`USO05191576A
`
`[191
`United States Patent
`Pommier et al.
`[45] Date of Patent: Mar. 2, 1993
`
`[11] Patent Number:
`
`5,191,576
`
`
`
`[54] METHOD FOR BROADCASTING OF
`DIGITAL DATA, NOTABLY FOR RADIO
`BROADCASTING AT HIGH THROUGHPUT
`RATE TOWARDS MOBILE RECEIVERS,
`WITH TIME FREQUENCY INTERLACING
`AND ANALOG SYNCI-IRONIZATION
`
`[75]
`
`Inventors: Daniel Pommier, Breal Sur
`Montfort; Bernard LeFloch,
`Rennes, both of France
`
`[73] Assignees: Etat Francais and Telediffusion de
`France S.A., France; Etat Francais
`and Telediffusion de France S.A.,
`France
`
`[21] Appl. No.: 777,463
`
`[22] Filed:
`
`Oct. 17, 1991
`
`Mobile Radio-I. Sabbagh, B.Sc., M.Sc., and D.G. Ap-
`pleby, B.Sc. (Engl), C.Eng. M.l.E.E., IEE Proceedings,
`vol. 132, Pt.F. No. 5, Aug. 1985.
`PC Communications: The Revolution is Coming, Fea-
`ture, Brig. Gen. H. R. Johnson, USAF (Ret.).
`Pommier, et al., “New Prospects for High Quality Digi-
`tal Satellite Sound Broadcasting to Mobile, Portable,
`and Fixed Radio Receivers,” IBC ’88 Brighton, 23-27
`Sep. 1988 (IEE Conference Publication No. 293).
`Principles of Digital Communication and Coding,
`Viterbi and Omura, McGraw-Hill, 1979, pp. 78-83,
`150-159, 242-253.
`'
`1
`Viterbi, “convolutional Codes and Their Performance
`in Communication Systems”, IEEE Transactions on
`Communications Technology, vol. Com-19, No. 5, Oct.
`1971.
`.
`
`Foreign Application Priority Data
`[30]
`Nov. 18, 1988 [FR]» France ................................ 88 15216
`
`Primary Examiner-—Benedict V. Safourek
`Assistant Examiner—Alpus H. Hsu
`
`Int. Cl.5 ................ .. H04J 11/00; H04L 27/28
`[51]
`[52] U.S. Cl. ...................................... .. 370/18; 370/21;
`370/50; 370/69.1; 370/70; 375/38; 375/58;
`455/59
`[58] Field of Search ..................... .. 370/18, 19, 21, 23,
`370/50, 69.1, 70, 100.1, 101, 105.4, 105.5, 111;
`375/38, 40, 58, 60, 99, 101, 107, 108, 111, 112,
`113; 455/50, 54, 59, 63; 358/12, 142, 143;
`381/2, 13, 14; 371/43, 46
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`........................ .. 370/70
`9/1971 Cutter et al.
`3,605,019
`370/101
`l/1987 I-iatabe ........
`4,638,478
`
`375/113
`1/1989 Laurent
`4,799,241
`358/142
`4,884,139 ll/1989 Pommier
`4,922,483
`5/1990 Kobayashi ............................ 370/50
`
`FOREIGN PATENT DOCUMENTS
`1443881
`5/1966 France .
`.
`88/00417
`1/1988 PCT Int’l Appl.
`703247
`2/1954 United Kingdom .
`
`OTHER PUBLICATIONS
`
`[57]
`
`'
`
`ABSTRACT
`
`A method for the diffusion of digital data designed to be
`received notably by mobile receivers moving in an
`urban environment, that is,
`in the presence of stray
`signals and jamming, and in conditions of multiple prop-
`agation (Rayleigh process) Providing an optimized
`mode of setting up the frame structure of the broadcast
`signal, so as to derive the maximum benefit from the
`resistance of the system to pulsed stray signals and to
`jamming. The header of each frame has a first empty
`synchronization symbol and eventually a second, un-
`modulated wobbled signal forming a two-stage analog
`synchronization system. So, the recovery of synchroni-
`zation is achieved in an analog way, without prior ex-
`traction of a clock signal at the binary level. The consti-
`tution of the sequence of the useful symbols in the frame
`results from temporal and sequential interlacing opera-
`tions, combined to obtain an implicit de-interlacing at
`the receiver. The empty symbol may be further used for
`the extraction of the jamming affecting the transmission
`channel.
`
`Adaptive Slow Frequency-Hopping System for Land
`
`13 Claims, 5 Drawing Sheets
`
`
`
`Aruba Networks et al. Exhibit 1006 Page 00001
`
`Aruba Networks et al. Exhibit 1006 Page 00001
`
`
`
`U.S. Patent
`
`Mar. 2, 1993
`
`Sheet 1 of 5
`
`5,191,576
`
`Symbols of s nchronization
`
`Channel
`
`10
`
`.
`
`
`
`CHANNEL
`
`Time-frequency
`
`
`interlacing
`
`
`
`ULTIPLEXING
`
`
`‘'‘‘"“°‘ ‘N
`MODULATION
`(own)
`
`
`
`Fig 1
`
`Physical synchronization
`of the _12,5.KHz receiver
`( periodicity 8095 i
`
`22
`
`
`
`Page 00002
`
`Page 00002
`
`
`
`U.S. Patent
`
`Mar.2, 1993 '
`
`Sheet2of5
`
`A 5,191,576
`
`24
`
`‘
`
`DETECTION 25
`
`ANALYSIS
`
`
`
`TIME
`
`BASE
`
`\52
`
`ANALOG — 50
`E
`I
`SYNCHRONIZATION
`15' STAGE or
`:
`2"” STAGE or
`SYNCHRONIZATION
`I
`SYNCHRONIZATION
`IFIIG. filb
`
`.
`
`I
`
`Page 00003
`
`Page 00003
`
`
`
`U.S. Patent
`
`Mar. 2, 1993
`
`’
`
`Sheet 3 of 5
`
`5,191,576
`
`Page 00004
`
`Page 00004
`
`
`
`U.S. Patent
`
`Mar. 2, 1993
`
`Sheet 4 of 5
`
`5,191,576
`
`i
`
`0
`
`1
`
`2
`
`3
`
`4
`S
`6
`
`7
`
`a
`
`9
`
`10
`
`11
`
`12
`
`13
`
`19
`E
`
`A34
`
`A35
`
`036
`
`037
`
`A33
`
`439
`
`A00
`
`041
`
`402
`
`A43
`
`khk
`
`has
`
`Ahé
`
`407
`
`Page 00005
`
`i
`
`0 1 2
`
`3
`
`A
`
`5
`
`5
`'7.
`0
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`17
`!
`495
`
`496
`
`497
`
`490
`
`A99
`
`500
`
`501
`
`502
`
`503
`
`504
`
`505
`
`506
`
`507
`
`500
`
`509
`
`sun
`
`Sn
`
`Page 00005
`
`
`
`U.S. Pateot
`
`Mar.‘2, 1993
`
`Sheet 5 of 5
`
`5,191,576
`
`Analog-Digital
`
`conversation
`
`Time-Frequency
`' de—inferlacing
`
`
`
`V‘777]/
`Weighting/
`of metrics’
`
`FFT* with N
`complex points
`
`>PageV00006'
`
`-
`
`Page 00006
`
`
`
`1.
`
`5,191,576
`
`METHOD FOR BROADCASTING OF DIGITAL
`DATA, NOTABLYFOR RADIO BROADCASTING
`AT HIGH THROUGHPUT RATE TOWARDS
`MOBILE RECEIVERS, WITH TIME FREQUENCY
`INTERLACING AND ANALOG
`SYNCHRONIZATION
`
`This application is a continuation of application Ser.
`No.’ 07/439,275, filed Nov. 20, 1989, now abandoned.
`BACKGROUND OF THE INVENTION
`
`5
`
`l0
`
`2
`f/time t space and the emission of frames of symbols
`each formed by a multiplex of orthogonal carrier fre-
`quencies modulated by a set of digital elements and
`broadcast simultaneously on M parallel channels,
`wherein the recovery of synchronization of each
`frame received is achieved by analog synchronization
`means without prior extraction of a clock signal.
`According to an advantageous characteristic of the
`invention, the header of each of said frames of symbols
`comprises an interval of silence, with the duration of a
`digital symbol of the frame. This symbol of silence is
`capable of being used as a means of synchronization of
`the demodulation and/or as a means of analysis of the .
`pulsed noise and of the jamming which are characteris-
`tic of the channel.
`
`1. Field of the Invention
`The field of the invention is that of the broadcasting
`of digital data designed to be received notably by mov-
`ing receivers in an urban environment, namely in the
`presence of interferences or jamming, under conditions
`of multiple propagation (RAYLEIGH process) gener-
`ating a phenomenon of fading.
`The invention can be applied more particularly, but
`not exclusively, to a system of digital sound broadcast-
`ing as described in the French patent applications Nos.
`86 09622 ofJul. 2, 1986 and 86 13271 of Sep. 23, l986, on
`behalf of the same applicants. This system of digital
`broadcasting, presented in these prior patent applica-
`tions, is based on the combined use of a channel coding
`device and a method known as the COFDM system
`(coding orthogonal frequency division multiplex sys-
`tern).
`2. Description of the Prior Art
`The modulation method proper of this prior art sys-
`tem consists in providing for the distribution of the
`constituent digital elements of the data signal in the
`frequency-time f-t space and in simultaneously emitting
`sets of digital elements on M parallel broadcasting chan-
`nels by means of a multiplex of orthogonal carrier fre-
`quencies. This type of modulation makes it possible to
`prevent two successive elements of the data train from
`being emitted at the same frequency. This enables the
`absorption of the fluctuating selectivity in frequency of 40
`the channel, by frequentially dispersing the initially
`adjacent digital elements during the broadcasting.
`The prior art encoding method seeks, for its part, to
`enable the processing of the samples coming from the
`demodulator to absorb the effect of variation in ampli-
`tude of the signal received, due to the RAYLEIGH
`process. This encoding is advantageously a convolutive
`encoding, possibly concatenated with a REED-SOLO-
`MON type encoding.
`In a known way, the encoded digital elements are
`furthermore interlaced, in time as well as frequency, in
`order to maximize the statistical independence of the
`samples with respect to the Rayleigh process and the
`selective character of the channel.
`
`15
`
`20
`
`25
`
`30
`
`35
`
`45
`
`50
`
`55
`
`SUMMARY OF THE INVENTION
`
`An aim of the present invention is to provide an opti-
`mized embodiment of the frame structure of the broad-
`cast signal so as to derive the maximum benefit from the
`self-synchronization
`properties of
`the COFDM
`method, and to maximize the resistance of the system to
`the pulsed interferences and jamming.
`This aim as well as others which shall appear subse-
`quently are achieved by means of a method for the
`broadcasting of digital data, notably for sound broad-
`casting at a high_ throughput rate towards mobile re-
`ceivers, of the type providing for the distribution of said
`data in the form of digital elements in the frequency
`
`According to another characteristic advantage of the
`invention, said frame header comprises an unmodulated
`multiplex of said M orthogonal carrier frequencies, with
`the duration of a digital symbol of the frame. This un-
`modulated symbol may be used as a synchronization
`means and/or as a phase reference for the J phase-
`modulated carriers of the digital train.
`Preferably, said symbols are formed by means of a
`frequential interlacing operation using a reversible de-
`terministic function,
`said function consisting in a
`method for shuffling the indices of said frequencies with
`a maximization of the dispersal of the frequencies asso-
`ciated with adjacent digital elements of the source data
`signal. Said shuffling of indices advantageously consists
`in applying a function of the bit inversion type to said
`binary encoded indices.
`Preferably, the phase modulation done on the carriers
`is of the type with four phase states, each carrier being
`modulated by a pair of digital elements, and said pairs
`are formed by a source sequence of digit_al elements in
`forming packets of 2J consecutive elements in said se-
`quence, and in associating the elements two by two in
`each packet according to a criterion of maximization of
`dispersal of the adjacent digital elements of the source
`sequence.
`In an advantageous mode of the invention, the pairs
`are formed by splitting each of the said data packets into
`two half-packets and by pairing the same-order digital
`elements in each half-packet.
`The frequential interlacing thus defined is advanta-
`geously combined with a temporal interlacing achieved
`by the application of delays, the value of which is as-
`signed to each digital ‘element by the application of a
`reversible function of the index of the digital element,
`the delay function (F) being such that the deinterlacing
`in the initial order and the recovery of each element of
`the source sequence are achieved by the application, to
`each digital element with a same index in the sequence
`received, of a complementary delay value with respect
`to the depth of the maximum interlacing of the delay
`function.
`According to another characteristic of the invention,
`' the method includes a jamming extraction process com-
`prising the following steps:
`the received signal is analyzed during said symbol of
`silence, on the spectrum covered by the J orthogonal
`carriers;
`the frequencies affected by complex Gaussian noise are
`identified;
`a correction and/or cancellation processing is done of
`the useful signal received by said detected jammed
`signals.
`
`65
`
`Page 00007 ‘
`
`Page 00007
`
`
`
`3
`Advantageously, said spectral analysis is comple-
`mented by a two-dimensional filtering step in the time/-
`frequency space, providing for a smoothing of the re-
`sults of said analysis on the useful extent of the analyzed
`spectrum.
`-
`According to a complementary characteristic of the
`invention, the method is of the type implementing a
`convolutive encoding of the data at the transmitter, and
`a soft decision decoding through maximization of likeli-
`hood at the receiver,
`wherein said processing of correction and/or cancel-
`lation of the useful signal received for said detected,
`jammed signals consists in informing said soft decision
`making by means of the noise power detected at each of
`said frequencies.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`Other features and advantages of the invention will
`appear from the following description of an embodi-
`ment given by way of a non-restrictive example, and
`from the appended drawings, of which:
`FIG. 1 is a block diagram of a transmission-reception
`chain implementing the method of the invention;
`FIG. 2 gives a schematic view of the structure of a
`frame as broadcast by the system of the invention;
`FIGS. 3a, 3b respectively give a schematic view of,
`firstly, a standard acquisition chain of synchronization
`by clock signal extraction and, secondly, the principle
`of analog recovery of synchronization with two stages
`according to the invention;
`‘
`‘
`FIG. 4 gives a schematic view of a time-frequency
`interlacing/deinterlacing chain, optimal in cooperation
`with the synchronization principle of the invention;
`FIG. 5 shows an advantageous mode of a convolutive
`temporal interlacing that can be implanted in the chain
`of FIG. 4;
`'
`FIG. 6 illustrates an advantageous mode of a shuf-
`fling of frequency indices compatible with the frequen-
`tial interlacing of the chain of FIG. 4;
`FIG. 7 gives a schematic view of a reception chain
`with extraction of jamming, ‘compatible with the high
`throughput rate digital broadcasting method of the
`invention.
`
`.
`
`DESCRIPTION OF A PREFERRED
`EMBODIMENT
`
`The different aspects of the embodiment which shall
`be described hereinafter more particularly concern digi-
`tal sound broadcasting towards mobile receivers, as
`defined notably in the EUREKA Digital Audio Broad-
`casting (DAB) program.
`However, it is clear that the high throughput digital
`broadcasting principle of the invention can be applied
`to any type of communications, notably in channels
`subjected to the Rayleigh process such as, for example,
`aircraft-satellite or other types of communications.
`In the digital sound broadcasting application of the
`DAB, one aim may be, for example, the transmission of
`sixteen stereophonicprograms in an 8 MHz wide fre-
`quency band with a digital throughput rate of the order
`of 100 kbits (after compression).
`A transmission chain of the type described in the
`patent applications mentioned in the introduction is
`shown in FIG. 1.
`Each of the N(16) channels Co to C,._1 undergoes an
`encoding 10 in parallel, then a time-frequency interlac-
`ing II on a separate channel, before beingsubjected
`
`5,191,576
`
`4
`
`,
`
`l0
`
`l5
`
`20
`
`25
`
`30
`
`35
`
`jointly to a process 12 of temporal multiplexing and
`OFDM modulation.
`The encoding 10 is advantageously of the convolu-
`tive type. The time-frequency interlacing 11 is aimed at
`shuffling the digital elements of each channel in order to
`give them maximum independence with respect to in-
`terferences and to the jamming of the broadcasting
`channel 13.
`'
`The OFDM modulation consists in the modulation of
`symbols each formed by a multiplex of orthogonal fre-
`quencies broadcast simultaneously on J channels. This
`operation can be achieved by a Fourier transform on
`the encoded and interlaced digital sequence of each
`channel C,-.
`By way of example, in an 8 MHz frequency band, it is
`- possible to define 512 separate 15 625 Hz carrier fre-
`quencies. Of these, 448 are usable, after elimination of
`the central frequency of the spectrum and of the lateral
`carriers (fith of the spectrum) to take the filtering con-
`straints into account.
`,
`The reception chain comprises the steps of channel
`selection 14, demodulation 15, frequency de-interlacing
`16 and decoding 17 of the de-interlaced channel.
`The channel selection operation 14 is performed ad-
`vantageously by Fast Fourier Transform (FFT) so as to
`decimate the set of suitably interlaced carriers to apply
`the OFDM_ demodulation operation only to the carriers‘
`of the selected channel (see addition certificate No. 86
`13721 already referred to). After the time-frequency
`de-interlacing 16, a “soft” decision Viterbi decoding 17
`is advantageously applied.
`The data frame, as broadcast through the channel 13
`presents, according to the invention, the structure of
`FIG. 2.
`The frame is formed by a header 21 and N elementary
`channels 22 marked Coto C,,_1 each formed by K sym-
`bols 23, marked Soto Sk_1. Each symbol 23 is formed
`by a multiplex of J orthogonal carriers. Each channel
`C,- represents a particular data flow independent of the
`information transmitted on the other channels.
`The header 21 of the frame includes an “empty” or
`“blank” interval 24 which is advantageously used to
`perform both an analog synchronization of the frame
`and an extraction of the jamming of the broadcasting
`channel.
`'
`The possibility of achieving an analog synchroniza-
`tion recovery on an “empty” symbol is a fundamental
`characteristic of the invention.
`For, in existing systems working at a high throughput
`rate and as shown in FIG. 3a, the recovery of synchro-
`nization is usually achieved in synchronization at the
`binary level, on the received train, by means of a clock
`51 working with synchronization means 52 with lock-
`ing. The recovered synchronization drives a time base
`system 53 which opens windows 54 in the wave train
`received to extract the useful frames therefrom. This
`. type of chain with locking of synchronization is made
`necessary by the need to work with very high precision,
`typically of the order of :5 ns for throughput rates of
`10 Mbits per second.
`For equal throughput rates, the broadcasting method
`of the invention makes it possible to work with consid-
`erably lower precision during the recovery of synchro-
`nization. In effect, since each symbol is formed by a
`multiple of J orthogonal carriers, the synchronization is
`achieved on symbols with a width that
`is J times
`greater. Thus, in the case of the use of 448 carriers in
`
`45
`
`50
`
`55
`
`60
`
`65
`
`Page 00008
`
`Page 00008
`
`
`
`5J9l§76
`
`10
`
`20
`
`25
`
`30
`
`35
`
`5
`parallel, the precision required at the recovery of syn-
`chronization is about 4.5 as.
`The assembly of FIG. 3b corresponds to the imple-
`mentation of a two-stage synchronization by recovery
`of two successive symbols of synchronization. The first
`synchronization symbol recovered is the “blank” sym-
`bol 24 of the frame header 21. The detection of an enve-
`lope 55 of the blank symbol sets off the time base 56
`which may be generated by a simple quartz-oscillator
`57 at 12.5 kHz for symbols with a duration of 80 p.s. The
`time base 56 opens windows 58 in the digital train re-
`ceived so as to recover the second recovery symbol 25.
`This symbol is formed by an unmodulated multiplex
`of the J carrier frequencies. It advantageously takes the _
`15
`form of a wobbling on the entire spectrum covered by
`the carriers but may be formed by any multiplex with a
`substantially constant envelope.
`The aim of the second stage of synchronization is to
`make a more precise resetting of the synchronization
`acquired at the first stage, by analysis of the pulse re-
`sponse of the channel. The self-correlation of the wob-
`bled symbol 25 thus enables increased precision in syn-
`chronization to be obtained. The detection of an enve-
`lope 61 of the second synchronization signal, after anal-
`ysis 62, resets the time base 56, and hence the sequenc-
`ing of the window 58 openings in the wave train re-
`ceived. The recovery of this symbol, with a duration of
`80 1.1.5, accomodates a precision of :2 us, and is there-
`fore compatible with an analog recovery chain.
`The analysis of the pulsed response of the channel
`makes it possible to take into account echo phenomena
`for the synchronization. Furthermore, a safety interval
`is advantageously provided between each symbol of the
`frame, with a view to absorbing this echo effects and
`limiting the intersymbol interference phenomenon. The
`safety interval typically has a value of 16 us, reducing
`the useful symbol period to 64 us.
`The synchronization symbols 24, 25 of the header 21
`of the frame may further each have a distinct second
`function.
`The blank symbol 24 may, in effect, serve to analyze
`theinterferences and jamming that affect the transmis-
`sion channel in order to take them into account, at the
`receiver, in the soft decision module as shall be seen
`further below.
`The wobbled synchronization symbol may, for its
`part, serve as a phase reference for the decoding of the
`useful signals 23 of the frame. In effect, advantageously,
`the reference phase of each of the J carriers of the multi-
`plex is locked in a distinct and specific way, so as to
`make it possible to restore each component of the multi-
`plex to the receiver in differential demodulation. Ad-
`vantageously, the locking of the reference phases is
`expressed by the formula:
`‘
`
`6
`powerful local transmitter, by means of a specific de-
`vice for the analysis of this information.
`In all, the principle of analog and implicit synchroni-
`zation of the frames, in the invention, makes it possible
`to avoid the drawbacks of the existing systems using
`synchronization words that are recognized at the binary
`level (consumption of throughput, risks of poor recog-
`nition of the word, total loss of the frame in the event of
`synchronization error). This determining advantage is
`added on to the optional possibility of a bi-functional
`use of the symbols of synchronization, as has just been
`presented.
`.
`The frame structure thus achieved results, according
`to the invention, from a dual operation of temporal
`interlacing and frequential
`interlacing of the source
`sequence (FIG. 4).
`The number of binary elements per channel of one
`and the same frame coming to the input of the interlac-
`ing system depends on the number J of carriers per
`symbol, the number K of symbols per channel and the
`number of states of the modulation applied to each
`carrier. In the case of a modulation with four phase
`states, the size of the blocks P1 of data (1 designating the
`index of the frame) presented at each frame at the input
`of this system is 2-J-K bits.
`By way of example, if J=448 and K=9 (k=0 to 8
`designating the order number of the symbol in the chan-
`nel), we obtain blocks P1 of data equal to 8 064 bits.
`Let P1,; be the index i bit of the block P1 (i=0 to
`2-J-K-1)
`.
`The temporal interlacing consists in forming a block
`Q1, of the same size as P1, the index i of which, marked
`Q1,,~is defined as follows:
`
`Ql,i= P1-110,1.
`
`On the choice of the function f(.) depends the depth
`and efficiency of the interlacing. In general, the image
`of f is the set F={0,1 .
`.
`.
`, m— 1}, where m designates
`the temporal depth of the interlacing.
`An example of an open-ended temporal interlacing is
`shown in FIG. 5. The example shown is of the type
`applying to a process with inversion of bits defined by
`the following interlacing function:
`fr: reciprocal number associated with n:
`if n is a number varying from O to 2p_ 1, written in the
`form:
`
`.
`-1
`]=
`n=[f}2 n,-2'n,‘=0orl
`
`the associated number n is equal to:
`
`.
`n,-2P“"'
`
`1
`0
`
`IIMI
`
`P i
`
`F1:
`
`The block P1 is interlaced according to the diagram of
`FIG. 5 so as to form a block Q1. The double change-
`over switch 81, 82 symbolizes the application of the
`interlacing function by successive switching over of
`each of the delay blocks 83. Let q1..1 be the i”' element
`of the block Q1. We have the relationship:
`
`411- 1 =PI—R(i/l6),i~
`
`The depth of interlacing is therefore 16 frames.
`
`Page00009
`
`45
`
`50
`
`¢k=1rkz/N
`
`with k=0 to N: index of each frequency.
`N:
`total number of frequencies of the multiplex
`(N =5l2 in the present example).
`Any other mode of computation of the locking phases
`is suitable, provides that it makes it possible to discrimi-
`nate the information conveyed by each of the carriers of
`the multiplex.
`'
`If necessary, the header 2] of the frame has a third
`symbol 26 which is a carrier of information such as the
`list of the local frequencies of emissions for the channel
`considered. A mobile receiver is then capable of getting
`automatically and permanently locked into the most
`
`55
`
`60
`
`65
`
`Page 00009
`
`
`
`8
`splitting each of the data packets into two half-packets
`and by matching the same order digital elements in each
`half-packet.
`According to an essential characteristic of the inven-
`tion, provided that the temporal interlacing function F
`and the frequential interlacing function G are appropri-
`ately chosen with respect to each other, the de-interlac-
`ing operation is done implicitly by the application of the
`combined function (GoF)'-1. This results, firstly, from
`the simplicity of synchronization described further
`above, which enables the implicit obtaining of the se-
`quence received from the transmission channel 43 with
`immediate knowledge of the index of the symbols in the
`sequence and, secondly, the complementarity of the
`two functions of temporal interlacing and frequential
`interlacing.
`The reconstruction of the blocks Q1is then done very
`simply in the de-interlacing module in using the bijec-
`tive character of the frequential interlacing. With the
`receiver using a differential demodulation, the data are
`restored without any problem of phase ambiguity.
`The principle of the temporal de-interlacing 45 con-
`sists in the application, to the binary elements of each
`block Q1, of the complementary delay with respect to
`the depth of the interlacing of the delay undergone at
`transmission. The knowledge of this complementary
`delay is expressed by m—f(i)—-1, and is deduced di-
`rectly from the index i of the binary element and a priori
`0 knowledge of the function f(.). No synchronization
`other than that of the multiplex itself is needed to do the
`de-interlacing.
`The diagram of FIG. 4 shows this mechanism for a
`single symbol of modulation.
`Should the channel decoder work in soft decision
`mode, the de-interlacing is actually applied not to bi-
`nary elements but to words (generally four-bit words)
`representing the estimation, by the demodulator, of the
`bits received.
`As mentioned further above, the blank symbol 24 of
`the frame header 21 can be used to identify and charac-
`terize the jamming of the transmission channel, and to
`take it into account in the restoration of the signal re-
`ceived notably within a soft decision decoding process.
`FIG. 7 shows the reception branch of the transmis-
`sion system of the invention. The hatched modules
`therein illustrate the implementation of this supplemen-
`tary function of taking the jamming into account with
`respect to the known reception chain of FIG. 1.
`The explanation of FIG. 7 first of all requires re-
`minder of the chief characteristics of the signal trans-
`mitted in general.
`The signal transmitted is formed by a sequence of
`modulation symbols forming a multiplex of N orthogo-
`nal carriers.
`~
`Let fk be the set of carrier frequencies considered
`with:
`
`l0
`
`15
`
`20
`
`25
`
`35
`
`40
`
`45
`
`50
`
`55
`
`5,191,576
`
`7
`Clearly, this example is given purely as a non-restric-
`tive illustration.
`The elements of the block Q1 are assigned to_the K
`symbols of the channel considered in the frame 1 as
`follows:
`The block Q1 is split up into K packets of 2J bits in
`ascending order of the index i and these packets are
`assigned to the K symbols of the channel considered,
`according to the following principle:
`
`No. of (am. |v‘lI.2.I— i)(q/,2). |.qI,u— 1) -
`Symbol
`0
`1
`
`-
`
`- (q1,2,rx~1), |-ql.2JK— 1)
`K — 1
`
`The frequential interlacing consists in assigning the 2J
`bits of each packet to the J carriers forming the symbol
`associated with the packet considered. These 21 bits are
`assembled in J pairs which are_ bijectively associated
`with the carriers of the symbol, according to the partic-
`ular relationship that defines the interlacing.
`An example of frequential interlacing is illustrated in
`FIG. 6 wherein:
`
`i represents the index of the elements of the sequence
`that have undergone the temporal interlacing 41 and are
`introduced in the frequential interlacing module 42;
`1 represents the index of the elements after frequen-
`tial interlacing 42;
`the column r=F(i) illustrates the implementation of
`the interlacing function by inversion of bits on the indi-
`ces of the 512 carrier frequencies forming each multi-
`plex.
`It will be noted that each elementary symbol of mod-
`ulation is formed by a selection of 448 carriers forming
`a sub-set of the set:
`
`{fl-=fo+jDf}(j=0 to 51 l)f,, designates an arbitrary
`frequency and Df the difference between each carrier.
`This sub-set is the set of carriers fj, the index j of
`which meets the condition (1):
`
`3§j§480,j:;&256
`
`(I)
`
`This choice is warranted by constraints related to the
`feasibility of certain analog functions of the receiver.
`The elimination of the central carrier overcomes the
`problem of the continuous drift of the analog digital
`circuits, and the elimination of the lateral carriers of the
`spectrum (I of the total spectrum) overcomes the edge
`effects of the cut-off filters.
`Let j be the i+ 1”’ number meeting the condition (1) in
`the list of the indices 0 to 511 classified in their recipro-
`cal ascending order. This relationship defines the func-
`tion j =F(i). The frequential interlacing is characterized
`by by the relationships:
`'
`if E(i/448) is an even value, then uj,k=q1,,-
`if E(i/448) is an odd value, then v,;k=q1,g
`with i=0 to 8 063, k=E(i/896) and j=F(R(i/448))
`and with E(p/q): integer part of p/q.
`R(p/q): remainder of the division of p by q.
`In these relationships, (u,-,1‘, vj_k) designates the couple
`of binary elements determining the phase of the carrier
`ii of the order k symbol. (Each carrier undergoes a four
`phase state modulation).
`In other words, each of the pairs of binary elements is
`formed by the digital elements source sequence in form-
`ing packets of 2M constituent elements in said sequence,
`and in associating the elements two by two in each
`packet according to a criterion of maximization of dis-
`persal of the adjacent digital elements of the source
`sequence. The pairs of digital elements are formed by
`
`fk=j},+k/T,k=0 to N——l
`
`65
`
`where T, represents the duration allocated to a modula-
`tion symbol.
`We then define an orthogonal base of elementary
`signals
`.
`\lIj,k(t) with k=0 to N—l, j= — co to + co
`IlI;,k(t)=8k(t-J'Ts)
`with 0§t§T,:gk(t)=e2"‘”/7"
`elsewhere: gk(t) = O.
`
`Page 00010
`
`Page 00010
`
`
`
`9
`Let us then take a set of complex numbers C,-J. taking
`its values in a finite alphabet, and representing the signal
`of transmitted data.
`The associated OFDM signal is written:
`
`5,191,576
`
`,
`.
`.
`N51 c
`:0) = ..+2°°
`jl-ee k-O ]'N‘l'k()
`
`In the case concerning this application, and cohe-
`sively with the preceding descriptions, the transmitted
`signals C,-,1‘ have a constant modulus. This means, in
`other words, that each of the carriers of the multiplex
`undergoes a phase modulation.
`The transmission channel can be modelized accord-
`ing to the relationship:
`'
`
`l0
`
`15
`
`l'f,;k=l1j.kCj,k+1V,-+ k
`
`where Hj_k is the complex response of the channel at
`the point (i,k) of the time-space frequency, and NM
`is a complex Gaussian noise with:
`
`20
`
`1\{,-_k=N1j_k+iNQ,;k
`and
`
`E(NIj,k)2=E(NQ‘,k)2=¢T2j.k
`
`’
`
`where E0 represents the mathematical expectation.
`It can then be shown that the implementation of a
`decoding according to a maximum likelihood criterion
`a posteriori consists in the maximization, on C,-,1‘, under
`the constraint of the code of linkage of the symbols C_,-,;.
`of the expression:
`
`Re(Y,;,:.H',~,:.c',-,:./«r2,-,:.)
`
`25
`
`30
`
`35
`
`where Re(.) represents the real part of a complex
`number.
`The essential element of this analysis relates to the
`fact that the noise power 0'2,-,1; at every point (j,k) of the
`time-frequency space coming into play in the decoding
`process. When the code linking the elements CM is a
`convolutive code and when the decoder used is a soft
`decision Viterbi decoder, the knowledge of the noise
`power generated by the channel and by the reception
`device therefore forms a major weighting parameter
`with respect to the optimization of the decoding. This
`parameter does not come into play in the particular case
`of a white noise, such that:
`02-,k=a'2, irrespectively of j and k.
`the
`the signal,
`However,
`if a jammer affects
`weighting has the effect of “erasing” the corresponding
`carriers in varying degrees in the same way as a fading
`on these same carriers. This property is a specific fea-
`ture of the COFDM system which makes it extremely
`attractive in channels highly disturbed by industrial
`stray signals or noise, the nature of which may be pulsed
`or recurrent in frequency (localized field of the time-
`frequency space).
`The implementation of this process of measurement 60
`and identification of the jamming, then of weighting of
`the resemblance coefficients consists in performing a
`spectral analysis of the noise on the empty symbol 21.
`This analysis is achieved through a discrete Fourier
`transform 71 using the digitized signal 72 obtained at the
`output of an ADC 73.
`If {fk}k=o, .
`_
`.
`_ 1v_1 designates the set of carriers used
`in the COFDM signal, it would appear to be necessary
`
`45
`
`50
`
`55
`
`65
`
`10
`to analyze