the above-described invention in no
`way restricts the scope of the invention. There may be
`,
`.
`_
`,
`_
`_
`_
`any number of sources of information or data trains to 50 pnsingf
`be processed with distinct protection levels. The pro-
`dlsftnbutmg mans f°’ dmnbumg smd d‘5‘t"1 data
`teenen level my be adapted by nenng either on the
`into at least two sets of data elements, a first of the
`code efficiency used or on the type of modulation.
`5°‘5_°f dam °l°m°“t-5 ’°‘i““'m3 3 5'“ l°V°1 °f Pm‘
`Moreover, the invention is applicable not only to the
`t°°“9‘_‘ and 5'1"“ °“° 5°°°nd 5°‘ 9f dam °l°m"-ms
`broadcasting of digital images but also to sound broad- 55
`rociumns a_-second level of protection;
`i
`_
`.
`casting and, more generally, to the broadcasting of any
`firs‘ m°d“l3t1°n_m°311Ss °°“Pl°d *0 313 dlsmbunng
`type of digital information. It enables the differentiated
`11153“? ‘P 1'°¢¢1V° 31° firs‘ 5°‘ °f dim’ Clenlentss ‘'3'
`processing, not only of sub-sets of one and the same
`353°C13tmS 31¢ first 5“ 0f data ¢l¢m¢m5 ‘V1111 3 fifst
`program but also of completely independent programs.
`5¢1'l¢-5 0f m°d“l3tl°n Symbols S¢l°°t°d 3°°°1'¢ll118 t0
`In another embodiment as illustrated in FIG. 7, the 60
`3 firs‘ m°<l11l3il°11 Symbol 8lPll3b€t
`°°FT¢SP0I1€llI18
`modulation and/or the encoding efiiciency assigned to
`to a first type of modulation providing for said first
`each carrier or set of carriers may be variable, for exam-
`l¢V¢l Of Pf0t¢¢tl°n;
`ple as a function of the importance of the information to
`firs‘ lnt¢1'l€3ViIl8 m¢a!1-S: 0009185 10 thfi first modula-
`be transmitted at each instant. A selector means 71
`tion means to supply each frequency carrier of a
`delivers the signals to be channel encoded to one or the 65
`first selected set of said plurality of orthogonal
`other of two channel encoders 721, 722 and two modula-
`frequency carriers with a distinct succession of
`modulation symbols picked out of said first series
`tion means 731 and 732. This selector means 71 is con-
`of modulation symbols, for interleaving in both
`trolled by a selection module 74, taking into accou