FR2886789A1 - DEVICE AND METHOD FOR SELECTING A RECEIVER BASED ON THE ENVIRONMENT - Google Patents

Abstract

This signal receiving device (1) originating from a transmission channel comprises means (10, 20, 40) for obtaining samples of a signal, means (50) for calculating a value dependent on the transmission channel. autocorrelation of a series of samples, means (50) for comparing this value with a predetermined threshold, and means (50, 81, 82) for selecting, based on the result of the comparison, and for processing samples signal, an equalizer (70) or a filter (60) adapted to the channel.

Description

2886789 1

  BACKGROUND OF THE INVENTION The field of the invention is that of digital communications.

  More particularly, the invention relates to a method and a device for selecting, according to the transmission conditions, a receiver selected from an equalizer and a filter adapted to the transmission channel.

  The invention finds particular application in the field of digital radio communications between a base station and a mobile terminal.

  But it can also apply to any type of transmission (copper line, acoustic, underwater, etc.).

  The technical problem solved by the invention will now be explained in the context of direct sequence spread spectrum mobile communication techniques (better known by the acronym DSCDMA for "Direct Sequence Multiple Access Code Division").

  The receiver traditionally used in DS-CDMA systems, in base stations, as in mobile terminals, is a rake receiver, better known by the English name of RAKE.

  It is known that RAKE-type receivers offer limited performance when systems are heavily loaded and / or when spreading factors are low.

  In the particular case where the base station transmits and the mobile receives, this link being called "downlink", a receiver more efficient than the RAKE type receiver is obtained by the combination of a chip level equalizer with a suitable correlator to the desired code.

  It is recalled here that, in accordance with CDMA technology, each information symbol is issued as a series of sub-symbols resulting from the spread spectrum, and that each of these sub-symbols is traditionally designated by the term "chip".

  Such a receptor is in particular described in the document "Downlink specific linear equalization for selective frequency CDMA cellular systems", T. P. Krauss, W. J. Hillery and M. D. Zoltowski, J. VLSI Signal Process, Vol. 30, pages 143-161, March 2002.

  In practice, and especially for reasons related to the structure of the signals processed by the equalizer, the most used form of equalizer is the linear equalizer minimizing the mean square error (MMSE in English, for "Minimum Mean Square Error "). The resulting receiver is usually called the MMSE linear receiver or LMMSE receiver.

  The LMMSE receivers have a major disadvantage in that they are significantly more complex than a RAKE receiver, which results in increased consumption of the mobile terminal.

  No. 4 2004/0042537, entitled "Spread spectrum receiver apparatus and method" by CD Frank, March 2004, discloses a device implementing in parallel a rake receiver of RAKE type and an MMSE equalizer. adaptive.

  This device comprises means for selecting one of these two receivers according to the actual performance of these receivers in a given environment.

  This method, which requires the use of both receivers in parallel, is necessarily expensive in terms of complexity and consumption.

  In order to solve this problem, J. Black and MA Howard have described (US Patent Application No. 2004/0240531, "Communication receiver with hybrid equalizer", December 2004) a method for deciding when to activate the equalizer, the receiver of RAKE type remains permanently active to allow deciding when to switch back to the RAKE receiver alone.

  In this solution, the two receivers operate in parallel during the decision periods.

  OBJECT AND SUMMARY OF THE INVENTION The object of the invention is to overcome the above disadvantages.

  For this purpose, the invention relates to a device for receiving a signal carrying digital information, said signal being derived from a transmission channel, this device comprising means for obtaining samples of this signal, an equalizer and a filter adapted to the channel.

  This device comprises means for calculating a value dependent on the autocorrelation of a series of samples, means for comparing this value with a predetermined threshold and means for selecting, in order to process signal samples, the equalizer or filter adapted according to the result of the comparison.

  Those skilled in the art understand that according to the invention, the calculation of the value dependent on the autocorrelation is carried out without any prior treatment, by the equalizer or by the adapted filter, of the samples used for this calculation.

  Thus, in accordance with the invention, the selection of the equalizer or filter adapted to the channel is carried out without these elements being implemented, including during the decision period.

  Due to the fact that the autocorrelation is taken into account, the noise level influences the decision to use the equalizer or the filter adapted to the channel, this parameter having a not inconsiderable impact on the performance of the equalizer.

  Indeed, it is known to those skilled in the art that the performances of an equalizer are close to those of the filter adapted to the channel when the signal-to-noise ratio (SNR) decreases. that is, when the power of the noise is large relative to the power of the signal received from the transmitting base station.

In all the description:

  the notation x * denotes the complex conjugate of the scalar x; the notation lx 'denotes the modulus of the scalar x; - the notation E {x} denotes the mathematical expectation of the random variable x; - MT designates the transpose of the matrix M; M "denotes the conjugate transpose of the matrix M, and the operator 11.112 denotes the square of a matrix norm.

  In a preferred embodiment, the value dependent on the autocorrelation is of the form J = al I [3R-1112, where a and J3 are weighting coefficients, I the identity matrix, and R is the temporal autocorrelation matrix. of the signal.

  As demonstrated later, this characteristic advantageously makes it possible to evaluate the difference in gain that can be made by the equalizer with respect to the filter adapted to the channel.

  Thus, the invention makes it possible to select the equalizer only when a significant gain can be obtained with respect to the filter adapted to the channel. It is understood that it is preferable not to select the equalizer when it provides a limited gain, because the use of the equalizer is much more resource consuming for the terminal.

  Those skilled in the art therefore understand that the choice of the predetermined threshold is made according to a performance / consumption compromise.

  Preferably, the weighting coefficients a and p are respectively equal to 1 / Lp and 1 / ro, where ro and Lp respectively represent the power of the received signal and the number of coefficients (or length) of the equalizer.

  As explained later, this choice of the coefficient p frees the value of the criterion of the power of the received signal.

  Moreover, this choice of the parameter a makes it possible to limit the dependence between the value of the criterion and the length of the equalizer, for a given transmission environment.

  The signal received by the device according to the invention is transmitted by a base station. In a preferred embodiment, the parameters defining the autocorrelation matrix depend on: the estimated coefficients of the transmission channel; - the estimated power of the noise; and the ratio of the transmit power of a pilot signal from that base station to the total transmit power of that station.

  This characteristic advantageously makes it possible to limit the number of operations necessary to obtain the autocorrelation matrix, since it is constructed directly from this information.

  Correlatively, the invention relates to a method of receiving a signal carrying a digital information, said signal being derived from a transmission channel. This method comprises: a step of obtaining samples of this signal; A step of calculating a value dependent on the autocorrelation of a series of samples; a step of comparing this value with a predetermined threshold; a selection step, an equalizer or a filter adapted to the channel as a function of the result of the comparison; and a step of processing samples of the signal with the selected element.

  In a preferred implementation, the various steps of the method are determined by computer program instructions.

  Consequently, the invention also relates to a computer program on an information medium, this program being capable of being implemented in a reception device, this program comprising instructions adapted to the implementation of a receiving method as briefly described above.

  This program can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any other form form

desirable.

  The invention also relates to a readable information medium by a receiving device, and comprising instructions of a computer program as mentioned above.

  The information carrier may be any entity or device capable of storing the program. For example, the medium may comprise storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or a magnetic recording medium, for example a diskette (floppy disc) or a disk hard.

  On the other hand, the information medium may be a transmissible medium such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means. The program according to the invention can be downloaded in particular on an Internet type network.

  Alternatively, the information carrier may be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the method in question.

Brief description of the drawings

  Other features and advantages of the present invention will emerge from the description given below, with reference to the accompanying drawings which illustrate an embodiment having no limiting character. In the figures: - Figure 1 shows a receiving device according to the invention in a preferred embodiment; FIG. 2 represents, in flowchart form, the main steps of a reception method according to the invention; and Figures 3 and 4 illustrate the results obtained in one embodiment of the invention.

  Detailed description of an embodiment

  Figure 1 schematically shows a receiving device 1 according to the invention.

  In the preferred embodiment described herein, the receiver 1 is a receiver adapted to a downlink type transmission of a DS-CDMA system according to the W-CDMA standard. For more details on this standard, those skilled in the art will be able to refer to the document "H. Holm and A. Toskala, Eds., WCDMA for UMTS, ed., Wiley, 2002".

  In a known manner, this receiver 1 comprises an antenna 10, a radio frequency chain 20 making it possible in particular to bring the signal back to baseband carrier frequency and to carry out Nyquist root filtering, and an analog / digital converter 40. provides samples of the digital signal.

  In the embodiment described here, we will assume that the samples delivered by the analog / digital converter 40 are sampled at the chip rate.

  The device 1 also comprises an equalizer 70 and a filter adapted to the transmission channel 60.

  In the embodiment described here, the matched filter 60 is arranged in parallel with the equalizer 70.

  In the example described here, the equalizer 70 is an LMMSE receiver, and the matched filter 60 is a RAKE type receiver.

  According to the invention, the receiving device 1 comprises digital data processing means 50 of the type conventionally found in a mobile terminal.

  More specifically, these processing means 50 are adapted to execute a computer program PROG stored in an information medium 52, this computer program comprising instructions for carrying out the steps of a reception method according to the invention and whose flowchart is given in Figure 2.

  In the embodiment described herein, it will be assumed that the receiver (i.e. equalizer 70 or filter 60 adapted to the transmission channel) is selected every 0.66 milliseconds, or each slot of the W-CDMA standard.

  For this purpose, the processing means 50 collect, during a step E10, the samples of a slot n at the output of the means 10 to 40 for obtaining samples.

  This sample collection step E10 is followed by a step E20 during which the processing means 50 calculate a value J dependent on the autocorrelation of the samples of this slot.

  It is recalled firstly that the impulse response of the linear equalizer MMSE chip time can be expressed in the case of a single receiving antenna as p = [Q; H "H + t] 'Hyeo (1) where: p is the column vector of length Lp containing the samples of the impulse response of the equalizer; H is a Toeplitz matrix Lp x (Lh + Lp 1) gathering the elements of the impulse response of the transmission channel, denoted h = [ho, h11 ..., hLh_1] T in the form: H '= h, h0 h hi.n-1 0 0 ho h, hLh_1 - r' is the correlation matrix of the noise; a column vector of length Lp of which all the elements are zero except the A + lemma, where o is the delay of restitution of the equalizer, and a2d is the transmission power of the desired signal, namely, in the particular case of the downlink DS-CDMA systems, the total power of the signal transmitted by the base station.

  For further information, those skilled in the art may refer to T. P. Krauss, W. J. Hillery and M. D. Zoltowski, "Downlink specific linear equalization for frequency selective CDMA cellular systems", J. VLSI Signal Process, Vol. 30, pages 143-161, March 2002.

  In known manner, the RAKE type receiver is mathematically equivalent to a filter adapted to the transmission channel followed by a correlator.

  For more information on this subject, a person skilled in the art will be able to refer to the document C. Laot and E. Hardouin "Infinite-length implementation of linear chip-level equalizers by blind recursive filtering for the DS-CDMA downlink" in Proc . IEEE Int. Conf. Common. June 2004.

  However, the term H "eo of equation (2) is no more than a vector containing all or part of the impulse response of the filter adapted to the transmission channel, such a consideration has in particular been exploited in the aforementioned document in which it is shown that the LMMSE receiver can decompose into a prefilter followed by a RAKE-type receptor.

  The difference between the processing performed by the RAKE receiver and the LMMSE receiver therefore lies in the matrix R = 6; H "H + F (2) which is none other than the temporal autocorrelation matrix of the received signal.

  In particular, if the matrix R is proportional to the identity matrix, the LMMSE receiver is reduced to a receiver of RAKE type and no gain is provided by the equalizer.

  On the other hand, if the matrix R is very different from the identity, the receiver of the RAKE type is then very different from the LMMSE receiver, sign that the latter is able to bring a significant gain compared to the receiver of the RAKE type.

  It is thus understood that the gain provided by the equalizer is a function of the difference between the matrix R and the identity matrix, since the more this matrix approaches an identity matrix, the more the LMMSE receiver will tend towards a receiver of RAKE type.

  Also, in the preferred embodiment, the value J dependent on the autocorrelation of the samples of the slot is of the form: J = a'IIQR IIZ (3) where a and 13 are weights explained later.

  This value measures the standard of the difference between the autocorrelation matrix of the received signal R and an identity matrix in order to translate the gain that can be brought by the LMMSE equalizer 70 relative to the receiver of the RAKE 60 type.

  It will be noted that via the autocorrelation matrix this value J depends on the channel, but also on the power of the signal, the power of the noise and the correlation of the noise.

  Since the received signal yn is sampled at the chip rate, the temporal autocorrelation matrix R of this signal has a Toeplitz structure and a Hermitian symmetry of the form: ro r ro Where rra = E Y - ,,,}

R (4)

  The main diagonal of the autocorrelation matrix R carries the power of the received signal. So that the decision criterion is independent of the power of the received signal and is zero when R is proportional to the identity, we put Q = 1 / ro.

  In addition, it can be seen by developing formula (3) that the value of the criterion increases more than linearly with the length of the equalizer L for the same environment. This is not desirable because even if in practice the performance of the equalizer 70 depends on its length, the dependence is complex and far from linear. To alleviate this dependence, we set a = 1 / Lp.

  Using the Frobenius norm and the Hermitian Toeplitz structure of the autocorrelation matrix R, the criterion can finally be calculated practically by the formula: ## EQU1 ## It will be noted that the number of terms involved in the sum of the formula (5) can be reduced by considering the structure of the autocorrelation matrix R described by the formula (2) and the particular form of the HT matrix.

  Assuming that the lengths of the transmission channels associated respectively with the signal from the desired base station and the interfering signals are less than or equal to a given length Lmax, it can be shown that the elements rm of the temporal autocorrelation matrix R are zero for m Lmax.

  The resulting criterion is finally written: Énax _ 2 (Lp-l) r, 12

T (6) r / = 1

  In practice, the length Lmax can be determined from the chip rate and the average power profile of the transmission channels likely to be encountered in the environment of the application.

  In the case of the W-CDMA standard, for example Lmax = 10.

  Several methods can be used to determine the autocorrelation of the received signal.

  According to a first method, each autocorrelation value rm is estimated directly from a given time average by the formula: ## EQU1 ## where y is the nth sample at the chip rate of the received signal and N the length of the received signal sequence entering the estimate.

  In the application considered, N will typically be of the order of the length of the W-CDMA slot, or 2560 chips.

  This first method of calculation has the advantage of providing the true autocorrelation of the received signal, whatever the statistics of the interference.

  In return, it requires a significant computational complexity whose cost is felt all the more when the RAKE receiver 60 is finally selected since this receiver does not exploit the knowledge of autocorrelation of the signal.

  In a preferred embodiment, another method of calculating autocorrelation that is much more economical in terms of terminal consumption is used.

  This preferred variant will now be explained in detail. In order to simplify the calculation, it is assumed that the noise correlation matrix is diagonal, that is: t = a! I (8) where o- is the power of the noise.

  The autocorrelation matrix R is rewritten from the formula (2) then in the form: R = cT HHH + cT I (9) The elements of the autocorrelation matrix R can then be estimated at a real positive coefficient close to for example by the method described in E. Hardouin "Chip Level Equalization for the Downlink (7) of DS-CDMA Mobile Communications Systems", Ph.D. Thesis, University of Rennes 1, May 2004, pages 110-114.

  This method uses the pilot channel present in the W-CDMA and cdma2000 standards, which is a channel common to all users of the cell which carries only pilot symbols all identical and known in advance of the receiver. This method requires that the ratio between the power of the pilot channel and the total power transmitted by the base station is known by the receiver, for example by being transmitted by the base station.

  This ratio is noted p and is defined as: = P p / oh! where Pn ,,,, e is the power of the pilot channel transmitted by the base station and is the total power transmitted by the base station. The autocorrelation values are then obtained by: = 1 h, h, ,,, for m = 1, ..., L,, -1 (11) / = m (10) for m. Li, where the coefficients h, are the coefficients of the transmission channel estimated by correlation with the pilot channel, by the formula: = ap 1

OR

  rf snYn + fapn = where ap is the value of the pilot symbols carried by the pilot channel, sn the n th sample of the scrambling sequence such that ISpl = 1 and N the length of the received signal sequence on which the estimate. (12)

  2886789 13 N will typically be of the order of the length of the W-CDMA slot, or 2560 chips.

  The term QW is the noise power estimated by zz 2 o '= 6d where 6 is the power of the received signal and 6 is given by the formula z 1 ci 6d = Ihr P 1 = o Once the values of autocorrelation rm are estimated by one or other of the methods described, the value of the criterion J can be calculated by using formulas (5) or (6) according to the chosen formula.

  The step E20 of calculating the value J dependent on the autocorrelation is followed by a step E30 during which the processing means 50 compare this value J with a predetermined threshold S. The value of the threshold S will generally be chosen between 0 and 1, this value may however exceed 1.

  A value of 0.3 leads, for example, to performances close to those of the fixed LMMSE receiver while offering a significant savings in consumption when the LMMSE receiver offers only a small gain compared to the RAKE 60 receiver as described later. with reference to Figures 3 and 4.

  According to the invention, the comparison step E30 is followed by a step of selecting the equalizer 70 or the RAKE receiver 60 as a function of the result of this comparison.

  More precisely, the receiver 60 of the RAKE type is selected (step E40) if the value J dependent on the autocorrelation is lower than the predefined threshold S, and the equalizer 70 is selected (step E50) in the opposite case.

  For this purpose, the processing means 50 are adapted to position switches 81 and 82 at the input and output of the selected receiver. (13)

z (14).

  The step E40 or the step E50 of selecting the receiver is followed by a step E60 during which signals are processed with the selected receiver 60, 70.

  The processed samples may be in particular those of the slot used to select the receiver or those of the next slot.

  The description that has just been given has been made in the context of the CDMA where the choice is made between a receiver 60 of RAKE type and a receiver LMMSE 70.

  However, the same criterion can be applied in a non-CDMA environment to select between an equalizer and a simple filter adapted to the transmission channel.

  The selection method can be implemented in an identical manner, that is to say by calculating the criterion according to the received signal sampled at the chip rate, if the RAKE receiver processes a signal sampled at a rate greater than that of the chip rate, which is used in practice to take into account precisely the delays of the channel paths. Indeed, the performance difference between a RAKE-type receiver processing a chip-sampled signal and a RAKE-type receiver processing a sampled signal at a higher rate is small compared to the performance gain provided by an equalizer when this gain is significant. .

  The selection method can be implemented in the same way that the equalizer is performed synchronously or fractionally, that is to say by calculating the criterion according to the received signal sampled at the chip rate. Indeed, if an equalizer processing a sampled signal at the chip rate is likely to bring a gain compared to a receiver type RAKE, the gain provided by an equalizer processing a signal sampled at a higher rate will be at least equivalent.

  Alternatively, if the channel is too selectively frequency-selective, or if the signal-to-noise ratio is too low to allow a synchronous equalizer to bring a gain compared to a RAKE-type receiver, then the same will be true for a receiver. split equalizer.

  In other words, if the conditions are favorable (respectively, unfavorable) to the use of a synchronous equalizer, then they will be equally favorable for a fractional equalizer.

  Similar reasoning applies in the case of a terminal equipped with several receiving antennas. It is then sufficient to calculate the value J dependent on the autocorrelation of the samples obtained at the output of only one of the antennas: indeed, if the signal received on an antenna knows transmission conditions allowing the equalizer to bring a gain sufficient with respect to the receiver RAKE type, it will generally be the same for the signals received on the other antennas.

  The criterion thus calculated on a particular antenna therefore reflects the gain possibilities allowed by an equalizer exploiting the reception diversity compared to a receiver of RAKE type also exploiting the reception diversity.

  Those skilled in the art understand that, in the presence of several receiving antennas, the equalizer and the filter adapted to the channel may respectively integrate and be associated with a spatial cancellation cancellation treatment, for example of the "optimal combining" type. "for the adapted filter (RG Vaughan," On Optimum Combination at the Mobile ", IEEE Trans., Veh Technol., vol 37, n 4, Nov. 1988).

  The proposed selection criterion expressing the difference in structure between a filter adapted to the channel and an MMSE linear equalizer, it directly reflects the gain that can be observed between these two receivers.

  However, since the criterion expresses in this way the presence of a temporally structured interference in the received signal, it makes it possible to detect whether it is expedient to implement a receiver capable of combating this reception interference.

  In general, the equalizer can thus be other than an MMSE linear equalizer, such as a Maximum Likelihood Sequence Estimator (MLSE), a decision feedback equalizer or a turbo equalizer.

  In CDMA, the equalizer can also be a Multipath Interference Canceller (MPIC).

  When the equalizer is not the MMSE linear equalizer, the length of the equalizer entering the calculation of the criterion can be chosen equal to the length that would be chosen for the linear equalizer MMSE if the latter was used.

  However, when the equalizer is not the linear equalizer MMSE, the criterion no longer directly reflects the gain that can be observed between the equalizer and the filter adapted to the channel.

  We will now describe with reference to FIGS. 3 and 4 the results obtained by the selection method proposed in the context of an adaptive mono-antenna receiver of the RAKE 60 / LMMSE 70 type in a W-CDMA mobile communication system.

  The scenario considered reproduces an urban deployment of seven hexagonal macro cells: a host central cell surrounded by a ring of six adjacent interfering cells. The service in question is the transmission of voice at 12.2 kb / s. All base stations (BSs) transmit at full power (43 dBm) while the power allocated to the desired user by its BS, located in the center of the central cell, is set at 25 dBm. The transmission channel associated with each BS is the Rayleigh ITU Pedestrian B fading channel, which has 6 distinguishable paths whose maximum delay covers 15 chip durations. Note that the RAKE only takes into account the four most powerful paths. The mobile terminal moves from the center of the central cell to its edge, located 500 m from the BS. During the displacement, the nature of the dominant interference thus evolves from a strong intracellular interference structured by the multiple paths towards a strong weakly temporally weak extracellular interference.

  For each position of the mobile, it simulates the transmission of a large number of blocks of three slots, the transmission channels between the mobile and the different BS being drawn randomly and independently to each block and remaining fixed over the duration of the block. The selection of the receiver is performed at each three-slot block, based on the direct estimate of the autocorrelation of the received signal described above.

  FIG. 3 shows the performances obtained by the RAKE receiver, the LMMSE receiver and the RAKE / LMMSE adaptive receiver for several values of the threshold S as a function of the distance of the mobile to its base station. As expected, the lower the threshold, the more the adaptive receiver's performance is close to that of the fixed LMMSE receiver, since the LMMSE receiver is then more easily selected. On the other hand, a higher selection rate of the LMMSE receiver also results in increased receiver consumption. This confirms the influence of the choice of the threshold on the performance / consumption ratio of the RAKE / LMMSE adaptive receiver.

  Figure 4 details the selection percentage of the RAKE receiver for different positions of the mobile in the cell. Whatever the value of the threshold, it can be seen that the selection rate of the RAKE-type receiver increases as the mobile approaches the cell border, the value of the rate being controlled by the importance of the threshold. It should be noted in particular that even for S = 0.3, which offers the adaptive receiver very similar performance to that of a LMMSE receiver, the percentage of selection of the RAKE receiver exceeds 75% for distances greater than 400 m. represents a considerable economy of consumption compared to a non-adaptive LMMSE receiver.

  These results illustrate the relevance of the proposed criterion in order to select the most appropriate receiver between the RAKE receiver and the LMMSE receiver according to the radio environment, as well as the possibility of choosing the decision threshold according to the compromise. desired performance / consumption.

Claims (1)

18 Claims   1. Device (1) for receiving a signal carrying a digital information, said signal coming from a transmission channel, this device (1) comprising means (10, 20, 40) for obtaining samples of this signal an equalizer (70) and a filter adapted (60) to said channel, said receiving device (1) being characterized in that it comprises means (50) for calculating a value (J) dependent on the autocorrelation a sequence of samples, means (50) for comparing said value (J) with a predetermined threshold (S) and means (50, 81, 82) for selecting, in order to process signal samples, said equalizer (70) or said matched filter (60) depending on the result of said comparison.   2. Receiving device according to claim 1, characterized in that said value (J) dependent on the autocorrelation is of the form: J = aII3R-I112 where a and f3 are weighting coefficients, I the identity matrix and R a temporal autocorrelation matrix of said signal.   3. Reception device according to claim 2, characterized in that said weighting coefficients a and R are respectively equal to 1 / Lp and 1 / ro, where ro and Lp respectively represent the power of said signal and the length of said equalizer (70). ).   4. Receiving device according to claim 2 or 3, wherein said signal is transmitted by a base station, characterized in that parameters defining said autocorrelation matrix (R) depend on: estimated coefficients (h,) of transmission channel; an estimated power of the noise (a); and - a ratio (p) between the transmission power of a pilot signal transmitted by said base station and the total transmission power of this station.   5. A method of receiving a signal carrying digital information, said signal being derived from a transmission channel, this method including a step (E10) for obtaining samples of said signal and being characterized in that it comprises: a step (E20) of calculating a value (J) dependent on the autocorrelation of a series of samples; a step (E30) of comparing said value (J) with a predetermined threshold (S); a step (E40, E50) for selecting one of an equalizer (70) and a matched filter (60) for said channel as a function of the result of said comparison (E30); and - a step (E60) of processing samples of said signal, with the element (60, 70) selected.   6. Computer program on an information carrier, said program (PROG) being capable of being implemented in a receiving device (1), characterized in that it comprises instructions adapted to the implementation of a reception method according to claim 5, when said program is executed by this device.   An information carrier (52) readable by a receiving device, on which a computer program (PROG) is recorded which includes instructions for carrying out the steps of a reception method according to claim 5.