+

WO2005027367A1 - Method for receiving a cdma interference suppression signal and a corresponding receiver - Google Patents

Method for receiving a cdma interference suppression signal and a corresponding receiver Download PDF

Info

Publication number
WO2005027367A1
WO2005027367A1 PCT/FR2004/002364 FR2004002364W WO2005027367A1 WO 2005027367 A1 WO2005027367 A1 WO 2005027367A1 FR 2004002364 W FR2004002364 W FR 2004002364W WO 2005027367 A1 WO2005027367 A1 WO 2005027367A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
channel
interest
reception method
interference
Prior art date
Application number
PCT/FR2004/002364
Other languages
French (fr)
Inventor
Nicolas Ibrahim
Belkacem Mouhouche
Karim Abed Meraim
Philippe Loubaton
Original Assignee
Wavecom
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wavecom filed Critical Wavecom
Publication of WO2005027367A1 publication Critical patent/WO2005027367A1/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B1/00Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
    • H04B1/69Spread spectrum techniques
    • H04B1/707Spread spectrum techniques using direct sequence modulation
    • H04B1/7097Interference-related aspects
    • H04B1/7103Interference-related aspects the interference being multiple access interference
    • H04B1/7107Subtractive interference cancellation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B1/00Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
    • H04B1/69Spread spectrum techniques
    • H04B1/707Spread spectrum techniques using direct sequence modulation
    • H04B1/7097Interference-related aspects
    • H04B1/711Interference-related aspects the interference being multi-path interference

Definitions

  • the field of the invention is that of radiocommunications. More specifically, the invention relates to a technique for receiving a radio signal of the CDMA type (in English “Code Division Multiple
  • the invention applies in particular, but not exclusively, to the UMTS mobile radio system (in English “Universal Mobile Telecommunications System”, in French “Universal Mobile Telecommunications System”), and more particularly to the UMTS- multiple access system.
  • FDD for “Frequency Division Duplex” in downlink.
  • CDMA Code Division Multiple Access
  • the CDMA system is a broadband technique in which the radio frequency spectrum is used by a plurality of orthogonal channels. Each traffic channel, or lane, is spread by a unique code of predetermined length.
  • a user signal, or source signal can contain several traffic channels or a channel.
  • the CDMA system is therefore a “spread spectrum” technology.
  • the digital code applied, on the transmitter side, to the data symbols transmitted by a particular user is known by the appropriate receiver, so that the latter can despread the source signal which concerns him.
  • a scrambling code specific to each cell in the radio network, is applied to the signal. In this way, in reception, when the signal is despread, it is easy to identify the cell from which it comes, according to the level of correlation with the scrambling code considered.
  • the orthogonality between the traffic channels, or channels is ensured for a transmission channel without multipaths, as illustrated diagrammatically by FIG. 1.
  • the UMTS 1 signal can be symbolized by a white signal, corresponding to the sum 2 of a plurality of "colors" each corresponding to a source signal modulated by an appropriate spreading code C1 to C6.
  • These different spreading codes constitute orthogonal sequences.
  • the UMTS signal 1 does not undergo any distortion, so that the received signal 4 is still a “white” signal.
  • This received signal 4 can be processed by an optimal receiver 5, equivalent to an optical prism, which redecomposes the signal 4 into different colors C1 to C6 (namely the different spreading codes) constituting it.
  • the propagation channel contains several paths, characterized by delays and different complex gains, the orthogonality of the channels is no longer ensured, and interference between the different traffic channels appears, as illustrated by the figure. 2.
  • This interference depends on the transmission channel and can be very disabling for digital communication, inducing a strong degradation of the received signal.
  • the propagation channel 6 is multipath, the received signal is no longer a "white” signal like the UMTS signal transmitted 1, but can be seen as the sum of several "gray" signals referenced 7 to 9.
  • the decomposition obtained is no longer a decomposition into elementary colors Cl to C6: on the contrary, six new colors Cl 'to C6' are recovered each corresponding to the mixture of several starting colors Cl to C6.
  • a receiver called RAKE, which can be used in UMTS mobile terminals.
  • the RAKE 31 is a filter adapted to the propagation channel: in the example of FIG. 3, the channel of interest is the traffic channel No. 1.
  • the received and sampled signal 30 feeds the RAKE receiver 31, which comprises several parallel branches each associated with a delay ⁇ 0 to ⁇ L and a gain h 0 * to h L * .
  • Each of these branches corresponds to one of the discernible paths of the propagation channel.
  • the coefficients ⁇ 0 to ⁇ L and h 0 to h L are therefore respectively the delays and the gains of the propagation channel linked to each possible path.
  • the signals recovered from each of these branches which correspond to several equalized versions of the received signal, are supplied at the input of a block 32 grouping together the scrambling code (in English "scrambling") and the code spreading of the lane of interest n ° l.
  • This block 32 makes it possible to “scramble” and “despread” the received signal, so as to extract an estimated symbol 33 d, for the channel of interest No. 1.
  • a drawback of this technique of the prior art is that the RAKE receiver considers interference from other paths, or traffic channels, induced by multipaths, as white noise, which makes its elimination impossible.
  • the general principle of such a “RAKE + PIC” receiver is, in the case where all the spreading codes of all the channels, or traffic channels, are known, to use the decisions of the symbols of each channel , and pass them back through the estimated transmission channel, in order to subtract them from the received signal, and thus reduce the interference affecting the channel of interest.
  • the quality of the channel of interest obtained according to this technique is therefore greatly increased compared to the technique of the single RAKE receiver described above in relation to FIG. 3.
  • it is not necessary for the spreading codes of the routes other than the route of interest are known.
  • the signal received 30 then undergoes the delay operations 31 and gain of the RAKE receiver, characteristics of each of the paths, on each branch of the branches numbered from 0 to L.
  • FHT Hadamard Rapide 40 transform
  • FWT Hadamard Walsh Transform
  • the M most disruptive channels for example by according to their power.
  • the multi-path channel 46 can be regenerated, which can be subtracted 47 from the received signal 30, in order to refine the decision 42 allowing to extract the symbol b (i) 43 from the path of interest n ° i.
  • a signal b (i) more representative of the path of interest no. I since it is purified of certain interference terms.
  • a disadvantage of this technique of the prior art is that it is very costly in terms of computing resources, since it is necessary to perform a FWT type transform on each of the branches associated with one of the discernible paths of the channel of propagation.
  • this so-called “RAKE + PIC” technique does not allow sufficient cancellation of the interference, so that the curve representative of the bit error rate as a function of the signal to noise (see Figure 7) for this technique decreases too slowly.
  • a third technique for eliminating interference induced by the existence of multiple paths is an equalization technique, illustrated in FIG. 5.
  • the signal received 30 feeds an equalizer 50 of the MMSE type (in English
  • the pilot channel 51 is used by the mobile terminal to synchronize on the network and identify the cells it receives. There is one pilot channel 51 per cell.
  • the equalized signal is “unscrambled” 52 by application of the scrambling code characteristic of the cell considered, then undergoes a despreading 54, making it possible to take a decision 53 on the symbol b, of the route of interest considered, in this case route n ° l.
  • Such despreading 54 is for example a direct despreading by multiplication with the spreading sequence and accumulation over the length of this sequence.
  • This equalization technique has the advantage, compared to the RAKE and “RAKE + PIC” techniques, of being of very reduced complexity.
  • the invention particularly aims to overcome these various drawbacks of the prior art. More specifically, an objective of the invention is to provide a technique for receiving a CDMA signal making it possible to extract a signal more representative of the channel of interest than according to the techniques of the prior art. Another objective of the invention is to implement such a technique which makes it possible to increase the quality of the signal which is extracted. Another object of the invention is to provide such a technique which allows better cancellation of the interference induced by the existence of multiple paths in the propagation channel. The invention also aims to implement such a technique which has a reduced complexity, and which consumes little in terms of resources.
  • such a reception method comprises the following steps: - equalization of said received signal, delivering a single equalized signal; estimation of the interference induced by a source signal transmitted on at least one channel other than said channel of interest, from said single equalized signal; removing, from said received signal, the estimated interference induced by said source signals, for channels other than said channel of interest, to obtain a signal more representative of said channel of interest than said received signal; processing of said signal representative of said channel of interest.
  • the invention is based on a completely new and inventive approach to the reception of CDMA signals, and to the cancellation of the interference induced by the existence of multiple paths in the propagation channel.
  • the invention proposes to combine the use of an equalizer and an interference cancellation architecture.
  • the method of the invention is therefore much simpler than the methods of the prior art of the “RAKE” or “RAKE + PIC” type, since it is based in particular on the use of a single equalized signal, and not on the use of a plurality of equalized signals each corresponding to one of the discernible paths of the multipath propagation channel.
  • RAKE or “RAKE + PIC” type
  • the signal obtained according to the invention is much more representative of the path of interest than the signals obtained according to the techniques of the art. prior, thanks to a better cancellation of the interference induced by the other source signals.
  • the performances of such a method are clearly superior to the joint performances of conventional equalization techniques on the one hand, and of “RAKE + PIC” type techniques on the other hand.
  • each of said channels being obtained by multiplication of said corresponding source signal with a predetermined spreading code, the spreading code used for said at least one other channel is estimated to be blind, according to at least one predetermined criterion, during said estimation of the corresponding source signal. It is therefore not necessary, for the implementation of the invention, that the receiver knows all spread codes for all channels.
  • the receiver can obtain an estimate of the interference of the source signals of the other channels, by advantageously exploiting the orthogonality properties of the spreading codes used in CDMA.
  • said spreading code used is chosen, according to said at least one predetermined criterion, from the OVSF codes (in English "Orthogonal Variable Spreading Factor” for “orthogonal variable spreading factor”) orthogonal to the spreading code of said route of interest.
  • said at least one predetermined criterion is a correlation criterion, said spreading code used having a significant correlation with said equalized signal.
  • said steps of estimating and removing the interference induced by said source signals are carried out in series, of SIC type (in English “Sériai Interference Cancellation”).
  • said steps of estimating and removing the interference induced by said source signals are carried out in parallel, of PIC type (in English "Parallel Interference Cancellation”).
  • said interference estimation implements an inverse mathematical transformation feeding a set of decision modules in parallel, followed by a corresponding direct mathematical transformation. Unlike the "RAKE + PIC" method of the prior art, the invention therefore implements a single inverse mathematical transformation, acting on the single equalized signal after "unscrambling".
  • the invention is therefore much less consuming in terms of resources, in particular of computation, than the "RAKE + PIC" technique, for which a fast Walsh transform had to be performed on each of the L branches corresponding to the L discernible paths of the propagation channel. multipath.
  • the invention also implements a unique corresponding direct mathematical transformation. It is therefore particularly economical in computing resources.
  • said mathematical transformation is of the Hadamard Rapide Transform type, also called the Walsh Rapide Transform. It is of course also possible to envisage using any other type of mathematical transform making it possible to carry out the despreading of the equalized signal.
  • said equalization implements an algorithm of the MMSE type (in English “Minimum Mean Squared Error”).
  • said equalization implements a reduced rank filter belonging to the group comprising: the reduced rank filters of Wiener (in English "Stochastic Gradient Multistage Wiener Filter”); Conjugate Gradient Reduced Rank Filter.
  • Wiener in English "Stochastic Gradient Multistage Wiener Filter”
  • Conjugate Gradient Reduced Rank Filter the filtering methods with reduced rank have a reduced computational complexity compared to the Wiener filter with full rank.
  • said estimation of the interference induced by said source signals is iterative.
  • said decision is a flexible decision.
  • said signal is transmitted on a downlink of a UMTS system, in particular of a system of the UMTS-FDD type (for “Frequency Duplex Division ”).
  • said step of processing said signal representative of said channel of interest comprises at least some of the following substeps: an equalization substep; - a descrambling sub-step; a despreading sub-step; a decision sub-step, at the end of which symbols representative of the path of interest n ° i are recovered.
  • the invention also relates to a receiver of a CDMA signal comprising at least two channels, each corresponding to a separate source signal, including at least one channel of interest which it is desired to extract, said received signal being conveyed by a communication channel. multipath transmission.
  • such a receiver comprises the following means: - means for equalizing said received signal, delivering a single equalized signal; means for estimating the interference induced by a source signal transmitted on at least one channel other than said channel of interest, from said single equalized signal; means for removing, from said received signal, the estimated interference induced by said source signals, for channels other than said channel of interest, in order to obtain a signal more representative of said channel of interest than said received signal; means for processing said signal representative of said channel of interest.
  • FIG. 1 presents the general principle of the transmission of a CDMA signal via a perfect propagation, exposed by analogy with the domain of colorimetry;
  • FIG. 2 illustrates the principle of the transmission of the signal of FIG. 1 via a multi-path channel, again by analogy with the field of colorimetry;
  • FIG. 3, also previously commented on in relation to the prior art, describes the architecture of a RAKE type receiver;
  • FIG. 4, also presented previously, illustrates the architecture of a receiver of the prior art of the “RAKE + PIC” type;
  • - Figure 5 already described above, shows the architecture of a receiver of the prior art implementing an equalization technique;
  • FIG. 6 illustrates the general architecture of a receiver of the invention, jointly implementing equalization and interference cancellation techniques; - Figure 7 describes a variant of the receiver of Figure 6 with iterative processing; FIG. 8 presents the compared performances, in terms of bit error rate, of the techniques of the prior art and of the invention.
  • the general principle of the invention is based on the joint implementation of an equalization of the received signal and the cancellation of the interference induced by the existence of multiple paths in the propagation channel.
  • a preferred alternative embodiment is presented throughout the rest of the document in which the interference cancellation implemented has a PIC-type structure, in parallel. Those skilled in the art will easily extend this teaching to the case where interference cancellation is carried out in series, of the SIC type. Referring to FIG.
  • the received signal 30 feeds an equalizer 60, which implements the step of equalizing the signal received from the method of the invention, delivering an equalized signal unique.
  • the equalization 60 is done directly on the received signal, considered as a whole, and no longer on each of the L branches corresponding to the L discernible paths from the multipath propagation channel.
  • the equalization block 60 operates in two stages. During a first step (learning or updating), one proceeds to descrambling and then despreading of the received signal, the despreading being carried out with the spreading code of the pilot. The received signal is then shifted to unscramble and despread it again.
  • the descrambling and despreading outputs constitute a size G vector.
  • the reduced rank algorithm uses this vector as an input signal and the pilot symbol (known) as the desired output to update the equalizer.
  • the second step (direct equalization) consists in applying the updated equalizer to the input signal to give an equalized signal which will serve as an input to the block referenced 61.
  • the single equalized signal then undergoes "descrambling" 61, by multiplication term to term of the signal by the scrambling sequence, then feeds a module 62 in which it undergoes a Fast Reverse Hadamard Transform (or IFHT).
  • IFHT Fast Reverse Hadamard Transform
  • This module 62 implements a multiplication by the Hadamard matrix containing +1 or -1, which can be made rapid using the butterfly techniques used by the FFT (Fast Fourier Transform). . It will be noted that, according to this architecture, a single inverse mathematical transformation 62 must be implemented, on the equalized and unscrambled signal, which constitutes a significant saving in terms of computing resources compared to the “RAKE + PIC” technique according to which one FWT was to be performed on each of the L branches of the RAKE (see Figure 4). In the example of FIG. 6, we consider that the channel of interest that we are trying to extract is channel # 1.
  • the signals from IFHT 62 feed Kl decision modules in parallel, referenced 63 2 to 63 ⁇ , each corresponding to one of the channels 2 to K other than the channel of interest No. 1.
  • Decisions 63 2 to 63 ⁇ can be hard decisions or flexible decisions.
  • the sign of the result obtained by weighting the output of the IFHT operation 62 by the length of the spreading sequence considered can be adopted as a flexible decision criterion.
  • the estimate 66 of the interference induced by the channels other than the channel of interest is then subtracted 67 from the received signal 30, according to a simple mathematical subtraction, so as to recover only the signal representative of the channel of interest n ° l, which constitutes the step of removing the interference from the reception method of the invention. Thanks to IFHT 62, we therefore generate the Kl interference estimates of
  • Kl source signals corresponding to the most powerful Kl channels are subtracted (67) from the received signal 30 to obtain a better estimate of the transmitted signal.
  • the interference estimate can also relate only to M ⁇ K signals.
  • the signal obtained after subtraction 67 then undergoes the processing step of the reception method of the invention, during which it is equalized 68, according to a method similar to that implemented by the equalizer 60, then feeds the block 69 grouping the spreading and scrambling codes of the channel of interest.
  • the block referenced 69 implements a multiplication by the joint scrambling and spreading sequence of the signal of interest, followed by a summation over the length of the spreading sequence of the signal of interest.
  • a firm decision 70 is taken, on the sign of the quantity resulting from the block referenced 69, which makes it possible to extract the symbol b, transmitted on the channel of interest no.
  • FIG. 7 presents a variant of the receiver of FIG.
  • FIG. 6 in the case where an iteration of the steps of estimation and removal of the interference induced by the source signals is implemented.
  • the same elements of the receiver are designated by the same reference numerals in FIGS. 6 and 7.
  • the processing operations applied to the signal already described previously in connection with FIG. 6 are not described in more detail.
  • the firm decision 70 making it possible to extract the symbol b, emitted on the channel of interest no. 1 is taken at the output of the IFHT module 62.
  • the decisions 63 2 to 63 ⁇ are flexible decisions, as mentioned previously, and are repeated at each successive iteration of the processing.
  • a decision 70 on the path of interest No. 1 is taken at each iteration of the IFHT step 62.
  • FIG. 8 illustrates the compared performances of the different methods described above in relation to FIGS. 3 to 6, namely: the “RAKE” receiver of the prior art; the “RAKE + PIC” receiver of the prior art; the prior art equalization technique; the joint technique of equalization and interference cancellation of the invention.
  • the signal to noise ratio or SNR of the channel of interest varies between 0 dB and 15 dB, while the signal to noise ratio of the other 14 channels is fixed at 10 dB.
  • the simulation chain does not take into account channel coding.
  • the channels are modulated in
  • the values of the entrenched channels are the decisions 63 2 to 63 k made at the output of the inverse fast Hadamard transform
  • FIG. 8 shows that the technique of the invention makes it possible to obtain a reduction in the signal to noise ratio of 4 dB for a rate d bit error rate (BER) of 10 "3 , compared to the conventional equalization technique of the prior art.
  • the gain obtained, according to the invention, compared to the technique of the art so-called "RAKE" is even greater.
  • the performances of FIG. 8 were obtained for a variant of the invention in which a single step of estimation of the interfering pathways and removal of this interference was implemented, and in which the decisions taken by the modules 63 2 to 63 ⁇ are hard decisions. These performances can be further increased by implementing an iteration of the various steps of the method of the invention, as illustrated in FIG. 7, and by taking flexible decisions es.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
  • Noise Elimination (AREA)

Abstract

The invention relates to a method for receiving a CDMA signal comprising at least two channels each of which corresponds to a distinct source signal whose at least one channel is of interest for extraction, said received signal being transmitted by a multibeam transmission channel. The inventive reception method consists in equalising (60) the received signal by transmitting a single equalised signal, estimating (61-66) an interference induced by the source signal transmitted through al least one channel other than said channel of interest in order to obtain a more representative signal of said channel of interest than the received signal, in processing said representative signal for the channel of interest, each of said channels being obtained by multiplying said corresponding source signal with a predetermined asymmetry code, the asymmetry code used for said at least one other channel is estimated at random according to at lest one predetermined criterion during said estimation of the corresponding source signal.

Description

Procédé de réception d'un signal CDMA à annulation d'interférence et récepteur correspondant. Le domaine de l'invention est celui des radiocommunications. Plus précisément, l'invention concerne une technique de réception d'un signal de radiocommunication de type CDMA (en anglais « Code Division Multiple Method for receiving a CDMA signal with interference cancellation and corresponding receiver. The field of the invention is that of radiocommunications. More specifically, the invention relates to a technique for receiving a radio signal of the CDMA type (in English “Code Division Multiple
Access », en français AMDC pour « Accès Multiple par Différence de Code ») transmis via un canal de propagation à trajets multiples. L'invention s'applique notamment, mais non exclusivement, au système de radiotéléphonie mobile UMTS (en anglais « Universal Mobile Télécommunications System », en français « Système de Télécommunication Mobile Universel »), et plus particulièrement au système d'accès multiple UMTS- FDD (pour « Frequency Division Duplex ») en liaison descendante. On rappelle que le système CDMA est une technique large bande dans laquelle le spectre radiofréquence est utilisé par une pluralité de canaux orthogonaux. Chaque canal de trafic, ou voie, est étalé par un code unique de longueur prédéterminée. Un signal d'usager, ou signal source, peut contenir plusieurs canaux de trafic ou une voie. Le système CDMA est donc une technologie à « étalement de spectre ». Le code numérique appliqué, côté émetteur, aux symboles de données émis par un usager particulier est connu par le récepteur approprié, de façon que ce dernier puisse désétaler le signal source qui le concerne. En outre, un code de brouillage, propre à chacune des cellules du réseau de radiocommunication, est appliqué au signal. De cette façon, en réception, lorsqu'on désétale le signal, on peut facilement identifier la cellule dont il provient, en fonction du niveau de corrélation avec le code de brouillage considéré. Dans le système d'accès multiple UMTS en liaison descendante, l'orthogonalité entre les canaux de trafic, ou voies, est assurée pour un canal de transmission sans trajets multiples, ainsi qu'illustré schématiquement par la figure 1. Par analogie avec le domaine de la colorimétrie, le signal UMTS 1 peut être symbolisé par un signal blanc, correspondant à la somme 2 d'une pluralité de « couleurs » correspondant chacune à un signal source modulé par un code d'étalement Cl à C6 approprié. Ces différents codes d'étalement constituent des séquences orthogonales. Lors de la transmission du signal UMTS 1 par un canal de transmission parfait 3, le signal UMTS 1 ne subit aucune déformation, de sorte que le signal reçu 4 est encore un signal « blanc ». Ce signal reçu 4 peut être traité par un récepteur optimal 5, équivalent à un prisme optique, qui redécompose le signal 4 en différentes couleurs Cl à C6 (à savoir les différents codes d'étalement) le constituant. Cependant, lorsque le canal de propagation contient plusieurs trajets, caractérisés par des retards et des gains complexes différents, l'orthogonalité des voies n'est plus assurée, et une interférence entre les différents canaux de trafic apparaît, ainsi qu'illustré par la figure 2. Cette interférence dépend du canal de transmission et peut être très handicapante pour la communication numérique, en induisant une forte dégradation du signal reçu. Ainsi, lorsque le canal de propagation 6 est à trajets multiples, le signal reçu n'est plus un signal « blanc » comme le signal UMTS émis 1, mais peut être vu comme la somme de plusieurs signaux « gris » référencés 7 à 9. Lorsque ces signaux 7, 8 et 9 sont décomposés par le récepteur équivalent à un prisme optique 5, la décomposition obtenue n'est plus une décomposition en couleurs élémentaires Cl à C6 : on récupère au contraire six nouvelles couleurs Cl' à C6' correspondant chacune au mélange de plusieurs couleurs de départ Cl à C6. Plusieurs techniques ont d'ores et déjà été envisagées pour tenter de résoudre ce problème, et obtenir une réception de bonne qualité, en dépit de l'apparition d'interférences liées à l'existence de trajets multiples. On a tout d'abord conçu un récepteur, appelé RAKE, susceptible d'être utilisé dans les terminaux mobiles UMTS. Ainsi qu'illustré sur la figure 3, le RAKE 31 est un filtre adapté au canal de propagation : dans l'exemple de la figure 3, la voie d'intérêt est le canal de trafic n° 1. Le signal reçu et échantillonné 30 alimente le récepteur RAKE 31 , qui comprend plusieurs branches en parallèles associées chacune à un retard τ0 à τL et à un gain h0 * à hL *. Chacune de ces branches correspond à l'un des trajets discernables du canal de propagation. Les coefficients τ0 à τL et h0 à hL sont donc respectivement les retards et les gains du canal de propagation liés à chaque trajet possible. En sortie du récepteur RAKE 31, les signaux récupérés sur chacune de ces branches, qui correspondent à plusieurs versions égalisées du signal reçu, sont fournis en entrée d'un bloc 32 regroupant le code de brouillage (en anglais « scrambling ») et le code d'étalement (en anglais « spreading ») de la voie d'intérêt n°l. Ce bloc 32 permet de « dé-brouiller » et « désétaler » le signal reçu, de façon à extraire un symbole estimé 33 d, pour la voie d'intérêt n° 1. Un inconvénient de cette technique de l'art antérieur est que le récepteur RAKE considère les interférences des autres voies, ou canaux de trafic, induites par les trajets multiples, comme un bruit blanc, ce qui rend leur élimination impossible. Pour tenter de pallier cet inconvénient du récepteur RAKE, on a envisagé de le compléter au moyen d'un bloc d'annulation d'interférence en parallèle de type PIC (en anglais « Parallel Interférence Cancellation »), ainsi qu'illustré par la figure 4. Le principe général d'un tel récepteur « RAKE +PIC » est, dans le cas où tous les codes d'étalement de toutes les voies, ou canaux de trafic, sont connus, d'utiliser les décisions des symboles de chaque voie, et de les repasser dans le canal de transmission estimé, afin de les retrancher du signal reçu, et faire ainsi diminuer l'interférence affectant la voie d'intérêt. La qualité de la voie d'intérêt obtenue selon cette technique est donc fortement accrue par rapport à la technique du seul récepteur RAKE décrite ci- dessus en relation avec la figure 3. Il n'est cependant pas nécessaire que les codes d'étalement des voies autres que la voie d'intérêt soient connus. En effet, grâce à la méthodologie arborescente de sélection des codes d'étalement prévue dans le système UMTS, on peut considérer, en réception, que tous les codes d'étalement ont la même longueur que le code d'étalement de la voie d'intérêt, car, lorsqu'un code est utilisé, tous les codes qui le suivent dans l'arbre des codes ne le sont pas. Le désétalement peut donc se faire très facilement par une Transformée d'Hadamard Rapide. En d'autres termes, on estime l'interférence induite par tous les codes de la même longueur, et on retranche leur influence du signal reçu. Lorsqu'un code n'est pas utilisé, les symboles estimés à sa sortie sont faibles, et ont donc une influence négligeable sur le résultat final. Plus précisément, en relation avec la figure 4, on identifie tout d'abord tous les trajets discernables du canal de propagation, en l'occurrence L+l trajets numérotés de 0 à L. Le signal reçu 30 subit alors les opérations 31 de retard et de gain du récepteur RAKE, caractéristiques de chacun des trajets, sur cliacune des branches numérotées de 0 à L. On réalise ensuite une transformée d'Hadamard Rapide 40 (FHT pour « Fast Hadamard Transform », encore appelée « Fast Walsh Transform » ou FWT) sur chacune des branches 0 à L, de façon à réaliser le désétalement du signal. Les différentes versions égalisées du signal, disponibles sur cliacune des branches 0 à L du récepteur, sont combinées dans le bloc MRC 41 (en anglais « Maximum Ratio Combiner »), par addition, avec un retard adéquat. Si l'on considère un système comprenant K codes d'étalement distincts, on peut choisir d'éliminer l'interférence induite par M canaux de trafic parmi ces K. On sélectionne alors (44) les M voies les plus perturbatrices, par exemple en fonction de leur puissance. Après utilisation du code d'étalement à CM associé à chacune de ces voies, et combinaison 45, on peut régénérer 46 le canal multi- trajet, que l'on peut soustraire 47 au signal reçu 30, afin d'affiner la décision 42 permettant d'extraire le symbole b(i) 43 de la voie d'intérêt n°i. On obtient ainsi, par rapport à la technique RAKE décrite précédemment en relation avec la figure 3, un signal b(i) plus représentatif de la voie d'intérêt n°i, puisque épuré de certains termes d'interférences. Un inconvénient de cette technique de l'art antérieur est qu'elle est très coûteuse en termes de ressources de calcul, puisqu'il est nécessaire de réaliser une transformée de type FWT sur chacune des branches associées à l'un des trajets discernables du canal de propagation. En outre, bien que plus performante que la technique du RAKE, cette technique dite « RAKE+PIC » ne permet pas une annulation suffisante de l'interférence, de sorte que la courbe représentative du taux d'erreur binaire en fonction du rapport signal à bruit (voir figure 7) pour cette technique décroît trop lentement. Une troisième technique d'élimination de l'interférence induite par l'existence de trajets multiples est une technique d'égalisation, illustrée par la figure 5. Le signal reçu 30 alimente un égaliseur 50 de type MMSE (en anglaisAccess ”, in French AMDC for“ Multiple Access by Difference Code ”) transmitted via a multipath propagation channel. The invention applies in particular, but not exclusively, to the UMTS mobile radio system (in English “Universal Mobile Telecommunications System”, in French “Universal Mobile Telecommunications System”), and more particularly to the UMTS- multiple access system. FDD (for “Frequency Division Duplex”) in downlink. It will be recalled that the CDMA system is a broadband technique in which the radio frequency spectrum is used by a plurality of orthogonal channels. Each traffic channel, or lane, is spread by a unique code of predetermined length. A user signal, or source signal, can contain several traffic channels or a channel. The CDMA system is therefore a “spread spectrum” technology. The digital code applied, on the transmitter side, to the data symbols transmitted by a particular user is known by the appropriate receiver, so that the latter can despread the source signal which concerns him. In addition, a scrambling code, specific to each cell in the radio network, is applied to the signal. In this way, in reception, when the signal is despread, it is easy to identify the cell from which it comes, according to the level of correlation with the scrambling code considered. In the UMTS downlink multiple access system, the orthogonality between the traffic channels, or channels, is ensured for a transmission channel without multipaths, as illustrated diagrammatically by FIG. 1. By analogy with the domain of the colorimetry, the UMTS 1 signal can be symbolized by a white signal, corresponding to the sum 2 of a plurality of "colors" each corresponding to a source signal modulated by an appropriate spreading code C1 to C6. These different spreading codes constitute orthogonal sequences. During the transmission of the UMTS signal 1 by a perfect transmission channel 3, the UMTS signal 1 does not undergo any distortion, so that the received signal 4 is still a “white” signal. This received signal 4 can be processed by an optimal receiver 5, equivalent to an optical prism, which redecomposes the signal 4 into different colors C1 to C6 (namely the different spreading codes) constituting it. However, when the propagation channel contains several paths, characterized by delays and different complex gains, the orthogonality of the channels is no longer ensured, and interference between the different traffic channels appears, as illustrated by the figure. 2. This interference depends on the transmission channel and can be very disabling for digital communication, inducing a strong degradation of the received signal. Thus, when the propagation channel 6 is multipath, the received signal is no longer a "white" signal like the UMTS signal transmitted 1, but can be seen as the sum of several "gray" signals referenced 7 to 9. When these signals 7, 8 and 9 are decomposed by the receiver equivalent to an optical prism 5, the decomposition obtained is no longer a decomposition into elementary colors Cl to C6: on the contrary, six new colors Cl 'to C6' are recovered each corresponding to the mixture of several starting colors Cl to C6. Several techniques have already been envisaged in an attempt to resolve this problem, and to obtain good quality reception, despite the appearance of interference linked to the existence of multiple paths. We first designed a receiver, called RAKE, which can be used in UMTS mobile terminals. As illustrated in FIG. 3, the RAKE 31 is a filter adapted to the propagation channel: in the example of FIG. 3, the channel of interest is the traffic channel No. 1. The received and sampled signal 30 feeds the RAKE receiver 31, which comprises several parallel branches each associated with a delay τ 0 to τ L and a gain h 0 * to h L * . Each of these branches corresponds to one of the discernible paths of the propagation channel. The coefficients τ 0 to τ L and h 0 to h L are therefore respectively the delays and the gains of the propagation channel linked to each possible path. At the output of the RAKE receiver 31, the signals recovered from each of these branches, which correspond to several equalized versions of the received signal, are supplied at the input of a block 32 grouping together the scrambling code (in English "scrambling") and the code spreading of the lane of interest n ° l. This block 32 makes it possible to “scramble” and “despread” the received signal, so as to extract an estimated symbol 33 d, for the channel of interest No. 1. A drawback of this technique of the prior art is that the RAKE receiver considers interference from other paths, or traffic channels, induced by multipaths, as white noise, which makes its elimination impossible. In an attempt to overcome this drawback of the RAKE receiver, it has been envisaged to supplement it by means of a block of interference cancellation in parallel of the PIC type (in English "Parallel Interference Cancellation"), as illustrated by the figure. 4. The general principle of such a “RAKE + PIC” receiver is, in the case where all the spreading codes of all the channels, or traffic channels, are known, to use the decisions of the symbols of each channel , and pass them back through the estimated transmission channel, in order to subtract them from the received signal, and thus reduce the interference affecting the channel of interest. The quality of the channel of interest obtained according to this technique is therefore greatly increased compared to the technique of the single RAKE receiver described above in relation to FIG. 3. However, it is not necessary for the spreading codes of the routes other than the route of interest are known. Indeed, thanks to the methodology tree structure for selection of spreading codes provided in the UMTS system, it can be considered, on reception, that all the spreading codes have the same length as the spreading code of the channel of interest, because, when a code is used, all the codes that follow it in the code tree are not. Despreading can therefore be done very easily by a Rapid Hadamard Transformation. In other words, the interference induced by all the codes of the same length is estimated, and their influence is subtracted from the received signal. When a code is not used, the symbols estimated at its output are weak, and therefore have a negligible influence on the final result. More precisely, in relation to FIG. 4, first of all all the discernible paths of the propagation channel are identified, in this case L + l paths numbered from 0 to L. The signal received 30 then undergoes the delay operations 31 and gain of the RAKE receiver, characteristics of each of the paths, on each branch of the branches numbered from 0 to L. We then perform a Hadamard Rapide 40 transform (FHT for “Fast Hadamard Transform”, also called “Fast Walsh Transform” or FWT) on each of the branches 0 to L, so as to carry out the despreading of the signal. The different equalized versions of the signal, available on each of the branches 0 to L of the receiver, are combined in the block MRC 41 (in English “Maximum Ratio Combiner”), by addition, with an adequate delay. If we consider a system comprising K distinct spreading codes, we can choose to eliminate the interference induced by M traffic channels among these K. We then select (44) the M most disruptive channels, for example by according to their power. After using the spreading code at C M associated with each of these channels, and combination 45, the multi-path channel 46 can be regenerated, which can be subtracted 47 from the received signal 30, in order to refine the decision 42 allowing to extract the symbol b (i) 43 from the path of interest n ° i. One thus obtains, compared to the RAKE technique described previously in relation to FIG. 3, a signal b (i) more representative of the path of interest no. I, since it is purified of certain interference terms. A disadvantage of this technique of the prior art is that it is very costly in terms of computing resources, since it is necessary to perform a FWT type transform on each of the branches associated with one of the discernible paths of the channel of propagation. In addition, although more efficient than the RAKE technique, this so-called “RAKE + PIC” technique does not allow sufficient cancellation of the interference, so that the curve representative of the bit error rate as a function of the signal to noise (see Figure 7) for this technique decreases too slowly. A third technique for eliminating interference induced by the existence of multiple paths is an equalization technique, illustrated in FIG. 5. The signal received 30 feeds an equalizer 50 of the MMSE type (in English
« Minimum Mean Squared Error »), dont le calcul peut être simplifié grâce aux méthodes de filtrage à rang réduit, de type A-CGRRF (en anglais « Adaptive“Minimum Mean Squared Error”), the calculation of which can be simplified using reduced rank filtering methods, of the A-CGRRF type (in English “Adaptive
Conjuguate Gradient Reduced Rank Filter ») ou SG-MSWF (en anglaisConjuguate Gradient Reduced Rank Filter ”) or SG-MSWF (in English
« Stochastic Gradient MultiStage Wiener Filter »). Ces méthodes, qui, comme pour un égaliseur de Wiener à rang complet, n'ont besoin que du signal reçu 30 et du canal pilote 51 pour estimer l'égaliseur MMSE sont notamment décrites par B."Stochastic Gradient MultiStage Wiener Filter"). These methods, which, like for a full-range Wiener equalizer, only need the received signal 30 and the pilot channel 51 to estimate the MMSE equalizer are described in particular by B.
Mouhouche, N. Ibrahim, K. Abed Meraim et P. Loubaton dans « Reduced-RankMouhouche, N. Ibrahim, K. Abed Meraim and P. Loubaton in "Reduced-Rank
Adaptive Chip-Level MMSE Equalization for the Forward Link of Long CodeAdaptive Chip-Level MMSE Equalization for the Forward Link of Long Code
CDMA Systems », Proceedings of the International Symposium on Signal Processing Applications (ISSPA), Paris, 1-4 juillet 2003. On rappelle que, dans un système CDMA, le canal pilote 51 est utili se par le terminal mobile pour se synchroniser sur le réseau et identifier les cellules qu'il reçoit. Il y a un canal pilote 51 par cellule. En sortie de l'égaliseur 50, le signal égalisé est « débrouillé » 52 par application du code de brouillage caractéristique de la cellule considérée, puis subit un désétalement 54, permettant de prendre une décision 53 sur le symbole b, de la voie d'intérêt considérée, en l'occurrence la voie n°l. Un tel désétalement 54 est par exemple un désétalement direct par multiplication avec la séquence d'étalement et accumulation sur la longueur de cette séquence. Cette technique d'égalisation présente l'avantage, par rapport aux techniques de RAKE et de « RAKE+PIC », d'être de complexité très réduite.CDMA Systems ”, Proceedings of the International Symposium on Signal Processing Applications (ISSPA), Paris, 1-4 July 2003. It is recalled that, in a CDMA system, the pilot channel 51 is used by the mobile terminal to synchronize on the network and identify the cells it receives. There is one pilot channel 51 per cell. At the output of the equalizer 50, the equalized signal is “unscrambled” 52 by application of the scrambling code characteristic of the cell considered, then undergoes a despreading 54, making it possible to take a decision 53 on the symbol b, of the route of interest considered, in this case route n ° l. Such despreading 54 is for example a direct despreading by multiplication with the spreading sequence and accumulation over the length of this sequence. This equalization technique has the advantage, compared to the RAKE and “RAKE + PIC” techniques, of being of very reduced complexity.
Cependant, comme illustré par la figure 7, les résultats obtenus par égalisation traduisent des performances moindres que celles de la technique diteHowever, as illustrated by FIG. 7, the results obtained by equalization reflect lower performances than those of the so-called technique.
« RAKE+PIC ». L'invention a notamment pour objectif de pallier ces différents inconvénients de l'art antérieur. Plus précisément, un objectif de l'invention est de fournir une technique de réception d'un signal CDMA permettant d'extraire un signal plus représentatif de la voie d'intérêt que selon les techniques de l'art antérieur. Un autre objectif de l'invention est de mettre en œuvre une telle tecb-nique qui permette d'accroître la qualité du signal que l'on extrait. L'invention a encore pour objectif de fournir une telle technique qui permette une meilleure annulation de l'interférence induite par l'existence de trajets multiples dans le canal de propagation. L'invention a aussi pour objectif de mettre en œuvre une telle teclinique qui présente une complexité réduite, et qui soit peu consommatrice en termes de ressources. Ces objectifs, ainsi que d'autres qui apparaîtront par la suite, sont atteints à l'aide d'un procédé de réception d'un signal CDMA comprenant au moins deux voies, correspondant chacune à un signal source distinct, dont au moins une voie d'intérêt que l'on veut extraire, ledit signal reçu étant véhiculé par un canal de transmission à trajets multiples. Selon l'invention, un tel procédé de réception comprend les étapes suivantes : - égalisation dudit signal reçu, délivrant un signal égalisé unique ; estimation de l'interférence induite par un signal source émis sur au moins une voie autre que ladite voie d'intérêt, à partir dudit signal égalisé unique ; retrait, dudit signal reçu, de l'interférence estimée induite par lesdits signaux source, pour les voies autres que ladite voie d'intérêt, pour obtenir un signal plus représentatif de ladite voie d'intérêt que ledit signal reçu ; traitement dudit signal représentatif de ladite voie d'intérêt. Ainsi, l'invention repose sur une approche tout à fait nouvelle et inventive de la réception de signaux CDMA, et de l'annulation de l'interférence induite par l'existence de trajets multiples dans le canal de propagation. En effet, l'invention propose de combiner l'utilisation d'un égaliseur et d'une architecture d'annulation d'interférence. Le procédé de l'invention est donc beaucoup plus simple que les méthodes de l'art antérieur de type « RAKE » ou « RAKE + PIC », puisqu'il repose notamment sur l'utilisation d'un signal égalisé unique, et non sur l'utilisation d'une pluralité de signaux égalisés correspondant chacun à l'un des trajets discernables du canal de propagation multi-trajets. En outre, comme cela apparaîtra plus clairement dans la suite de ce document, en relation avec la figure 8, le signal obtenu selon l'invention est beaucoup plus représentatif de la voie d'intérêt que les signaux obtenus selon les techniques de l'art antérieur, grâce à une meilleure annulation de l'interférence induite par les autres signaux sources. On notera que les performances d'un tel procédé sont nettement supérieures aux performances conjointes des techniques d'égalisation classique d'une part, et des techniques de type « RAKE + PIC » d'autre part. Avantageusement, chacune desdites voies étant obtenue par multiplication dudit signal source correspondant avec un code d'étalement prédéterminé, le code d'étalement utilisé pour ladite au moins une autre voie est estimé en aveugle, selon au moins un critère prédéterminé, lors de ladite estimation du signal source correspondant. II n'est donc pas nécessaire, pour la mise en œuvre de l'invention que le récepteur connaisse tous les codes d'étalement de toutes les voies. Connaissant le seul code d'étalement de la voie d'intérêt, le récepteur peut obtenir une estimation de l'interférence des signaux sources des autres voies, en exploitant avantageusement les propriétés d'orthogonalité des codes d'étalement utilisés en CDMA. Ainsi, ledit code d'étalement utilisé est choisi, selon ledit au moins un critère prédéterminé, parmi les codes OVSF (en anglais « Orthogonal Variable Spreading Factor » pour « facteur d'étalement variable orthogonal ») orthogonaux au code d'étalement de ladite voie d'intérêt. Préférentiellement, ledit au moins un critère prédéterminé est un critère de corrélation, ledit code d'étalement utilisé présentant une corrélation importante avec ledit signal égalisé. Selon une première variante avantageuse de l'invention, lesdites étapes d'estimation et de retrait de l'interférence induite par lesdits signaux sources sont réalisées en série, de type SIC (en anglais « Sériai Interférence Cancellation »). Selon une deuxième variante avantageuse de l'invention, lesdites étapes d'estimation et de retrait de l'interférence induite par lesdits signaux sources sont réalisées en parallèle, de type PIC (en anglais « Parallel Interférence Cancellation »). De façon avantageuse, ladite estimation de l'interférence met en œuvre une transformation mathématique inverse alimentant un ensemble de modules de décision en parallèle, suivie d'une transformation mathématique directe correspondante. Contrairement à la méthode « RAKE + PIC » de l'art antérieur, l'invention met donc en œuvre une unique transformation mathématique inverse, agissant sur le signal égalisé unique après « débrouillage ». L'invention est donc beaucoup moins consommatrice en termes de ressources, notamment de calcul, que la technique « RAKE + PIC », pour laquelle une transformée de Walsh rapide devait être réalisée sur chacune des L branches correspondant aux L trajets discernables du canal de propagation multi-trajets. L'invention met également en œuvre une unique transformation mathématique directe correspondante. Elle est donc particulièrement économe en ressources de calcul. Préférentiellement, ladite transformation mathématique est de type Transformée d'Hadamard Rapide, encore appelée Transformée de Walsh Rapide. On peut bien sûr également envisager d'utiliser tout autre type de transformée mathématique permettant de réaliser le désétalement du signal égalisé. Avantageusement, ladite égalisation met en œuvre un algorithme de type MMSE (en anglais « Minimum Mean Squared Error »). De façon préférentielle, ladite égalisation met en œuvre un filtre à rang réduit appartenant au groupe comprenant : les filtres à rang réduit de Wiener (en anglais « Stochastic Gradient Multistage Wiener Filter ») ; les filtres à rang réduit à gradient conjugué (en anglais « Conjugate Gradient Reduced Rank Filter »). Le calcul de l'égaliseur se trouve ainsi fortement simplifié. En effet, comme pour le calcul d'un égaliseur de Wiener à rang complet, seuls le signal reçu et le canal pilote doivent être connus. En outre, les méthodes de Filtrage à rang réduit présentent une complexité de calcul réduite par rapport au filtre de Wiener à rang complet. Selon une variante avantageuse de l'invention, ladite estimation de l'interférence induite par lesdits signaux sources est itérative. On peut ainsi améliorer la qualité du signal représentatif de la voie d'intérêt, en procédant à plusieurs itérations des étapes d'estimation et de retrait de l'interférence, et de traitement du signal représentatif de la voie d'intérêt. Le signal ainsi obtenu est alors plus représentatif de la voie d'intérêt. Selon une autre variante avantageuse de l'invention, ladite décision est une décision souple. Avantageusement, ledit signal est transmis sur une voie descendante d'un système UMTS, notamment d'un système de type UMTS-FDD (pour « Frequency Division Duplex »). De façon préférentielle, ladite étape de traitement dudit signal représentatif de ladite voie d'intérêt comprend au moins certaines des sous-étapes suivantes : une sous-étape d'égalisation ; - une sous-étape de débrouillage ; une sous-étape de désétalement ; une sous-étape de décision, à l'issue desquelles on récupère des symboles représentatifs de la voie d'intérêt n°i. L'invention concerne aussi un récepteur d'un signal CDMA comprenant au moins deux voies, correspondant chacune à un signal source distinct, dont au moins une voie d'intérêt que l'on veut extraire, ledit signal reçu étant véhiculé par un canal de transmission à trajets multiples. Selon l'invention, un tel récepteur comprend les moyens suivants : - des moyens d'égalisation dudit signal reçu, délivrant un signal égalisé unique ; des moyens d'estimation de l'interférence induite par un signal source émis sur au moins une voie autre que ladite voie d'intérêt, à partir dudit signal égalisé unique ; - des moyens de retrait, dudit signal reçu, de l'interférence estimée induite par lesdits signaux source, pour les voies autres que ladite voie d'intérêt, pour obtenir un signal plus représentatif de ladite voie d'intérêt que ledit signal reçu ; des moyens de traitement dudit signal représentatif de ladite voie d'intérêt. D'autres caractéristiques et avantages de l'invention apparaîtront plus clairement à la lecture de la description suivante d'un mode de réalisation préférentiel, donné à titre de simple exemple illustratif et non limitatif, et des dessins annexés, parmi lesquels : la figure 1, déjà commentée en relation avec l'art antérieur, présente le principe général de la transmission d'un signal CDMA via un canal de propagation parfait, exposé par analogie avec le domaine de la colorimétrie ; la figure 2 illustre le principe de la transmission du signal de la figure 1 via un canal multi-trajets, à nouveau par analogie avec le domaine de la colorimétrie ; la figure 3, également commentée précédemment en relation avec l'art antérieur, décrit l'architecture d'un récepteur de type RAKE ; la figure 4, également présentée précédemment, illustre l'architecture d'un récepteur de l'art antérieur de type « RAKE + PIC » ; - la figure 5, déjà décrite ci-dessus, présente l'architecture d'un récepteur de l'art antérieur mettant en œuvre une technique d'égalisation ; la figure 6 illustre l'architecture général d'un récepteur de l'invention, mettant conjointement en œuvre des techniques d'égalisation et d'annulation d'interférence ; - la figure 7 décrit une variante du récepteur de la figure 6 à traitement itératif ; la figure 8 présente les performances comparées, en termes de taux d'erreur binaires, des techniques de l'art antérieur et de l'invention. Le principe général de l'invention repose sur la mise en œuvre conjointe d'une égalisation du signal reçu et de l'annulation de l'interférence induite par l'existence de trajets multiples dans le canal de propagation. Par souci de simplification, on présente, dans toute la suite du document, une variante de réalisation préférentielle dans laquelle l'annulation d'interférence mise en œuvre présente une structure de type PIC, en parallèle. L'Homme du Métier étendra sans difficulté cet enseignement au cas où l'annulation d'interférence est réalisée en série, de type SIC. On présente, en relation avec la figure 6, l'architecture d'un récepteur de l'invention. Le signal reçu 30 alimente un égaliseur 60, qui met en œuvre l'étape d'égalisation du signal reçu du procédé de l'invention, délivrant un signal égalisé unique. Ainsi, contrairement aux techniques de l'art antérieur de type « RAKE » ou « RAKE+PIC », l'égalisation 60 se fait directement sur le signal reçu, considéré dans sa globalité, et non plus sur chacune des L branches correspondant aux L trajets discernables du canal de propagation multi-trajets. Le bloc d'égalisation 60 fonctionne en deux étapes. Au cours d'une première étape (apprentissage ou mise à jour), on procède au débrouillage puis au désétalement du signal reçu, le désétalement étant réalisé avec le code d'étalement du pilote. On décale ensuite le signal reçu pour le débrouiller et le désétaler de nouveau. Cette opération est répétée G fois, où G est la taille de l'égaliseur. Les sorties du débrouillage et désétalement constituent un vecteur de taille G. L'algorithme à rang réduit utilise ce vecteur comme signal d'entrée et le symbole pilote (connu) comme sortie désirée pour mettre à jour l'égaliseur. La deuxième étape (égalisation directe) consiste à appliquer l'égaliseur mis à jour au signal d'entrée pour donner un signal égalisé qui servira d'entrée au bloc référencé 61. Le signal égalisé unique subit ensuite un « débrouillage » 61 , par multiplication terme à terme du signal par la séquence de brouillage, puis alimente un module 62 dans lequel il subit une Transformée d'Hadamard Rapide Inverse (ou IFHT). Ce module 62 met en œuvre une multiplication par la matrice d'Hadamard contenant des +1 ou des -1, qui peut être rendue rapide en utilisant les techniques de papillon utilisées par la FFT (Fast Fourier Transform pour « Transformée de Fourier rapide »). On notera que, selon cette architecture, une seule transformation mathématique inverse 62 doit être mise en œuvre, sur le signal égalisé et débrouillé, ce qui constitue une forte économie en termes de ressources de calcul par rapport à la technique « RAKE + PIC » selon laquel le une FWT devait être réalisée sur chacune des L branches du RAKE (voir figure 4). Dans l'exemple de la figure 6, on considère que la voie d'intérêt que l'on cherche à extraire est la voie n°l. Les signaux issus de la IFHT 62 alimentent K-l modules de décision en parallèle, référencés 632 à 63κ, correspondant chacun à l'une des voies 2 à K autres que la voie d'intérêt n°l. Les décisions 632 à 63κ peuvent être des décisions dures ou des décisions souples. Par exemple, on peut adopter comme critère de décision souple le signe du résultat obtenu en pondérant la sortie de l'opération IFHT 62 par la longueur de la séquence d'étalement considérée. Après décisions, l'ensemble des signaux subissent une transformée d'Hadamard directe, ou FHT 64 (selon une technique similaire à celle mise en œuvre dans le bloc référencé 62, mais avec une restructuration différente), suivie d'une nouvelle étape de brouillage 65 (similaire au brouillage 61), destinée à permettre de régénérer une estimée 66 des K-2 voies autres que la voie d'intérêt, susceptibles d'interférer avec cette dernière. Le bloc référencé 66 met ainsi en œuvre un filtrage par le filtre correspondant au canal estimé. Les blocs référencés 61 à 66 de la figure 6 réalisent donc l' étape d'estimation de l'interférence induite par les signaux sources des voies autres que la voie d'intérêt du procédé de réception de l'invention. L'estimation 66 de l'interférence induite par les voies autres que la voie d'intérêt est ensuite retranchée 67 au signal reçu 30, selon une soustraction mathématique simple, de façon à ne récupérer que le signal représentatif de la voie d'intérêt n°l, ce qui constitue l'étape de retrait de l'interférence du procédé de réception de l'invention. Grâce à la IFHT 62, on génère donc les K-l estimations d'interférence des"RAKE + PIC". The invention particularly aims to overcome these various drawbacks of the prior art. More specifically, an objective of the invention is to provide a technique for receiving a CDMA signal making it possible to extract a signal more representative of the channel of interest than according to the techniques of the prior art. Another objective of the invention is to implement such a technique which makes it possible to increase the quality of the signal which is extracted. Another object of the invention is to provide such a technique which allows better cancellation of the interference induced by the existence of multiple paths in the propagation channel. The invention also aims to implement such a technique which has a reduced complexity, and which consumes little in terms of resources. These objectives, as well as others which will appear subsequently, are achieved using a method for receiving a CDMA signal comprising at least two channels, each corresponding to a separate source signal, including at least one channel. of interest that we want to extract, said received signal being conveyed by a multipath transmission channel. According to the invention, such a reception method comprises the following steps: - equalization of said received signal, delivering a single equalized signal; estimation of the interference induced by a source signal transmitted on at least one channel other than said channel of interest, from said single equalized signal; removing, from said received signal, the estimated interference induced by said source signals, for channels other than said channel of interest, to obtain a signal more representative of said channel of interest than said received signal; processing of said signal representative of said channel of interest. Thus, the invention is based on a completely new and inventive approach to the reception of CDMA signals, and to the cancellation of the interference induced by the existence of multiple paths in the propagation channel. Indeed, the invention proposes to combine the use of an equalizer and an interference cancellation architecture. The method of the invention is therefore much simpler than the methods of the prior art of the “RAKE” or “RAKE + PIC” type, since it is based in particular on the use of a single equalized signal, and not on the use of a plurality of equalized signals each corresponding to one of the discernible paths of the multipath propagation channel. In addition, as will appear more clearly in the remainder of this document, in relation to FIG. 8, the signal obtained according to the invention is much more representative of the path of interest than the signals obtained according to the techniques of the art. prior, thanks to a better cancellation of the interference induced by the other source signals. It will be noted that the performances of such a method are clearly superior to the joint performances of conventional equalization techniques on the one hand, and of “RAKE + PIC” type techniques on the other hand. Advantageously, each of said channels being obtained by multiplication of said corresponding source signal with a predetermined spreading code, the spreading code used for said at least one other channel is estimated to be blind, according to at least one predetermined criterion, during said estimation of the corresponding source signal. It is therefore not necessary, for the implementation of the invention, that the receiver knows all spread codes for all channels. Knowing the only spreading code of the channel of interest, the receiver can obtain an estimate of the interference of the source signals of the other channels, by advantageously exploiting the orthogonality properties of the spreading codes used in CDMA. Thus, said spreading code used is chosen, according to said at least one predetermined criterion, from the OVSF codes (in English "Orthogonal Variable Spreading Factor" for "orthogonal variable spreading factor") orthogonal to the spreading code of said route of interest. Preferably, said at least one predetermined criterion is a correlation criterion, said spreading code used having a significant correlation with said equalized signal. According to a first advantageous variant of the invention, said steps of estimating and removing the interference induced by said source signals are carried out in series, of SIC type (in English “Sériai Interference Cancellation”). According to a second advantageous variant of the invention, said steps of estimating and removing the interference induced by said source signals are carried out in parallel, of PIC type (in English "Parallel Interference Cancellation"). Advantageously, said interference estimation implements an inverse mathematical transformation feeding a set of decision modules in parallel, followed by a corresponding direct mathematical transformation. Unlike the "RAKE + PIC" method of the prior art, the invention therefore implements a single inverse mathematical transformation, acting on the single equalized signal after "unscrambling". The invention is therefore much less consuming in terms of resources, in particular of computation, than the "RAKE + PIC" technique, for which a fast Walsh transform had to be performed on each of the L branches corresponding to the L discernible paths of the propagation channel. multipath. The invention also implements a unique corresponding direct mathematical transformation. It is therefore particularly economical in computing resources. Preferably, said mathematical transformation is of the Hadamard Rapide Transform type, also called the Walsh Rapide Transform. It is of course also possible to envisage using any other type of mathematical transform making it possible to carry out the despreading of the equalized signal. Advantageously, said equalization implements an algorithm of the MMSE type (in English “Minimum Mean Squared Error”). Preferably, said equalization implements a reduced rank filter belonging to the group comprising: the reduced rank filters of Wiener (in English "Stochastic Gradient Multistage Wiener Filter"); Conjugate Gradient Reduced Rank Filter. The calculation of the equalizer is thus greatly simplified. Indeed, as for the calculation of a full rank Wiener equalizer, only the received signal and the pilot channel must be known. In addition, the filtering methods with reduced rank have a reduced computational complexity compared to the Wiener filter with full rank. According to an advantageous variant of the invention, said estimation of the interference induced by said source signals is iterative. It is thus possible to improve the quality of the signal representative of the channel of interest, by carrying out several iterations of the steps of estimation and removal of the interference, and of processing of the signal representative of the channel of interest. The signal thus obtained is then more representative of the channel of interest. According to another advantageous variant of the invention, said decision is a flexible decision. Advantageously, said signal is transmitted on a downlink of a UMTS system, in particular of a system of the UMTS-FDD type (for “Frequency Duplex Division ”). Preferably, said step of processing said signal representative of said channel of interest comprises at least some of the following substeps: an equalization substep; - a descrambling sub-step; a despreading sub-step; a decision sub-step, at the end of which symbols representative of the path of interest n ° i are recovered. The invention also relates to a receiver of a CDMA signal comprising at least two channels, each corresponding to a separate source signal, including at least one channel of interest which it is desired to extract, said received signal being conveyed by a communication channel. multipath transmission. According to the invention, such a receiver comprises the following means: - means for equalizing said received signal, delivering a single equalized signal; means for estimating the interference induced by a source signal transmitted on at least one channel other than said channel of interest, from said single equalized signal; means for removing, from said received signal, the estimated interference induced by said source signals, for channels other than said channel of interest, in order to obtain a signal more representative of said channel of interest than said received signal; means for processing said signal representative of said channel of interest. Other characteristics and advantages of the invention will appear more clearly on reading the following description of a preferred embodiment, given by way of simple illustrative and nonlimiting example, and of the appended drawings, among which: FIG. 1 , already commented on in relation to the prior art, presents the general principle of the transmission of a CDMA signal via a perfect propagation, exposed by analogy with the domain of colorimetry; FIG. 2 illustrates the principle of the transmission of the signal of FIG. 1 via a multi-path channel, again by analogy with the field of colorimetry; FIG. 3, also previously commented on in relation to the prior art, describes the architecture of a RAKE type receiver; FIG. 4, also presented previously, illustrates the architecture of a receiver of the prior art of the “RAKE + PIC” type; - Figure 5, already described above, shows the architecture of a receiver of the prior art implementing an equalization technique; FIG. 6 illustrates the general architecture of a receiver of the invention, jointly implementing equalization and interference cancellation techniques; - Figure 7 describes a variant of the receiver of Figure 6 with iterative processing; FIG. 8 presents the compared performances, in terms of bit error rate, of the techniques of the prior art and of the invention. The general principle of the invention is based on the joint implementation of an equalization of the received signal and the cancellation of the interference induced by the existence of multiple paths in the propagation channel. For the sake of simplification, a preferred alternative embodiment is presented throughout the rest of the document in which the interference cancellation implemented has a PIC-type structure, in parallel. Those skilled in the art will easily extend this teaching to the case where interference cancellation is carried out in series, of the SIC type. Referring to FIG. 6, the architecture of a receiver of the invention is presented. The received signal 30 feeds an equalizer 60, which implements the step of equalizing the signal received from the method of the invention, delivering an equalized signal unique. Thus, unlike prior art techniques of the “RAKE” or “RAKE + PIC” type, the equalization 60 is done directly on the received signal, considered as a whole, and no longer on each of the L branches corresponding to the L discernible paths from the multipath propagation channel. The equalization block 60 operates in two stages. During a first step (learning or updating), one proceeds to descrambling and then despreading of the received signal, the despreading being carried out with the spreading code of the pilot. The received signal is then shifted to unscramble and despread it again. This is repeated G times, where G is the size of the equalizer. The descrambling and despreading outputs constitute a size G vector. The reduced rank algorithm uses this vector as an input signal and the pilot symbol (known) as the desired output to update the equalizer. The second step (direct equalization) consists in applying the updated equalizer to the input signal to give an equalized signal which will serve as an input to the block referenced 61. The single equalized signal then undergoes "descrambling" 61, by multiplication term to term of the signal by the scrambling sequence, then feeds a module 62 in which it undergoes a Fast Reverse Hadamard Transform (or IFHT). This module 62 implements a multiplication by the Hadamard matrix containing +1 or -1, which can be made rapid using the butterfly techniques used by the FFT (Fast Fourier Transform). . It will be noted that, according to this architecture, a single inverse mathematical transformation 62 must be implemented, on the equalized and unscrambled signal, which constitutes a significant saving in terms of computing resources compared to the “RAKE + PIC” technique according to which one FWT was to be performed on each of the L branches of the RAKE (see Figure 4). In the example of FIG. 6, we consider that the channel of interest that we are trying to extract is channel # 1. The signals from IFHT 62 feed Kl decision modules in parallel, referenced 63 2 to 63 κ , each corresponding to one of the channels 2 to K other than the channel of interest No. 1. Decisions 63 2 to 63 κ can be hard decisions or flexible decisions. For example, the sign of the result obtained by weighting the output of the IFHT operation 62 by the length of the spreading sequence considered can be adopted as a flexible decision criterion. After decisions, all the signals undergo a direct Hadamard transform, or FHT 64 (according to a technique similar to that implemented in the block referenced 62, but with a different restructuring), followed by a new scrambling step 65 (similar to jamming 61), intended to make it possible to regenerate an estimated 66 of the K-2 channels other than the channel of interest, liable to interfere with the latter. The block referenced 66 thus implements filtering by the filter corresponding to the estimated channel. The blocks referenced 61 to 66 in FIG. 6 therefore carry out the step of estimating the interference induced by the source signals of the channels other than the channel of interest of the reception method of the invention. The estimate 66 of the interference induced by the channels other than the channel of interest is then subtracted 67 from the received signal 30, according to a simple mathematical subtraction, so as to recover only the signal representative of the channel of interest n ° l, which constitutes the step of removing the interference from the reception method of the invention. Thanks to IFHT 62, we therefore generate the Kl interference estimates of
K-l signaux sources correspondant aux K-l plus puissants canaux (virtuels). Ces interférences (estimées ou décidées) sont retranchées (67) du signal reçu 30 pour obtenir une meilleure estimée du signal émis. L'estimation d'interférence peut également ne concerner que M≤K signaux. On notera que, dans les modules de décision 632 à 63κ, les codes d'étalement autres que celui de la voie d'intérêt ne sont pas connus, mais estimés en aveugle. On considère tous les codes OVSF (virtuels) orthogonaux au code d'étalement d'intérêt. On utilise pour ce faire le fait que les codes vus par la voie d'intérêt sont toujours orthogonaux à son code d'étalement, même s'ils ne sont pas « originellement » de la même longueur. Le signal obtenu après soustraction 67 subit ensuite l'étape de traitement du procédé de réception de l'invention, au cours de laquelle il est égalisé 68, selon une méthode similaire à celle mise en œuvre par l'égaliseur 60, puis alimente le bloc 69 regroupant les codes d'étalement et de brouillage de la voie d'intérêt. Le bloc référencé 69 met en œuvre une multiplication par la séquence conjointe de brouillage et d'étalement du signal d'intérêt, suivie d'une sommation sur la longueur de la séquence d'étalement du signal d'intérêt. A l'issue de ces différentes opérations, une décision ferme 70 est prise, sur le signe de la quantité résultant du bloc référencé 69, qui permet d'extraire le symbole b, émis sur la voie d'intérêt n°l. La figure 7 présente une variante du récepteur de la figure 6, dans le cas où l'on met en œuvre une itération des étapes d'estimation et de retrait de l'interférence induite par les signaux sources. Les mêmes éléments du récepteur sont désignés par les mêmes références numériques sur les figures 6 et 7. Par souci de simplification, on ne décrit pas plus en détail les traitements appliqués au signal déjà décrits précédemment en relation avec la figure 6. On notera cependant que la décision ferme 70 permettant d'extraire le symbole b, émis sur la voie d'intérêt n°l est prise en sortie du module d'IFHT 62. Les décisions 632 à 63κ sont des décisions souples, comme évoqué précédem ent, et sont répétées à chaque itération successive du traitement. De même, une décision 70 sur la voie d'intérêt n°l est prise à chaque itération de l'étape d'IFHT 62. La figure 8 illustre les performances comparées des différentes méthodes décrites précédemment en relation avec les figures 3 à 6, à savoir : le récepteur « RAKE » de l'art antérieur ; le récepteur « RAKE + PIC » de l'art antérieur ; la technique d'égalisation de l'art antérieur ; la technique conjointe d'égalisation et d'annulation d'interférence de l'invention. La figure 8 présente les courbes du taux d'erreur binaire en fonction du rapport signal à bruit pour un système présentant une longueur d'étalement N=16, un nombre de canaux utilisés K=15, et un canal de propagation à 6 trajets. Le rapport signal à bruit ou SNR du canal d'intérêt varie entre 0 dB et 15 dB, alors que le rapport signal à bruit des 14 autres canaux est fixé à 10 dB. La chaîne de simulation ne prend pas en compte le codage canal. Les voies sont modulées enKl source signals corresponding to the most powerful Kl channels (virtual). These interferences (estimated or decided) are subtracted (67) from the received signal 30 to obtain a better estimate of the transmitted signal. The interference estimate can also relate only to M≤K signals. It will be noted that, in the decision modules 63 2 to 63 κ , the spreading codes other than that of the channel of interest are not known, but estimated blind. We consider all OVSF (virtual) codes orthogonal to the code of spread of interest. To do this, we use the fact that the codes seen by the route of interest are always orthogonal to its spreading code, even if they are not "originally" of the same length. The signal obtained after subtraction 67 then undergoes the processing step of the reception method of the invention, during which it is equalized 68, according to a method similar to that implemented by the equalizer 60, then feeds the block 69 grouping the spreading and scrambling codes of the channel of interest. The block referenced 69 implements a multiplication by the joint scrambling and spreading sequence of the signal of interest, followed by a summation over the length of the spreading sequence of the signal of interest. At the end of these various operations, a firm decision 70 is taken, on the sign of the quantity resulting from the block referenced 69, which makes it possible to extract the symbol b, transmitted on the channel of interest no. FIG. 7 presents a variant of the receiver of FIG. 6, in the case where an iteration of the steps of estimation and removal of the interference induced by the source signals is implemented. The same elements of the receiver are designated by the same reference numerals in FIGS. 6 and 7. For the sake of simplification, the processing operations applied to the signal already described previously in connection with FIG. 6 are not described in more detail. It will however be noted that the firm decision 70 making it possible to extract the symbol b, emitted on the channel of interest no. 1 is taken at the output of the IFHT module 62. The decisions 63 2 to 63 κ are flexible decisions, as mentioned previously, and are repeated at each successive iteration of the processing. Similarly, a decision 70 on the path of interest No. 1 is taken at each iteration of the IFHT step 62. FIG. 8 illustrates the compared performances of the different methods described above in relation to FIGS. 3 to 6, namely: the “RAKE” receiver of the prior art; the “RAKE + PIC” receiver of the prior art; the prior art equalization technique; the joint technique of equalization and interference cancellation of the invention. FIG. 8 presents the curves of the bit error rate as a function of the signal to noise ratio for a system having a spreading length N = 16, a number of channels used K = 15, and a propagation channel with 6 paths. The signal to noise ratio or SNR of the channel of interest varies between 0 dB and 15 dB, while the signal to noise ratio of the other 14 channels is fixed at 10 dB. The simulation chain does not take into account channel coding. The channels are modulated in
BPSK. Pour la méthode de l'invention, les valeurs des voies retranchées sont les décisions 632 à 63k faites à la sortie de la transformée d'Hadamard rapide inverseBPSK. For the method of the invention, the values of the entrenched channels are the decisions 63 2 to 63 k made at the output of the inverse fast Hadamard transform
IFHT 62. Les performances du procédé de réception de l'invention apparaissent très clairement au vu de la figure 8, qui montre que la technique de l'invention permet d'obtenir une réduction du rapport signal à bruit de 4 dB pour un taux d'erreur binaire (BER, « Bit Error Rate ») de 10"3, par rapport à la technique d'égalisation classique de l'art antérieur. Le gain obtenu, selon l'invention, par rapport à la technique de l'art antérieur dite « RAKE » est encore plus grand. Les performances de la figure 8 ont été obtenues pour une variante de l'invention dans laquelle on a mis en œuvre une seule étape d'estimation des voies interférentes et de retrait de cette interférence, et dans laquelle les décisions prises par les modules 632 à 63κ sont des décisions dures. Ces performances peuvent encore être accrues en mettant en œuvre une itération des différentes étapes du procédé de l'invention, comme illustré en figure 7, et en prenant des décisions souples. IFHT 62. The performance of the reception method of the invention appears very clearly in the light of FIG. 8, which shows that the technique of the invention makes it possible to obtain a reduction in the signal to noise ratio of 4 dB for a rate d bit error rate (BER) of 10 "3 , compared to the conventional equalization technique of the prior art. The gain obtained, according to the invention, compared to the technique of the art so-called "RAKE" is even greater. The performances of FIG. 8 were obtained for a variant of the invention in which a single step of estimation of the interfering pathways and removal of this interference was implemented, and in which the decisions taken by the modules 63 2 to 63 κ are hard decisions. These performances can be further increased by implementing an iteration of the various steps of the method of the invention, as illustrated in FIG. 7, and by taking flexible decisions es.

Claims

REVENDICATIONS
1. Procédé de réception d'un signal CDMA comprenant au moins deux voies, correspondant chacune à un signal source distinct, dont au moins une voie d'intérêt que l'on veut extraire, ledit signal reçu étant véhiculé par un canal de transmission à trajets multiples, caractérisé en ce qu'il comprend les étapes suivantes : égalisation dudit signal reçu, délivrant un signal égalisé unique ; estimation de l'interférence induite par un signal source émis sur au moins une voie autre que ladite voie d'intérêt, à partir dudit signal égalisé unique ; retrait, dudit signal reçu, de l'interférence estimée induite par lesdits signaux source, pour les voies autres que ladite voie d'intérêt, pour obtenir un signal plus représentatif de ladite voie d'intérêt que ledit signal reçu ; traitement dudit signal représentatif de ladite voie d'intérêt, et en ce que, chacune desdites voies étant obtenue par multiplication dudit signal source correspondant avec un code d'étalement prédéterminé, le code d'étalement utilisé pour ladite au moins une autre voie est estimé en aveugle, selon au moins un critère prédéterminé, lors de ladite estimation du signal source correspondant.1. Method for receiving a CDMA signal comprising at least two channels, each corresponding to a separate source signal, including at least one channel of interest which it is desired to extract, said received signal being conveyed by a transmission channel to multipath, characterized in that it comprises the following steps: equalization of said received signal, delivering a single equalized signal; estimation of the interference induced by a source signal transmitted on at least one channel other than said channel of interest, from said single equalized signal; removing, from said received signal, the estimated interference induced by said source signals, for channels other than said channel of interest, to obtain a signal more representative of said channel of interest than said received signal; processing of said signal representative of said channel of interest, and in that, each of said channels being obtained by multiplication of said corresponding source signal with a predetermined spreading code, the spreading code used for said at least one other channel is estimated blind, according to at least one predetermined criterion, during said estimation of the corresponding source signal.
2. Procédé de réception selon la revendication 1, caractérisé en ce que ledit code d'étalement utilisé est choisi, selon ledit au moins un critère, parmi les codes2. Reception method according to claim 1, characterized in that said spreading code used is chosen, according to said at least one criterion, from the codes
OVSF (en anglais « Orthogonal Variable Spreading Factor » pour « facteur d'étalement variable orthogonal ») orthogonaux au code d'étalement de ladite voie d'intérêt.OVSF (in English "Orthogonal Variable Spreading Factor" for "orthogonal variable spreading factor") orthogonal to the spreading code of said channel of interest.
3. Procédé de réception selon l'une quelconque des revendications 1 et 2, caractérisé en ce que ledit au moins un critère prédéterminé est un critère de corrélation, ledit code d'étalement utilisé présentant une corrélation importante avec ledit signal égalisé.3. Reception method according to any one of claims 1 and 2, characterized in that said at least one predetermined criterion is a correlation criterion, said spreading code used having a significant correlation with said equalized signal.
4. Procédé de réception selon l'une quelconque des revendications 1 à 3, caractérisé en ce que lesdites étapes d'estimation et de retrait de l'interférence induite par lesdits signaux sources sont réalisées en série, de type SIC (en anglais « Sériai Interférence Cancellation »).4. Reception method according to any one of claims 1 to 3, characterized in that said steps of estimating and removing the interference induced by said source signals are carried out in series, of SIC type (in English "Sériai Interference Cancellation").
5. Procédé de réception selon l'une quelconque des revendications 1 à 3, caractérisé en ce que lesdites étapes d'estimation et de retrait de l'interférence induite par lesdits signaux sources sont réalisées en parallèle, de type PIC (en anglais « Parallel Interférence Cancellation »).5. Reception method according to any one of claims 1 to 3, characterized in that said steps of estimating and removing the interference induced by said source signals are carried out in parallel, of PIC type (in English "Parallel Cancellation Interference ”).
6. Procédé de réception selon la revendication 5, caractérisé en ce que ladite estimation de l'interférence met en œuvre une transformation mathématique inverse alimentant un ensemble de modules de décision en parallèle, suivie d'une transformation mathématique directe correspondante. 6. Reception method according to claim 5, characterized in that said interference estimation implements an inverse mathematical transformation feeding a set of decision modules in parallel, followed by a corresponding direct mathematical transformation.
7. Procédé de réception selon la revendication 6, caractérisé en ce que ladite transformation mathématique est de type Transformée d'Hadamard Rapide. 7. Reception method according to claim 6, characterized in that said mathematical transformation is of the Rapid Hadamard Transformation type.
8. Procédé de réception selon l'une quelconque des revendications 1 à 7, caractérisé en ce que ladite égalisation met en œuvre un algorithme de type MMSE (en anglais « Minimum Mean Squared Error »). 8. Reception method according to any one of claims 1 to 7, characterized in that said equalization implements an MMSE type algorithm (in English "Minimum Mean Squared Error").
9. Procédé de réception selon la revendication 8, caractérisé en ce que ladite égalisation met en œuvre un filtre à rang réduit appartenant au groupe comprenant : les filtres à rang réduit de Wiener (en anglais « Stochastic Gradient Multistage Wiener Filter ») ; - les filtres à rang réduit à gradient conjugué (en anglais « Conjugate Gradient Reduced Rank Filter ») ;9. A reception method according to claim 8, characterized in that said equalization implements a reduced rank filter belonging to the group comprising: Wiener reduced rank filters (in English "Stochastic Gradient Multistage Wiener Filter"); - Conjugate Gradient Reduced Rank Filter (in English)
10. Procédé de réception selon l'une quelconque des revendications 1 à 9, caractérisé en ce que ladite estimation de l'interférence est itérative.10. Reception method according to any one of claims 1 to 9, characterized in that said estimation of the interference is iterative.
11. Procédé de réception selon l'une quelconque des revendications 1 à 9, caractérisé en ce que ladite décision est une décision souple.11. Reception method according to any one of claims 1 to 9, characterized in that said decision is a flexible decision.
12. Procédé de réception selon l'une quelconque des revendications 1 à 11, caractérisé en ce que ledit signal est transmis sur une voie descendante d'un système UMTS.12. Reception method according to any one of claims 1 to 11, characterized in that said signal is transmitted on a downlink of a UMTS system.
13. Procédé de réception selon l'une quelconque des revendications 1 à 12, caractérisé en ce que ladite étape de traitement dudit signal représentatif de ladite voie d'intérêt comprend au moins certaines des sous-étapes suivantes : une sous-étape d'égalisation ; une sous-étape de débrouillage ; une sous-étape de désétalement ; - une sous-étape de décision.13. Reception method according to any one of claims 1 to 12, characterized in that said step of processing said signal representative of said path of interest includes at least some of the following substeps: an equalization substep; a descrambling sub-step; a despreading sub-step; - a decision sub-step.
14. Récepteur d'un signal CDMA comprenant au moins deux voies, correspondant chacune à un signal source distinct, dont au moins une voie d'intérêt que l'on veut extraire, ledit signal reçu étant véhiculé par un canal de transmission à trajets multiples, caractérisé en ce qu'il comprend les moyens suivants : des moyens d'égalisation dudit signal reçu ; des moyens d'estimation de l'interférence induite par un signal source émis sur au moins une voie autre que ladite voie d'intérêt ; des moyens de retrait, dudit signal reçu, de l'interférence estimée induite par lesdits signaux source, pour les voies autres que ladite voie d'intérêt, pour obtenir un signal plus représentatif de ladite voie d'intérêt que ledit signal reçu ; des moyens de traitement dudit signal représentatif de ladite voie d'intérêt, et en ce que, chacune desdites voies étant obtenue par multiplication dudit signal source correspondant avec un code d'étalement prédéterminé, il comprend également des moyens d'estimation en aveugle du code d'étalement utilisé pour ladite au moins une autre voie, selon au moins un critère prédéterminé, mis en œuvre lors de ladite estimation du signal source correspondant. 14. Receiver of a CDMA signal comprising at least two channels, each corresponding to a separate source signal, including at least one channel of interest which it is desired to extract, said received signal being conveyed by a multipath transmission channel , characterized in that it comprises the following means: means for equalizing said received signal; means for estimating the interference induced by a source signal transmitted on at least one channel other than said channel of interest; means for removing, from said received signal, the estimated interference induced by said source signals, for channels other than said channel of interest, in order to obtain a signal more representative of said channel of interest than said received signal; means for processing said signal representative of said channel of interest, and in that, each of said channels being obtained by multiplication of said corresponding source signal with a predetermined spreading code, it also comprises means for blind estimation of the code spread used for said at least one other channel, according to at least one predetermined criterion, implemented during said estimation of the corresponding source signal.
PCT/FR2004/002364 2003-09-18 2004-09-17 Method for receiving a cdma interference suppression signal and a corresponding receiver WO2005027367A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR03/10987 2003-09-18
FR0310987A FR2860110B1 (en) 2003-09-18 2003-09-18 METHOD FOR RECEIVING AN INTERFERENCE CANCELLATION CDMA SIGNAL AND CORRESPONDING RECEIVER.

Publications (1)

Publication Number Publication Date
WO2005027367A1 true WO2005027367A1 (en) 2005-03-24

Family

ID=34224342

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/FR2004/002364 WO2005027367A1 (en) 2003-09-18 2004-09-17 Method for receiving a cdma interference suppression signal and a corresponding receiver

Country Status (2)

Country Link
FR (1) FR2860110B1 (en)
WO (1) WO2005027367A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8463196B2 (en) 2007-08-09 2013-06-11 Freescale Semiconductor, Inc. Method and decoder for decoding a wireless transmission from a predefined user

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002031995A2 (en) * 2000-10-06 2002-04-18 Ericsson Inc Method and apparatus for subtracting multiple rays of multiple interfering received signals
EP1289162A2 (en) * 2001-08-28 2003-03-05 Texas Instruments Incorporated Combined equalizer and spread spectrum interference canceller method and implementation for the downlink of CDMA systems

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002031995A2 (en) * 2000-10-06 2002-04-18 Ericsson Inc Method and apparatus for subtracting multiple rays of multiple interfering received signals
EP1289162A2 (en) * 2001-08-28 2003-03-05 Texas Instruments Incorporated Combined equalizer and spread spectrum interference canceller method and implementation for the downlink of CDMA systems

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8463196B2 (en) 2007-08-09 2013-06-11 Freescale Semiconductor, Inc. Method and decoder for decoding a wireless transmission from a predefined user

Also Published As

Publication number Publication date
FR2860110B1 (en) 2006-01-13
FR2860110A1 (en) 2005-03-25

Similar Documents

Publication Publication Date Title
EP1189361B1 (en) Method of multiuser detection
FR2715523A1 (en) Receiver and receiving method for a code division multiple access system.
FR2742619A1 (en) MULTICARIZER EQUALIZATION METHOD FOR MULTICAPTOR RECEPTION IN THE PRESENCE OF PROPAGATION MULTIFRABY AND INTERFERENCE, AND RECEIVER FOR ITS IMPLEMENTATION
WO2002023841A1 (en) Method for optimal estimation of a propagation channel relying solely on pilot symbols and corresponding estimator
EP1221219B1 (en) Equalising method in receivers using a combination of multiple carrier and code division multiple access modulating techniques
FR2813464A1 (en) MULTI-USER DETECTION METHOD
FR2810820A1 (en) Interference cancellation technique for CDMA Rake receivers has recursive correction for non orthogonality due to propagation
FR2813465A1 (en) METHOD OF JOINT CHANNEL ESTIMATION AND ARRIVAL DIRECTION
EP0945996B1 (en) CDMA receiver with parallel interference suppression and weighting
EP0669729B1 (en) Method for implementing multichannel equalisation in a radio receiver in the presence of interferece and multipath propagation
WO2007028917A2 (en) Iterative estimation for mimo channel in a cdma network
EP1221206B1 (en) Method and device for demodulating signals from multiusers
EP1283605A1 (en) Method for data date increase in a communication system with N transmitters and M receivers
FR2803467A1 (en) METHOD OF ESTIMATING A TRANSMISSION OR TELECOMMUNICATION CHANNEL
WO2005027367A1 (en) Method for receiving a cdma interference suppression signal and a corresponding receiver
EP1244243A1 (en) Global MMSE equalisation
WO2005109708A1 (en) Method for determination of code spread used in a cdma signal and corresponding communication device
WO2005062478A1 (en) Cdma data decoding device and method, and corresponding system
EP1221207B1 (en) Method and device for sequencing signals from multiusers
EP1135855A1 (en) Digital filter with parallel architecture and spread-spectrum signal receiver using same
FR2849306A1 (en) Digital word packet radio propagation compensation having iterative feedback estimator/symbol detector with operations providing distinct channel estimates/removing word groups received dependent transmission conditions
EP1693969B1 (en) CDMA signal reception method with parallel interference suppression and its corresponding stage and receiver
FR3137231A1 (en) Methods for transmitting and receiving spectrally spread wireless communication with high potential spectral efficiency and associated devices
WO2006131665A1 (en) Device for selecting a receiver based on the environment
WO2007048964A1 (en) Method and device for hybrid reception comprising a channel equalizer and an interference canceller

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BW BY BZ CA CH CN CO CR CU CZ DK DM DZ EC EE EG ES FI GB GD GE GM HR HU ID IL IN IS JP KE KG KP KZ LC LK LR LS LT LU LV MA MD MK MN MW MX MZ NA NI NO NZ PG PH PL PT RO RU SC SD SE SG SK SY TJ TM TN TR TT TZ UA UG US UZ VN YU ZA ZM

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): BW GH GM KE LS MW MZ NA SD SZ TZ UG ZM ZW AM AZ BY KG MD RU TJ TM AT BE BG CH CY DE DK EE ES FI FR GB GR HU IE IT MC NL PL PT RO SE SI SK TR BF CF CG CI CM GA GN GQ GW ML MR SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
122 Ep: pct application non-entry in european phase
点击 这是indexloc提供的php浏览器服务,不要输入任何密码和下载