TRANSMISSION/RECEPTION METHODS AND MODULES FOR A MULTIPLE-CARRIER MULTIPLE-ANTENNA SYSTEM USING TRAINING SEQUENCES

Abstract

The present invention relates to a transmission method and device suitable for a system using Nt transmit antennas TXi and Nr receive antennas RXj. The symbols transmitted are inserted in a time-frequency frame that contains a training sequence. Each training sequence is constructed from replicas of a basic training sequence. At least one of the training sequences includes at least two replicas at least one of which is phase offset.

Description

The field of the present invention is telecommunications. Within that field, the invention relates more particularly to digital communication. Digital communication includes wireless radio communication and cable communication. The communication transmission medium is commonly called the transmission or propagation channel, originally with reference to an aerial channel and by extension with reference to any channel.

The invention relates to transmission and reception techniques specifically adapted to estimate transmission channels in a multi-antenna system of the MIMO (multiple-input multiple-output) or MISO (multiple-input single-output) type using multiple carriers and equalization in the frequency domain or the time domain. In a system with multiple transmit or receive antennas, there are as many transmission channels as there are transmit antenna/receive antenna pairs; these channels are sometimes called sub-channels. Estimation consists in estimating the impulse response of each of these channels. The invention applies to multiple-carrier multiple-antenna systems typically using at least two transmit antennas. However, the invention applies equally to a system using only one transmit antenna.

These systems use a frame of content, at the input of the transmitter, that consists of payload data symbols, i.e. symbols that code the information of an input signal, and at the output of the same device, that is divided in time and in frequency for transmission on a plurality of carriers. FIG. 1 is an illustration of an OFDM (orthogonal frequency division multiplex) frame with payload data symbols D in white and pilot symbols SP in black. A time-frequency frame determines the placement in time of the payload data symbols and the pilot symbols on the various carriers. Furthermore, the presence of a plurality of transmit antennas enables spatial diversity to be introduced by multiplexing the payload data between the antennas. Below in the present document the term “data” refers to payload data.

The invention can be applied to uplink communication (from a terminal to a base station) and to downlink communication (from a base station to a terminal).

One example of application of the invention is the field of fixed or mobile radio communication, notably for fourth generation and later systems typically referred to as B3G (Beyond 3^rd Generation) systems. These systems include MC-CDMA (multi-carrier code division multiple access) systems and OFDMA (orthogonal frequency division multiple access) downlink or uplink systems using an MIMO (multiple-input multiple-output) transmission scheme in which data to be transmitted is divided between time-frequency frames including pilot symbols and possibly null carriers. The invention applies in particular to any type of system using OFDM modulation, for example OFDMA or LP-OFDM systems, and to IFDMA systems.

Standard transmission methods include a modulation step. If differential modulation (non-coherent system) is not used, it is imperative for the receiver to estimate the propagation channel (coherent system) in order to be able to equalize the received signal and detect the bits transmitted. Differential modulation as applied to multiple-antenna systems is not at present considered a promising candidate for high bit rate communications systems. It doubles the noise level, which degrades performance by around 3 decibels (dB).

The channel estimation function is therefore particularly important for multi-antenna systems because the performance of these systems is directly linked to channel estimation in the receiver. In multiple-antenna systems, this function must estimate the various channels connecting a transmit antenna to a receive antenna independently of each other.

Several transmission techniques are already known that are suitable for transmission channel estimation in a multiple-antenna system including a plurality of transmit antennas. These techniques include those that rely on reference symbols, called pilot symbols, inserted into the time-frequency frame on transmission. These pilot symbols are known to the receiver and they enable it to estimate the transmission channels corresponding to each transmit antenna. Theoretically, the capacity of MIMO systems increases linearly with whichever is the lower: the number of transmit antennas and the number of receive antennas. In practice, the usable spectral efficiency falls as the number of antennas increases because of the necessary presence of pilot symbols in the frame.

Various techniques exist for inserting pilot symbols into the time-frequency frame transmitted by an antenna. The set of all pilot symbols for the same time-frequency frame forms a training sequence. In a multiple-antenna system there are as many training sequences as there are transmit antennas.

A known technique for inserting pilot symbols and shown in FIG. 2 transmits from each transmit antenna a training pulse, or more generally a training sequence, that is time offsetbetween antennas by a value Δt so that the receiver connected to each receive antenna can isolate the impulse responses of the various transmission channels in the time domain. For example, {tilde over (c)}(p),p=0, . . . , Np−1 is the pilot symbol transmitted on the p^th carrier frequency by the transmit antenna TX₁. Np is the number of pilot carriers. The pilot symbol transmitted on the same OFDM symbol by the antenna TX_i is therefore given by the equation:

${\tilde{c}\left(p\right)}_i=\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\left(i-1\right)\Delta t}{\mathrm{Np}}}$

where Δt is the time offset between the pulses transmitted by the various antennas. For the impulse responses of the various sub-channels not to interfere with each other, the condition L≦Δt must apply, where L is the maximum spread of the delays of the various sub-channels. Moreover, a condition for determining all impulse responses is Nt×L≦Np, where Nt is the number of transmit antennas TX_i. To meet this latter condition, the time offset Δt is chosen to satisfy the condition Nt×Δt≦Np.

The paper by M-S. Baek, H-J. Kook, M-J. Kim, Y-H. You, and H-S. Song, “Multi-Antenna Scheme for High Capacity Transmission in the Digital Audio Broadcasting”, IEEE Transactions on Broadcasting, Vol. 51, No. 4, December 2005 and the paper by I. Barhumi, G. Leus, and M. Moonen, “Optimal Training Design for MIMO OFDM Systems in Mobile Wireless Channels”, IEEE Transactions on Signal Processing, Vol. 51, No. 6, June 2003 rely on this technique. In the first paper, all the sub-carriers of an OFDM symbol are dedicated to channel estimation, which enables the receiver to recover the various impulse responses ahead of OFDM demodulation. In contrast, in the second paper, multiplexing the payload data symbols and the pilot symbols, with the pilot systems shared over one or more OFDM symbols to form a training sequence, implies that the operation of recovering the impulse responses on reception is effected after OFDM demodulation. This operation employs a matrix A constructed from the training sequence and a Fourier matrix of appropriate dimension. Estimating the coefficients of the various impulse responses entails multiplying the received signal demodulated by the pseudo-inverse matrix of this matrix A. The methods disclosed have the advantage of preserving the disposition of the pilot symbols of a single antenna-frame between the various transmit antennas, using the same set of sub-carriers for all pilot frequencies, and avoiding the mandatory presence of null pilot symbols. They therefore offer high spectral efficiency. Their advantage is that they estimate the channel for all modulated carriers.

An objective of the invention is to be even more efficient than known methods of transmitting time-frequency frames with inserted pilot symbols forming a training sequence when used in a system with at least one antenna, in order to improve performance.

To this end, the present invention consists in a transmission method, a transmission module, a channel estimation method, and a channel estimator module.

A transmission method according to the invention is suitable for a system with Nt transmit antennas TX₁ to TX_Nt and Nr receive antennas, where Nt and Nr are greater than or equal to 1, a transmit antenna being separated from a receive antenna by a transmission sub-channel, and is implemented by Nt multiple-carrier transmitters each including a multiplexing and modulation module using N_FFT orthogonal functions forming orthogonal symbols intended to be transmitted by the Nt transmit antennas. The method is noteworthy in that it includes the steps of:

• • determining a basic training sequence {tilde over (c)}(p) determined by the position of Np pilot symbols in a time-frequency frame;
  • determining

$\mathrm{Kn}=\left(\sum _{i=1}^{\mathrm{Nt}}K_i\right)$

replicas of the basic training sequence {tilde over (c)}(p) such that at least one replica is time offset from the basic sequence, with K_i≧1 and at least one K_i≧2, subject to the constraints that the extreme time offset between replicas must be less than the number Np of pilot carriers and that the minimum time offset between replicas of two training sequences is greater than or equal to the maximum spread of the delays of the sub-channels; and

• • determining a training sequence for each transmit antenna TX_i as the sum of the K_i replicas.

A transmission module according to the invention includes a multiplexing and modulation module using N_FFT orthogonal functions, suitable for a multiple-antenna system using Nt transmit antennas TX₁ to TX_Nt, Nr receive antennas, where Nt and Nr are greater than or equal to 1, and a time-frequency frame for each transmit antenna including pilot symbols and data symbols of a payload signal, the data symbols and the pilot symbols being frequency-modulated by the multiplexing and modulation module to form orthogonal symbols that are transmitted by a transmit antenna in the form of a multiple-carrier signal using N_FFT carriers and using Np pilot carriers. This module is noteworthy in that it includes:

• • a module for determining a basic training sequence determined by the position of Np pilot symbols in the time-frequency frame associated with a particular transmit antenna;
  • a module for determining Kn, where

$\mathrm{Kn}=\left(\sum _{i=1}^{\mathrm{Nt}}K_i\right),$

replicas of the basic training sequence such that at least one replica is time offset from the basic sequence, with K_i≧1 and at least one K_i≧2, subject to the constraints that the extreme time difference between replicas is less than the number Np of pilot carriers and that the minimum time offset between replicas of two training sequences is greater than or equal to the maximum spread of the delays of the sub-channels connecting a transmit antenna to a receive antenna; and

• • a module for determining a training sequence for each transmit antenna as the sum of K_i replicas.

A method according to the invention for estimating transmission sub-channels in a multiple-antenna system using Nt transmit antennas, where Nt is greater than or equal to 1, at least one receive antenna, and a time-frequency frame for each transmit antenna including Np pilot symbols forming a training sequence and data symbols of a payload signal, each of the Nt training sequences being the sum of K_i replicas of a basic training sequence, either time offset or not, such that the values K_i are greater than or equal to 1 and at least one value K_i is greater than or equal to 2, being known to the receiver and enabling the receiver to estimate an impulse response corresponding to the transmission channel separating a transmit antenna from a receive antenna concerned, the data symbols and the pilot symbols being frequency-modulated by a multiplexing and modulation module using N_FFT orthogonal functions to form orthogonal symbols that are transmitted by the transmit antennas in the form of a multiple-carrier signal with N_FFT carriers. This method is noteworthy in that it includes:

• • a step of computing a matrix A constructed in the form of blocks from the training sequences and from the Fourier matrix of dimensions N_FFT×N_FFT, the number of blocks of which is equal to the product of the number of orthogonal symbols of the basic training sequence by the sum of the values K_i, a block being determined by the product of a diagonal matrix formed of the pilot symbols contained in an orthogonal symbol of the training sequence associated with the block multiplied by the Fourier matrix, andin that it includes, for a receive antenna under consideration:
  • a step of computing K_i estimates of at least one of the Nt impulse responses in the time domain by multiplying by the pseudo-inverse matrix of the matrix A Np pilot symbols extracted from a frequency-domain signal obtained after demodulation of a time-domain signal received by the receive antenna concerned by means of a Fourier transform of size N_FFT; and
  • a step of computing an average over the K_i estimates of the impulse response concerned.

A module according to the invention for estimating transmission channels in a multiple-antenna system using Nt transmit antennas, where Nt is greater than or equal to 1, at least one receive antenna and a time-frequency frame for each transmit antenna including pilot symbols forming a training sequence and data symbols of a payload signal, the Nt training sequences being known to the receiver and enabling the receiver to estimate Nt impulse responses corresponding to the Nt transmission channels respectively separating one of the transmit antennas from the receive antenna concerned, the data symbols and the pilot symbols being frequency-modulated by a multiplexing and modulation module using N_FFT orthogonal functions to form orthogonal symbols that are transmitted by the transmit antennas in the form of a multiple-carrier signal using N_FFT carriers with Np pilot carriers. This module is noteworthy in that it includes:

• • means for computing a matrix A constructed in the form of blocks from the training sequences and from the Fourier matrix of dimensions N_FFT×N_FFT, the number of blocks of which is equal to the product of the number of orthogonal symbols of the basic training sequence by the sum of the values K_i, a block being determined by the product of a diagonal matrix formed of the pilot symbols contained in an orthogonal symbol of the training sequence associated with the block multiplied by the Fourier matrix, andin that it includes, for a receive antenna under consideration:
  • means for computing K_i estimates of at least one of the Nt impulse responses in the time domain by multiplying by the pseudo-inverse matrix of the matrix A Np pilot symbols extracted from a frequency-domain signal obtained after demodulation of a time-domain signal received by the receive antenna concerned by means of an FFT of size N_FFT; and
  • means for computing the average of the K_i estimates of the impulse response concerned.

The transmission method, the estimation method, and the associated modules and devices of the invention solve the stated problem. Transmitting at least one training sequence including at least two replicas of a basic training sequence at least one of which is time offset produces at least two estimates of the same impulse response and so an average of those estimates can be computed. This improves the estimate of the impulse response of a channel.

In one particular embodiment, a guard interval is added before transmitting an orthogonal symbol and the constraint on the minimum value of the time offset is complied with by choosing a minimum time offset equal to the value of the guard interval.

In one particular embodiment, the replicas are advantageously determined in the frequency domain and the time offsets for the replicas are obtained by inserting a phase offset that is easy to compute. In a first embodiment, the phase offset values are chosen in a particular order, which is known to at least one receiver associated with a receive antenna of the system. In a second embodiment, the phase offset is a function of a carrier index.

In one particular embodiment, the basic training sequence is spread over a plurality of orthogonal symbols.

In one particular embodiment, each of the impulse response estimates is computed from the Np pilot carriers over a time interval Δt. This two-fold limitation advantageously reduces the computations required and therefore reduces the computation power necessary to implement the method.

In one particular embodiment, computation of the impulse responses is repeated for each receive antenna concerned of the multiple-antenna system. This embodiment is suitable for systems including a plurality of receive antennas.

The invention consists further in a transmitter for a multiple-antenna system. The transmitter includes at least one transmission module as defined above.

The invention consists further in a receiver for a multiple-antenna system. The receiver includes at least one transmission channel estimator module as defined above.

The invention consists further in a multiple-antenna system including at least one receiver and/or one transmitter as defined above.

In a preferred embodiment, the multiple-antenna system typically has at least two transmit antennas and is a MIMO system or a MISO system. In the remainder of this document, a MIMO system is to be understood as a system with a plurality of transmit antennas, regardless of the number of receive antennas (Nr≧1).

In a preferred embodiment, the steps of the method are determined by instructions of a transmission, respectively estimation, program in an electronic circuit such as a microchip that can be included in an electronic device such as a transmitter, respectively, a receiver. The transmission, respectively estimation, method of the invention can equally well be executed when this program is loaded into a computation unit such as a processor or the like the operation of which is then controlled by the execution of the program.

Consequently, the invention applies equally to a computer program, notably a computer program on or in an information medium, adapted to implement the invention. This program can use any programming language and take the form of source code, object code or a code intermediate between source code and object code, such as a partially-compiled form, or any other form desirable for implementing a method of the invention.

The information medium can be any entity or device capable of storing the program. For example, the medium can include storage means such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or magnetic storage means, for example a floppy disk or a hard disk.

Alternatively, the information medium can be an integrated circuit incorporating the program and adapted to execute the method in question or to be used in its execution.

Moreover, the transmission or estimation program can be translated into a transmissible form such as an electrical or optical signal, which can be routed via an electrical or optical cable, by radio or by other means. The program of the invention can in particular be downloaded over an Internet-type network or the like.

Other features and advantages of the invention become apparent in the course of the following description given with reference to the appended figures provided by way of non-limiting example:

FIG. 1 is an illustration of an OFDM (orthogonal frequency division multiplex) frame;

FIG. 2 is an illustration of transmission by a prior art system with four antennas for transmitting training sequences, each consisting of a time pulse that is time offset between antennas;

FIG. 3 is an illustration of transmission by a system of the invention using one antenna for transmitting a training sequence consisting of two replicas, one of which is time offset, of a basic training sequence equal to one time pulse;

FIG. 4 is an illustration of transmission by a system of the invention using four antennas (Nt=4) for transmitting a training sequence consisting of replicas of a basic training sequence equal to one time pulse, the sequence for the antenna TX₁ consisting of two replicas, one of which is time offset;

FIG. 5 is a flowchart of a transmission method of the invention;

FIG. 6 is a flowchart of an estimation method of the invention;

FIG. 7 is a diagram of a particular transmission system of the invention implementing a method of the invention; and

FIG. 8 is a graphical representation of simulation results.

FIG. 3 is an illustration of transmission in accordance with the invention of a training sequence consisting of two replicas of a time pulse, one of which is phase offset.

The situation shown is one with one transmit antenna and one receive antenna for transmitting time-frequency frames using training sequences. According to the invention, the training sequence associated with the transmit antenna consists of at least two replicas of a basic training sequence, that are time offset relative to each other. In this illustration, the basic training sequence is a time pulse. Each replica is used to determine an estimate of the response of the transmission channel between the transmit antenna and the receive antenna. According to the invention, the impulse response of the channel is then determined as the average of those two estimates.

FIG. 4 is an illustration of transmission by a system of the invention using four antennas to transmit training sequences consisting of replicas of a basic training sequence equal to one time pulse, the training sequence for the antenna TX₁ consisting of two replicas, one of which is time offset.

According to the invention, in a multiple-antenna system using a plurality of transmit antennas, the training sequence of at least one transmit antenna consists of at least two replicas of a basic training sequence that are time offset relative to each other. In this illustration, the basic training sequence is a time pulse and the training sequence for the antenna TX₁ consists of two replicas that are time offset relative to each other. Each of these replicas is used to determine an estimate of the response of the transmission channel between the transmit antenna TX₁ and the receive antenna RX_j concerned. According to the invention, the response of the transmission channel between the transmit antenna TX₁ and the receive antenna RX_j is then determined as the average of these two estimates.

FIG. 5 is a flowchart of a transmission method 1 of the invention.

A transmission method 1 of the invention is suitable for a system using Nt transmit antennas and Nr receive antennas, where Nt and Nr are both greater than or equal to 1. A transmit antenna is separated from a receive antenna by a transmission sub-channel. The method is implemented by Nt multi-carrier transmitters each including a multiplexing and modulation module using N_FFT orthogonal functions forming orthogonal symbols to be transmitted by the Nt transmit antennas TX₁ to TX_Nt. The system uses for each transmit antenna one time-frequency frame containing pilot symbols and payload data symbols. The data symbols and the pilot symbols are frequency-modulated by the multiplexing and modulation modules to form orthogonal symbols that are transmitted by a transmit antenna in the form of a multi-carrier signal using N_FFT carriers with Np pilot carriers.

A first step 2 of the method 1 determines a basic training sequence {tilde over (c)}(p), where p is the index of the pilot carrier.

This sequence is typically determined in the frequency domain by determining the Np pilot carriers and sharing them over one or more orthogonal symbols. These orthogonal symbols are OFDM symbols if the transmit orthogonal modulation is OFDM modulation. If the basic training sequence is shared over a plurality of orthogonal symbols, the pilot carriers of the orthogonal symbols can be identical, different, or partially overlapping. For example, one basic training sequence is one time pulse as illustrated by FIGS. 2, 3, and 4, and such a sequence can be written in the form {tilde over (c)}(p)=1 for 1≦p≦Np.

A second step 3 of the method determines under constraint a number Kn replicas of index n of the basic training sequence such that at least one replica is time offset from the basic training sequence, where:

$\begin{array}{cc}\mathrm{Kn}=\left(\sum _{i=1}^{\mathrm{Nt}}K_i\right)& \left(1\right)\end{array}$

The values K_i are equal to or greater than 1 and at least one value K_i is greater than or equal to 2. The constrained determination of replicas, whether time offset or not, is typically effected in the frequency domain by introducing a phase offset of value φ_n by multiplying the basic training sequence with a phase-rotation factor conforming to the expression:

$\begin{array}{cc}\tilde{c}\left(p\right)\times {}^{-{\mathrm{j\varphi }}_n}=\tilde{c}\left(p\right)\times {}^{-\mathrm{j2\pi }p\frac{\left(n-1\right)}{\mathrm{Np}}\Delta t}& \left(2\right)\end{array}$

where 1≦n≦Kn and with the following constraints satisfied.

A first constraint is that the extreme time offset between the various replicas of the basic training sequence determined beforehand is less than the number Np of pilot carriers, in other words:

Kn×Δt≦Np  (3)

A second constraint is that the minimum time offset between two replicas of two training sequences associated with two transmit antennas i and j, {tilde over (c)}(p)_i and {tilde over (c)}(p)_j, with i≠j, is greater than or equal to the maximum spread of the delays of the various channels:

L≦Δt  (4)

where L is the maximum spread of the delays of the various sub-channels.

Moreover, a condition for determining the set of all impulse responses is

Nt×L≦Np  (5)

This condition is automatically satisfied if the time offset Δt chosen satisfies the above two constraints.

In a third step 4, the method determines the training sequence for each transmit antenna TX_i.

For at least one transmit antenna TX_i the training sequence is the sum of K_i replicas of the basic training sequence, where K_i≧2, with at least one replica from the basic sequence time offset.

In a preferred embodiment of the invention, the training sequences are given by the following expression, on condition that constraints (3), (4), and (5) are satisfied:

$\begin{array}{cc}{\tilde{c}\left(p\right)}_i=\tilde{c}\left(p\right)\sum _{k=1}^{K_i}{}^{-\mathrm{j2\pi }p\frac{\left(i+\left(k-1\right)\mathrm{Nt}-1\right)}{\mathrm{Np}}\Delta t}& \left(6\right)\end{array}$

In the particular situation where the value K_i≧2 is equal to 2, the training sequence of at least one antenna TX_i corresponds to computing the expression:

$\begin{array}{cc}{\tilde{c}\left(p\right)}_i=\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\left(i-1\right)\Delta t}{\mathrm{Np}}}+\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\left(i+\mathrm{Nt}-1\right)\Delta t}{\mathrm{Np}}}& \left(7\right)\end{array}$

For each of the other antennas TX_i, the training sequence includes at least one replica of the basic training sequence phase offset by φ_i, with φ_i≧0, which corresponds to computing the expression:

$\begin{array}{cc}{\tilde{c}\left(p\right)}_i=\tilde{c}\left(p\right)\sum _{k=1}^{K_i}{}^{-{\mathrm{j\varphi }}_{i,k}}=\tilde{c}\left(p\right)\sum _{k=1}^{K_i}{}^{-\mathrm{j2\pi }p\frac{\left(i+\left(k-1\right)Nt-1\right)}{\mathrm{Np}}\Delta t}& \left(8\right)\end{array}$

with K_i≧1 being the number of replicas constituting the training sequence of the antenna TX_i and subject to the constraints (3), (4), and (5) being satisfied.

In one particular embodiment, the training sequence associated with each transmit antenna is the sum of the K replicas of the basic training sequence. For any transmit antenna TX_i, the training sequence is preferably given by the following expression, on condition that the constraints (3), (4), and (5) are satisfied:

$\begin{array}{cc}{\tilde{c}\left(p\right)}_i=\tilde{c}\left(p\right)\sum _{k=1}^K{}^{-\mathrm{j2\pi }p\frac{\left(i+\left(k-1\right)\mathrm{Nt}-1\right)}{\mathrm{Np}}\Delta t}& \left(9\right)\end{array}$

Here Kn=Nt×K.

The next example is given by way of illustration. The parameters of the system take the following values: N_FFT=N_mod=Np=8, Δt=2. N_FFT, N_mod, and Np are respectively the size of the FFT, the number of modulated carriers, and the number of pilot carriers. The number g of orthogonal symbols over which the training sequence is spread is taken as equal to 1. The system is a multiple-antenna system with two transmit antennas (Nt=2). Each training sequence is constructed from K₁=K₂=K=2 replicas. This yields:

$\hspace{1em}\{\begin{array}{c}{\tilde{c}\left(p\right)}_1=\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(1-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(1+\mathrm{Nt}-1\right)\times 2}{8}}=\tilde{c}\left(p\right)+\tilde{c}\left(p\right){}^{-\mathrm{j\pi }p}\\ {\tilde{c}\left(p\right)}_2=\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(2-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(2+\mathrm{Nt}-1\right)\times 2}{8}}=\tilde{c}\left(p\right){}^{-j\frac{\pi }{2}p}+\tilde{c}\left(p\right){}^{-j\frac{3\pi }{2}p}\end{array}$

where {tilde over (c)}(p) is the pilot symbol from the basic training sequence transmitted at the carrier frequency with index p. The constraint Kn×Δt≦Np is satisfied since (2+2)×2=8≦8. The pilot symbol {tilde over (c)}(p) is arbitrarily chosen to be equal to 1 for all pilot carriers. The pattern of the OFDM symbol containing the basic training sequence is represented in Table 1 in Appendix A. The first column of Table 1 represents the frequency indices and the second column the associated data type (P=pilot symbol).

This finally yields for the training sequences of the two antennas:

$\hspace{1em}\{\begin{array}{c}{\tilde{c}\left(p\right)}_1=1+{}^{-\mathrm{j\pi }p}\\ {\tilde{c}\left(p\right)}_2={}^{-j\frac{\pi }{2}p}+{}^{-j\frac{3\pi }{2}p}\end{array}$

FIG. 6 is a flowchart of an estimation method 10 of the invention.

The estimation method 10 of the invention estimates transmission channels in a multiple-antenna system using at least one transmit antenna and one receive antenna, Nt≧1 and Nr≧1. A transmission channel separates a transmit antenna TX_i from a receive antenna RX_j. Each transmit antenna of a multiple-antenna system transmits one time-frequency frame. A time-frequency frame determines the placement in time of the payload data symbols and the pilot symbols on the various carriers. The time-frequency frame can further include null symbols, but as these are not involved in the estimation process they are not referred to in the description. Nor does the description cover guard carriers, because they are not involved in the estimation process either. The pilot symbols for the same transmit antenna form a training sequence. The estimation method of the invention is suitable for signals transmitted by a transmission method conforming to a first embodiment of the invention but can be used for other types of signals, although they would be transmitted with lower efficiency.

The description that follows of the estimation method of the invention assumes that the signals are transmitted by a transmission method conforming to a first embodiment of the invention.

The replicas constituting the Nt training sequences are time offset relative to each other by at least one time slot Δt. The training sequences are known to the receiver and enable it to estimate Nt impulse responses that correspond to the Nt transmission channels. For a given transmit antenna, the data symbols and the pilot symbols are frequency-modulated by a multiplexing and modulation module using N_FFT orthogonal functions, commonly called an orthogonal multiplexer, to form orthogonal symbols that are transmitted by the transmit antenna connected to the orthogonal multiplexer in the form of a multiple-carrier signal using N_FFT carriers and using Np pilot carriers. Depending on the time-frequency frame concerned, the pilot symbols can be divided over one or more orthogonal symbols. In one particular embodiment the modulation is OFDM modulation.

An estimation method of the invention is applied in the receiver after demodulation of a time signal received by the receive antenna RX_j concerned and uses an FFT of a certain size to obtain a frequency-domain signal R_j(n) and extract Np pilot symbols from it. In one particular embodiment, the demodulation is OFDM demodulation. The size of the FFT is typically determined as a function of the size N_FFT of the FFT used for transmission. These sizes are typically chosen to be identical. On transmission the FFT is of the inverse type whereas on reception the FFT is of the direct type.

At time n, the signal from the receive antenna RX_j can be expressed in the frequency domain, i.e. after suppressing the guard interval and after OFDM demodulation, in the form of a column vector of dimension N_FFT:

$\begin{array}{cc}{R}_j\left(n\right)=\sum _{i=1}^{\mathrm{Nt}}\mathrm{diag}\left\{X_i\left(n\right)\right\}{\mathrm{Fh}}_{j,i}\left(n\right)+{\Xi }_j\left(n\right)& \left(10\right)\end{array}$

where X_i(n) is a column vector of dimension N_FFT, the OFDM symbol transmitted at the time n by the antenna TX_i;

• • F is the Fourier matrix of size N_FFT×N_FFT;
  • h_j,i is the column vector representing the L samples of the frequency response of the sub-channel connecting the transmit antenna TX_i to the receive antenna RX_j; andΞ_j(n) is the column vector of dimension N_FFT representing the Fourier transform of additive Gaussian white noise.

To simplify the computation, which is advantageous, the estimation method uses the demodulated signal according to equation (10) only over the duration Δt:

$\begin{array}{cc}{R}_j\left(n\right)=\sum _{i=1}^{Nt}\mathrm{diag}\left\{X_i\left(n\right)\right\}F^{\prime }h_{j,i}^{\prime }\left(n\right)+{\Xi }_j\left(n\right)& \left(11\right)\end{array}$

where F′ is the matrix consisting of the Δt first columns of the Fourier matrix F of size N_FFT×N_FFT and h_j,i′ is a column vector of length Δt of which the row elements beyond L are at zero:

h_j,i′=[h_j,i^T,0_Δt-L]^T  (12)

The notation diag{x} denotes a diagonal matrix having the column vector x on its diagonal:

$\begin{array}{cc}\mathrm{diag}\left[\begin{array}{c}a\\ b\\ c\end{array}\right]=\left[\begin{array}{ccc}a& 0& 0\\ 0& b& 0\\ 0& 0& c\end{array}\right]& \left(13\right)\end{array}$

The Fourier matrix F is square, with dimensions N_FFT×N_FFT, and of the following form:

$\begin{array}{cc}F=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}\left[\begin{array}{ccccc}1& 1& 1& \dots & 1\\ 1& w_{N_{\mathrm{FFT}}}& w_{N_{\mathrm{FFT}}}^2& \dots & w_{N_{\mathrm{FFT}}}^{N_{\mathrm{FFT}}-1}\\ ⋮& ⋮& ⋮& \dots & ⋮\\ 1& w_{N_{\mathrm{FFT}}}^{N_{\mathrm{FFT}}-1}& w_{N_{\mathrm{FFT}}}^{2\left(N_{\mathrm{FFT}}-1\right)}& \dots & w_{N_{\mathrm{FFT}}}^{\left(N_{\mathrm{FFT}}-1\right)\left(N_{\mathrm{FFT}}-1\right)}\end{array}\right]\mathrm{where}w_{N_{\mathrm{FFT}}}={}^{-j\frac{2\pi }{N_{\mathrm{FFT}}}}.& \left(14\right)\end{array}$

The matrix F′ is deduced directly from the matrix F:

$\begin{array}{cc}{F}^{\prime }=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}\left[\begin{array}{ccccc}1& 1& 1& \dots & 1\\ 1& w_{N_{\mathrm{FFT}}}& w_{N_{\mathrm{FFT}}}^2& \dots & w_{N_{\mathrm{FFT}}}^{\Delta t-1}\\ ⋮& ⋮& ⋮& \dots & ⋮\\ 1& w_{N_{\mathrm{FFT}}}^{N_{\mathrm{FFT}}-1}& w_{N_{\mathrm{FFT}}}^{2\left(N_{\mathrm{FFT}}-1\right)}& \dots & w_{N_{\mathrm{FFT}}}^{\left(\Delta t-1\right)\left(\Delta t-1\right)}\end{array}\right]& \left(15\right)\end{array}$

The estimation method 10 of the invention includes a step 11 of computing a matrix A and, for a receive antenna RX_j, a step 12 of computing in the time domain estimates of the Nt impulse responses of the Nt channels respectively separating a transmit antenna TX_i from the receive antenna RX_j concerned. The method further includes a step 13 of computing the average of the impulse response estimates.

To obtain the Nr×Nt channels, steps 12 and 13 must be repeated for each of the receive antennas RX_j.

The computation step 11 computes a matrix A constructed in the form of blocks from training sequences and the Fourier matrix with dimensions N_FFT×N_FFT. The number of blocks is equal to the product of the number of orthogonal symbols of the basic training sequence by the sum of the values K_i. Each block is associated with a transmit antenna and an orthogonal symbol over which the training sequence is spread. A block is determined by the product of the diagonal matrix formed by the pilot symbols linked to the block and the Fourier matrix. To limit the computation, which is advantageous, the Fourier matrix is limited to the pilot carriers and to Δt time samples, because Δt is greater than or equal to the maximum spread of the temporal response of the channel.

Each OFDM symbol transmitted at the time n by the transmit antenna TX_i, X_i(n) can be broken down into a vector containing the data symbols and a vector containing the pilot symbols:

X _i(n)=S_i(n)+B_i(n)  (16)

where:

S_i(n) is the payload data symbol vector of dimension N_FFT and B_i(n) is the pilot symbol vector of the same dimension.

The equation (11) can therefore be expressed in the form:

$\begin{array}{cc}{R}_j\left(n\right)=\sum _{i=1}^{Nt}\mathrm{diag}\left\{S_i\left(n\right)+B_i\left(n\right)\right\}F^{\prime }h_{j,i}^{\prime }\left(n\right)+{\Xi }_j\left(n\right)& \left(17\right)\end{array}$

Knowing that Δt≧L, it is possible to impose Δt=L. With this condition, the accumulation of the received vectors corresponding to the g orthogonal symbols transmitted by each transmit antenna containing pilot symbols constituting a training sequence can be expressed in the form:

$\begin{array}{cc}\begin{array}{c}{R}_j={\left[{R_j\left(0\right)}^T\dots {R_j\left(g\text{-}1\right)}^T\right]}^T\\ ={\mathrm{Th}}_j^{\prime }+{\mathrm{Ah}}_j^{\prime }+{\Xi }_j\end{array}& \left(18\right)\end{array}$

where:

R_j is a column vector of dimension N_FFT×g, T is a matrix of dimensions (N_FFT×g)×(Kn×Δt) containing the payload data symbols during the training sequences of the Nt transmit antennas;

$\begin{array}{cc}T=\left[\begin{array}{ccc}{S}_{{1}_{\mathrm{diag}}}\left(0\right)F^{\prime }& \dots & S_{{\mathrm{Kn}}_{\mathrm{diag}}}\left(0\right)F^{\prime }\\ ⋮& & ⋮\\ S_{{1}_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }& \dots & S_{{\mathrm{Kn}}_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }\end{array}\right]& \left(19\right)\end{array}$

A is a matrix of dimensions (N_FFT×g)×(Kn×Δt) containing the pilot symbols of the training sequences of the Nt transmit antennas;

$\begin{array}{cc}A=\left[\begin{array}{ccc}{B}_{{1}_{\mathrm{diag}}}\left(0\right)F^{\prime }& \dots & B_{{\mathrm{Kn}}_{\mathrm{diag}}}\left(0\right)F^{\prime }\\ ⋮& & ⋮\\ B_{{1}_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }& \dots & B_{{\mathrm{Kn}}_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }\end{array}\right]& \left(20\right)\end{array}$

h_j′ is the column vector of dimension Kn×Δt containing the Kn expressions of the estimates of the Nt impulse responses;

h_j′=[h_j,1^T, . . . , h_j,Kn^T]^T  (21)

Ξ_j is the column vector of dimension N_FFT×g containing g representations of the Fourier transform of additive Gaussian white noise.

The matrix A determined during the step 11 is therefore expressed in the form:

$\begin{array}{cc}A=\hspace{1em}\left[\begin{array}{cccccc}{B}_{1_{\mathrm{diag}}}\left(0\right)F^{\prime }& \dots & B_{{\mathrm{Nt}}_{\mathrm{diag}}}\left(0\right)F^{\prime }& B_{\mathrm{Nt}+1_{\mathrm{diag}}}\left(0\right)F^{\prime }& \dots & B_{\sum _{i=1}^{\mathrm{Nt}}K_i*N_{t_{\mathrm{diag}}}}\left(0\right)F^{\prime }\\ ⋮& & ⋮& ⋮& & ⋮\\ B_{{1}_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }& \dots & B_{{\mathrm{Nt}}_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }& B_{\mathrm{Nt}+1_{\mathrm{diag}}}\left(g\text{-}1\right)F^{\prime }& \dots & B_{\sum _{i=1}^{\mathrm{Nt}}K_i*N_{t_{\mathrm{diag}}}}\left(g\text{-}1\right)F^{\prime }\end{array}\right]& \left(22\right)\end{array}$

The step 12 of computing the Kn estimates of the Nt impulse responses in the time domain multiplies the vector R_j, the expression for which in the frequency domain at the output of the demodulation corresponds to the Np pilot carriers, by the pseudo-inverse matrix of the matrix A using the Barhumi technique:

ĥ _j =A ⁺ R _j  (23)

in which the symbol ⁺ signifies the following “pseudo-inverse” operation:

A ⁺=(A^HA)⁻¹A^H  (24)

To eliminate interference terms caused by the data present in the OFDM symbol, it is necessary to have the product of A⁺ by T equal to a null matrix, which is made possible by imposing disjoint sets of pilot symbols and data. Moreover, the computation of the impulse responses can be simplified by considering in the equations only the frequencies dedicated to channel estimation, i.e. by limiting the computations to the pilot carriers:

ĥ _j =Ã ⁺ {tilde over (R)} _j  (25)

where:

{tilde over (R)}_j is the matrix corresponding to R_j considering only the pilot carriers, and where:

the expression for the matrix à is as follows:

$\begin{array}{cc}\tilde{A}=\left[\begin{array}{ccc}{\tilde{B}}_{1_{\mathrm{diag}}}\left(0\right){\tilde{F}}^{\prime }& \dots & {\tilde{B}}_{{\mathrm{Kn}}_{\mathrm{diag}}}\left(0\right){\tilde{F}}^{\prime }\\ ⋮& & ⋮\\ {\tilde{B}}_{1_{\mathrm{diag}}}\left(g\text{-}1\right){\tilde{F}}^{\prime }& \dots & {\tilde{B}}_{{\mathrm{Kn}}_{\mathrm{diag}}}\left(g\text{-}1\right){\tilde{F}}^{\prime }\end{array}\right]& \left(26\right)\end{array}$

or:

$\begin{array}{cc}\tilde{A}=\hspace{1em}\left[\begin{array}{cccccc}{\tilde{B}}_{{1}_{\mathrm{diag}}}\left(0\right){\tilde{F}}^{\prime }& \dots & {\tilde{B}}_{{\mathrm{Nt}}_{\mathrm{diag}}}\left(0\right){\tilde{F}}^{\prime }& {\tilde{B}}_{\mathrm{Nt}+1_{\mathrm{diag}}}\left(0\right){\tilde{F}}^{\prime }& \dots & {\tilde{B}}_{\sum _{i=1}^{\mathrm{Nt}}K_i*N_{t_{\mathrm{diag}}}}\left(0\right){\tilde{F}}^{\prime }\\ ⋮& & ⋮& ⋮& & ⋮\\ {\tilde{B}}_{{1}_{\mathrm{diag}}}\left(g\text{-}1\right){\tilde{F}}^{\prime }& \dots & {\tilde{B}}_{{\mathrm{Nt}}_{\mathrm{diag}}}\left(g\text{-}1\right){\tilde{F}}^{\prime }& {\tilde{B}}_{\mathrm{Nt}+1_{\mathrm{diag}}}\left(g\text{-}1\right){\tilde{F}}^{\prime }& \dots & {\tilde{B}}_{\sum _{i=1}^{\mathrm{Nt}}K_i*N_{t_{\mathrm{diag}}}}\left(g\text{-}1\right){\tilde{F}}^{\prime }\end{array}\right]& \left(27\right)\end{array}$

The vector expression of the first estimate of the impulse response ĥ_j,i of the channel separating the particular transmit antenna i, for i taking a value between 1 and Nt, from the particular receive antenna j, for j taking a value between 1 and Nr, is obtained by effecting a selection of the Np successive samples of the column vector ĥ_j computed for the receive antenna j starting with the sample (i−1)Δt. For the second estimate of the impulse response of the channel ĥ_j,i, selection starts from the sample Nt×Δt+(i−1)Δt=(Nt+i−1)Δt. For the n^th estimate of the impulse response of the channel ĥ_j,i, selection starts from the sample (n×Nt+i)Δt.

The step 13 computes the average in the time domain of the K_i impulse response estimates. Alternatively, this average is determined in the frequency domain. The step is preferably repeated for all the channels for which K_i is strictly greater than 1.

In a method conforming to one particular embodiment of the invention, the computation of the Nt×Nr impulse responses described above for the Np pilot carriers is complemented by interpolation that can be linear in time or in frequency (one dimensional (1D) interpolation) to obtain the coefficients of each channel for all carriers modulated with payload data. Interpolation can be of an order greater than 1. Interpolation can be effected in the frequency domain by multiplying the Nt vectors of Δt points by a matrix of size N_FFT×Δt extracted from a Fourier matrix of size N_FFT×N_FFT.

FIG. 7 is a diagram of a particular transmission system for implementing a method of the invention.

The transmission system SY includes a multiple-carrier transmitter device EM and a receiver device RE. The transmitter device EM is connected to Nt transmit antennas TX₁ to TX_Nt, where Nt≧2. The receiver device RE is connected to Nr receive antennas RX₁ to RX_Nr, where Nr≧1.

In the example shown, the transmitter EM includes a channel coding module CdC, a bit interleaver EB, a binary symbol coding module CBS, a space-time coding module CET, and as many OFDM multiplexers MX as there are transmit antennas.

The channel coding module CdC codes input source data corresponding to one or more signals, typically telecommunication signals, to provide at the output coded data dc, for example data coded using a convolutional code.

The bit interleaver interleaves the bits of the coded data in accordance with a particular interleaving law to provide interleaved coded data dce.

The binary symbol coder module CBS converts the interleaved coded data dce into complex data symbols sc, for example by BPSK, QPSK or 16QAM modulation.

The space-time coding module CET determines from the complex data symbols sc a two-dimensional matrix of data symbols sd, for example using Alamouti-type coding, and generates pilot symbols.

Each OFDM multiplexer modulates data symbols sd and pilot symbols sp inserted at the input of the multiplexer by sinusoidal sub-carriers whose orthogonal functions are the conjugate Fourier components of an inverse Fourier transform of size N_FFT corresponding to the number of carriers of an OFDM multiplex. The N_FFT carriers include N_mod modulated data carriers and Np pilot carriers. The set of N_FFT carriers is called an OFDM symbol. The OFDM symbol can include null carriers and unmodulated guard carriers. The output of an OFDM multiplexer constitutes the OFDM signal in the time domain with no guard interval.

The transmitter device inserts a guard interval into the OFDM signal before transmission.

The signals transmitted include training sequences. For each transmit antenna concerned, a training sequence includes at least one replica of a basic training sequence, either phase offset or not. This basic training sequence is determined by the positioning of Np pilot symbols contained in one or more OFDM symbols, the number of which OFDM symbols is g. The basic training sequence is therefore determined by the number g, the amplitude of the pilot symbols, and the pilot carriers. Between two training sequences associated with two transmit antennas, the replicas are time offset relative to each other by a minimum of one interval Δt least equal to the maximum length L of the transmission channels, Δt≧L, so that the impulse responses of the various channels do not interfere with each other. The length of a channel is the transmission duration of the channel or the transmission delay introduced by the channel.

Throughout this document, the interval Δt must be understood as being a plurality of time slots separating samples. The time offset is typically applied before modulation, for example OFDM modulation, by phase offsetting the replicas in the frequency domain.

The basic training sequence is expressed in the form of a series of pilot symbols: {tilde over (c)}(p),p=0, . . . , Np−1 where {tilde over (c)}(p) is the pilot symbol transmitted at the carrier frequency of index p. This basic training sequence is spread over g orthogonal symbols of a time-frequency frame. Moreover, the condition for determining the set of impulse responses is:

Kn×Δt≦Np

where:

$\mathrm{Kn}=\left(\sum _{i=1}^{\mathrm{Nt}}K_i\right)$

For example, g is chosen as equal to 1, Nt as equal to 2, N_FFT as equal to 8, Np as equal to 8, K₁ and K₂ as equal to 2, and Δt as equal to 2. These parameters satisfy the above condition because (2+2)×2≦8.

The training sequence transmitted by the transmit antenna TX₁ consists of two replicas of the basic training sequence, one of which is phase offset. Its expression is as follows:

$\begin{array}{c}{\tilde{c}\left(p\right)}_1=\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(1-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(1+\mathrm{Nt}-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right)+\tilde{c}\left(p\right){}^{-\mathrm{j\pi }p}\end{array}$

The training sequence transmitted by the transmit antenna TX₂ also consists of two replicas of the basic training sequence, both of which are phase offset. Its expression is as follows:

$\begin{array}{c}{\tilde{c}\left(p\right)}_2=\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(2-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-\mathrm{j2\pi }\frac{p\times \left(2+\mathrm{Nt}-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right){}^{-j\frac{\pi }{2}p}+\tilde{c}\left(p\right){}^{-j\frac{3\pi }{2}p}\end{array}$

In the example illustrated by FIG. 7, the receive device RE includes as many OFDM demultiplexers DX₁ to DX_Nr as receive antennas RX₁ to RX_Nr. It further includes a channel estimator module EsT, an equalizer module EgL, a binary symbol decoder module DbS, a bit de-interleaver module Dbt, and a channel decoder module DEC.

Each signal received by a receive antenna is processed by an OFDM demultiplexer. An OFDM demultiplexer performs the inverse function of an OFDM multiplexer. Accordingly an OFDM demultiplexer applies a direct Fourier transform of size N_FFT to the received signal after suppression of the guard interval. An OFDM demultiplexer demodulates a time-domain signal received by the receive antenna RX_j concerned by means of an FFT of size N_FFT to obtain a frequency-domain signal R_j(n) combining the Np pilot symbols of the training sequences of the various transmit antennas.

The outputs of the OFDM demultiplexers provide frequency-domain OFDM signals R_j(n) that are used by the estimator module and the equalizer module.

The channel estimator module includes means for computing the matrix A according to equation (22) or (27). These means are typically program instructions which in a preferred embodiment are incorporated into an electronic circuit, a computation unit such as a processor or the like the operation of which is then controlled by the execution of the program.

The channel estimator module includes means for computing the K_i estimates of each of the Nt impulse responses ĥ_j according to equation (25). These means are typically program instructions which in a preferred embodiment are incorporated into an electronic circuit, a computation unit such as a processor or the like the operation of which is then controlled by the execution of the program.

Knowing the parameters of the multiple-antenna system (the size N_FFT of the FFT, the number N_mod of modulated carriers, the offset Δt between reference sequences, the number g of pilot OFDM symbols forming a reference sequence, the amplitude {tilde over (c)}(p) of the pilot symbols), an estimator method and module of the invention advantageously pre-compute the matrix corresponding to the following expression in which the matrix à is given by equation (26) or (27):

à ⁺=(Ã^HÃ)⁻¹Ã^H  (28)

In a preferred embodiment, the estimation method, respectively the estimator module, multiplies this pre-computed matrix product by the demodulated OFDM signal to obtain the K_i estimates of each channel combined in the expression ĥ_j according to equation (25).

The vector expression of the first estimate ĥ_j,i of the impulse response of the channel separating the particular transmit antenna i, for i between 1 and Nt, from the particular receive antenna j, for j between 1 and Nr, is obtained by selecting Np successive samples of the column vector ĥ_j computed for the receive antenna j, starting with the sample (i−1)Δt. For the second estimate ĥ_j,i of the impulse response of the channel, selection starts from the sample Nt×Δt+(i−1)Δt=(Nt+−1)Δt. For the n^th estimate ĥ_j,i of the impulse response of the channel, selection starts from the sample (n×Nt+i)Δt.

To obtain the estimates for the Nr×Nt channels, the computation of ĥ_j,i must be repeated for j varying from 1 to Nr.

The channel estimator module includes means for computing an average of the K_i impulse response estimates, thus obtaining Nt impulse responses.

The following two examples illustrate the computation of the expression for Ã.

First example. The transmission system includes Nt=2 transmit antennas and Nr=1 receive antenna. The parameters of the system take the following values:

• • N_FFT=8, N_mod=8, Δt=2, K₁=K₂=2.

The condition

$\sum _{i=1}^{\mathrm{Nt}}K_i\times \Delta t\le \mathrm{Np}$

is satisfied because (2+2)×2=8≦8.

A single OFDM symbol is assumed necessary for the channel estimation (g=1) and the pattern of this OFDM symbol, which constitutes the basic training sequence, is represented in Table 1. The first column represents the frequency domain indices and the second column the associated data type (P=pilot symbol). Equation (6) gives the expression for the pilot symbol modulating the carrier p where the training sequence transmitted the transmit antenna TX_i is concerned, for i=1 or 2:

$\begin{array}{cc}\begin{array}{c}{\tilde{c}\left(p\right)}_1=\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(1-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(1+\mathrm{Nt}-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right)+\tilde{c}\left(p\right){}^{-j\pi p}\end{array}& \\ \begin{array}{c}{\tilde{c}\left(p\right)}_2=\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(2-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(2+\mathrm{Nt}-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right){}^{-j\frac{\pi }{2}p}+\tilde{c}\left(p\right){}^{-j\frac{3\pi }{2}p}\end{array}& \end{array}$

Thus c(p)₁, is the pilot symbol modulating the carrier frequency p for the training sequence transmitted by the transmit antenna TX₁ and c(p)₂ is the pilot symbol modulating the carrier frequency p for the training sequence transmitted by the transmitter antenna TX₂.

For all the pilot carriers of this example c(p)_i=1. The vectors of the pilot symbols B₁ and B₃, respectively B₂ and B₄, corresponding to the two estimates of the impulse response of the channel connecting the antenna TX₁ to the antenna RX_j, respectively the antenna TX₂ to the antenna RX_j, are then expressed in the form:

$B_1=\left[\begin{array}{c}1\\ 1\\ 1\\ 1\\ 1\\ 1\\ 1\\ 1\end{array}\right];B_2=\left[\begin{array}{c}1\\ {}^{-j\frac{\pi }{2}}\\ -1\\ {}^{-j\frac{3\pi }{2}}\\ 1\\ {}^{-j\frac{\pi }{2}}\\ -1\\ {}^{-j\frac{3\pi }{2}}\end{array}\right];B_3=\left[\begin{array}{c}1\\ -1\\ 1\\ -1\\ 1\\ -1\\ 1\\ -1\end{array}\right];B_4=\left[\begin{array}{c}1\\ {}^{-j\frac{3\pi }{2}}\\ -1\\ {}^{-j\frac{\pi }{2}}\\ 1\\ {}^{-j\frac{3\pi }{2}}\\ -1\\ {}^{-j\frac{\pi }{2}}\end{array}\right]$

The Fourier matrix F is as follows:

$F=\frac{1}{\sqrt{8}}\left[\begin{array}{cccccccc}1& 1& 1& 1& 1& 1& 1& 1\\ 1& 0.707-0.707j& -j& -0.707-0.707j& -1& -0.707+0.707j& j& 0.707+0.707j\\ 1& -j& -1& j& 1& -j& -1& j\\ 1& -0.707-0.707j& j& 0.707-0.707j& -1& 0.707+0.707j& -j& -0.707+0.707j\\ 1& -1& 1& -1& 1& -1& 1& -1\\ 1& -0.707+0.707j& -j& 0.707+0.707j& -1& 0.707-0.707j& j& -0.707-0.707j\\ 1& j& -1& -j& 1& j& -1& -j\\ 1& 0.707+0.707j& j& -0.707+0.707j& -1& -0.707-0.707j& -j& 0.707-0.707j\end{array}\right]$

in which the symbol j represents the solution of the equation j²=−1. The matrix F′ is expressed in the following form:

$F^{\prime }=\frac{1}{\sqrt{8}}\left[\begin{array}{cc}1& 1\\ 1& 0.707-0.707j\\ 1& -j\\ 1& -0.707-0.707j\\ 1& -1\\ 1& -0.707+0.707j\\ 1& j\\ 1& 0.707+0.707j\end{array}\right]$

The matrix à is then expressed in the following form:

$\tilde{A}=\frac{1}{\sqrt{8}}\left[\begin{array}{cccccccc}{\tilde{B}}_{1\mathrm{diag}}& {\tilde{F}}^{\prime }& {\tilde{B}}_{2\mathrm{diag}}& {\tilde{F}}^{\prime }& {\tilde{B}}_{3\mathrm{diag}}& {\tilde{F}}^{\prime }& {\tilde{B}}_{4\mathrm{diag}}& {\tilde{F}}^{\prime }\end{array}\right]$

i.e.:

$\tilde{A}=\frac{1}{\sqrt{8}}\left[\begin{array}{cccc}1& 1& 1& 1\\ 1& {}^{-j\frac{3\pi }{4}}& -1& {}^{j\frac{\pi }{4}}\\ 1& j& 1& j\\ 1& {}^{-j\frac{\pi }{4}}& -1& {}^{j\frac{3}{4}}\\ 1& -1& 1& -1\\ 1& {}^{j\frac{\pi }{4}}& -1& {}^{-j\frac{3\pi }{4}}\\ 1& -j& 1& -j\\ 1& {}^{j\frac{3\pi }{4}}& -1& {}^{-j\frac{\pi }{4}}\end{array}\right]$

Second example. The transmission system includes Nt=3 transmit antennas and Nr=1 receive antenna. The parameters of the system take the following values:

• • N_FFT=8, N_mod=Np=8, Δt=2, K₁=2 et K₂=K₃=1.

The condition

$\sum _{i=1}^{\mathrm{Nt}}K_i\times \Delta t\le Np$

is satisfied because (2+1+1)×2=8≦8.

A single OFDM symbol is assumed necessary for the channel estimation (g=1), and the pattern of this OFDM symbol, which constitutes the basic training sequence, is represented in Table 1. Equation (6) gives the expression for the pilot symbol modulating the carrier p for the training sequence transmitted by the transmit antenna TX_i, for i=1, 2 or 3:

$\begin{array}{cc}\begin{array}{c}{\tilde{c}\left(p\right)}_1=\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(1-1\right)\times 2}{8}}+\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(1+\mathrm{Nt}-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right)+\tilde{c}\left(p\right){}^{-j\frac{3}{2}\pi p}\end{array}& \\ \begin{array}{c}{\tilde{c}\left(p\right)}_2=\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(2-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right){}^{-j\frac{\pi }{2}p}\end{array}& \\ \begin{array}{c}{\tilde{c}\left(p\right)}_3=\tilde{c}\left(p\right){}^{-j2\pi \frac{p\times \left(3-1\right)\times 2}{8}}\\ =\tilde{c}\left(p\right){}^{-j\pi p}\end{array}& \end{array}$

Thus c(p)₁ is the pilot symbol modulating the carrier frequency p for the training sequence transmitted by the transmit antenna TX₁, c(p)₂ is the pilot symbol modulating the carrier frequency p for the training sequence transmitted by the transmitter antenna TX₂, and c(p)₃ is the pilot symbol modulating the carrier frequency p for the reference sequence transmitted by the transmit antenna TX₃.

For example, c(p)=1 for all the pilot carriers. The vectors of the pilot symbols B₁ and B₄, respectively B₂ and B₃, corresponding to the two estimates of the impulse response of the channel connecting the antenna TX₁ to the antenna RX_j, respectively to the estimate of the impulse response of the channel connecting the antenna TX₂ to the antenna RX_j, and to the estimate of the impulse response of the channel connecting the antenna TX₃ to the antenna RX_j, are then expressed in the following form:

$B_1=\left[\begin{array}{c}1\\ 1\\ 1\\ 1\\ 1\\ 1\\ 1\\ 1\end{array}\right];B_2=\left[\begin{array}{c}1\\ {}^{-j\frac{\pi }{2}}\\ -1\\ {}^{-j\frac{3\pi }{2}}\\ 1\\ {}^{-j\frac{\pi }{2}}\\ -1\\ {}^{-j\frac{3\pi }{2}}\end{array}\right];B_3=\left[\begin{array}{c}1\\ -1\\ 1\\ -1\\ 1\\ -1\\ 1\\ -1\end{array}\right];B_4=\left[\begin{array}{c}1\\ {}^{-j\frac{3\pi }{2}}\\ -1\\ {}^{-j\frac{\pi }{2}}\\ 1\\ {}^{-j\frac{3\pi }{2}}\\ -1\\ {}^{-j\frac{\pi }{2}}\end{array}\right]$

The Fourier matrix F is as follows:

$F=\frac{1}{\sqrt{8}}\left[\begin{array}{cccccccc}1& 1& 1& 1& 1& 1& 1& 1\\ 1& 0.707-0.707j& -j& -0.707-0.707j& -1& -0.707+0.707j& j& 0.707+0.707j\\ 1& -j& -1& j& 1& -j& -1& j\\ 1& -0.707-0.707j& j& 0.707-0.707j& -1& 0.707+0.707j& -j& -0.707+0.707j\\ 1& -1& 1& -1& 1& -1& 1& -1\\ 1& -0.707+0.707j& -j& 0.707+0.707j& -1& 0.707-0.707j& j& -0.707-0.707j\\ 1& j& -1& -j& 1& j& -1& -j\\ 1& 0.707+0.707j& j& -0.707+0.707j& -1& -0.707-0.707j& -j& 0.707-0.707j\end{array}\right]$

in which the symbol j represents the solution of the equation j²=−1. The matrix F′ is expressed in the following form:

$F^{\prime }=\frac{1}{\sqrt{8}}\left[\begin{array}{cc}1& 1\\ 1& 0.707-0.707j\\ 1& -j\\ 1& -0.707-0.707j\\ 1& -1\\ 1& -0.707+0.707j\\ 1& j\\ 1& 0.707+0.707j\end{array}\right]$

The matrix à is then expressed in the following form:

$\tilde{A}=\frac{1}{\sqrt{8}}\left[\begin{array}{cccccccc}{\tilde{B}}_{1\mathrm{diag}}& {\tilde{F}}^{\prime }& {\tilde{B}}_{2\mathrm{diag}}& {\tilde{F}}^{\prime }& {\tilde{B}}_{3\mathrm{diag}}& {\tilde{F}}^{\prime }& {\tilde{B}}_{4\mathrm{diag}}& {\tilde{F}}^{\prime }\end{array}\right]$

i.e.:

$\tilde{A}=\frac{1}{\sqrt{8}}\left[\begin{array}{cccc}1& 1& 1& 1\\ 1& {}^{-j\frac{3\pi }{4}}& -1& {}^{j\frac{\pi }{4}}\\ 1& j& 1& j\\ 1& {}^{-j\frac{\pi }{4}}& -1& {}^{j\frac{3\pi }{4}}\\ 1& -1& 1& -1\\ 1& {}^{j\frac{\pi }{4}}& -1& {}^{-j\frac{3\pi }{4}}\\ 1& -j& 1& -j\\ 1& {}^{j\frac{3\pi }{4}}& -1& {}^{-j\frac{\pi }{4}}\end{array}\right]$

FIG. 8 gives simulation results comparing the performance of the known Barhumi technique and a method of the invention. The simulations relate to a MIMO OFDM system and apply to a channel BRAN E that is selective in time and in frequency and is suited to a MIMO context of the external environment type, with the characteristics set out in Table 2 in Appendix A. The frame taken into account is that described in the Barhumi paper; the set of all modulated carriers is divided into two disjoint sub-sets: a set of pilot carriers and a set of payload data. The basic training sequence is a time pulse. The same transmitted pilot symbol power is imposed for both the methods compared. The figure shows the trace of the binary error rate (BER) as a function of the ratio Eb/No for QPSK modulation with an offset Δt=128. The estimation of the channels effected at the pilot frequencies is complemented by time-linear interpolation for estimating all the coefficients for the various modulated frequencies. The curve 1, c1, corresponds to a perfect estimate, the curve 2, c2, corresponds to an estimate according to Barhumi with one iteration, the curve 3, c3, corresponds to an estimate according to Barhumi with five iterations, the curve 4, c4, corresponds to an estimate according to the invention with K=2 and one iteration, and the curve 5, c5 corresponds to an estimate according to the invention with K=2 and five iterations. The iterations correspond to the iterations of the interference suppresser during equalization. Comparing the curves shows that the method of the invention has the advantage of a lower bit error rate than the Barhumi method of the same signal-to-noise ratio.

APPENDIX A

TABLE 1
0 P
1 P
2 P
3 P
4 P
5 P
6 P
7 P

	TABLE 2
	Nt	2
	Nr	2
	Carrier frequency	5200	MHz
	Sampling frequency	50	MHz
	Carrier spacing	48.828125	kHz
	Cyclic prefix ratio	0.211
	OFDM symbol time	24.8	μs
	Convolutional code length	7
	Code efficiency	1/2
	Channel decoding algorithm	SOVA
	FFT size	1024
	Modulated carriers	1024
	Space-time code	Spatial multiplexing

Claims

1. A transmission method suitable for a system with Nt transmit antennas and Nr receive antennas, where Nt and Nr are at least 1, a transmit antenna being separated from a receive antenna by a transmission sub-channel, implemented by Nt multiple-carrier transmitters each including a multiplexing and modulation module using NFFT orthogonal functions (MX1 to MXNt) forming orthogonal symbols intended to be transmitted by the Nt transmit antennas TX1 to TXNt, the method including: a step of determining a basic training sequence ({tilde over (c)}(p)) from the position of Np pilot symbols in a time-frequency frame;a step of determining

2. A symbol transmission method according to claim 1, wherein a guard interval is added before transmitting an orthogonal symbol and the constraint on the minimum value of the time offset is complied with by choosing a minimum time offset equal to the value of the guard interval.

3. A symbol transmission method according to claim 1, wherein each training sequence consists of K replicas of the basic training sequence ({tilde over (c)}(p)).

4. A symbol transmission method according to claim 1, wherein the replicas are determined in the frequency domain and the time offset is determined by a phase offset value.

5. A symbol transmission method according to claim 4, wherein the phase offset values are chosen in a particular order.

6. A symbol transmission method according to claim 5, wherein the order is known to at least one receiver associated with a receive antenna of the system.

7. A symbol transmission method according to claim 4, wherein the phase offset is a function of a carrier index.

8. A symbol transmission method according to claim 1, wherein the basic training sequence is spread over a plurality of orthogonal symbols.

9. A method for estimating transmission sub-channels in a multiple-antenna system using Nt transmit antennas (TX1 to TXNt), where Nt is at least 1, at least one receive antenna (RX1 to RXNt), and a time-frequency frame for each transmit antenna including Np pilot symbols (sp) forming a training sequence and data symbols of a payload signal, each of the Nt training sequences being the sum of Ki replicas of a basic training sequence ({tilde over (c)}(p)), either time offset or not, such that the values Ki are greater than or equal to 1 and at least one value Ki is greater than or equal to 2, being known to the receiver, and enabling the receiver to estimate an impulse response corresponding to the transmission channel separating a transmit antenna (TX1 to TXNt) from a receive antenna RXj concerned, the data symbols and the pilot symbols being frequency-modulated by a multiplexing and modulation module using NFFT orthogonal functions (MX1 to MXNt) to form orthogonal symbols that are transmitted by the transmit antennas in the form of a multiple-carrier signal using NFFT carriers, the method including: a step of computing a matrix A constructed in the form of blocks from the training sequences and from the Fourier matrix of dimensions NFFT×NFFT, the number of blocks of which is equal to the product of the number of orthogonal symbols of the basic training sequence by the sum of the values Ki, a block being determined by the product of a diagonal matrix formed of the pilot symbols contained in an orthogonal symbol of the training sequence associated with the block multiplied by the Fourier matrix and including, for a receive antenna (RXj) concerned;a step of computing Ki estimates of at least one of the Nt impulse responses in the time domain by multiplying by the pseudo-inverse matrix of the matrix A Np pilot symbols extracted from a frequency-domain signal (Rj(n)) obtained after demodulation of a time-domain signal received by the receive antenna (RXj) concerned by means of a Fourier transform of size NFFT;a step of computing an average over the Ki estimates of the impulse response concerned.

10. A transmission channel estimation method according to claim 9, wherein the computation of the Nt impulse responses is extended by means of interpolation to the carriers modulated by data.

11. A transmission channel estimation method according to claim 9, wherein the step of computing the Nt impulse responses is repeated for each receive antenna concerned of the multiple-antenna system.

12. A transmission module including a multiplexing and modulation module using NFFT orthogonal functions (MX1 to MXNt), suitable for a multiple-antenna system using Nt transmit antennas (TX1 to TXNt), Nr receive antennas, where Nt and Nr are at least 1, and a time-frequency frame for each transmit antenna including pilot symbols and data symbols of a payload signal, the data symbols and the pilot symbols being frequency-modulated by the multiplexing and modulation module (MX1 to MXNt) to form orthogonal symbols that are transmitted by a transmit antenna in the form of a multiple-carrier signal using NFFT carriers and using Np pilot carriers, the module including: a module for determining a basic training sequence ({tilde over (c)}(p)) from the position of Np pilot symbols in the time-frequency frame associated with a particular transmit antenna (TX1);a module for determining Kn, where

13. A transmitter for a multiple-antenna system which includes at least one transmission module according to claim 12.

14. A module for estimating transmission channels in a multiple-antenna system using Nt transmit antennas (TX1 to TXNt), where Nt is at least 1, at least one receive antenna (RX1 to RXNt), and a time-frequency frame for each transmit antenna including pilot symbols (sp) forming a training sequence and data symbols of a payload signal, the Nt training sequences being known to the receiver and enabling the receiver to estimate Nt impulse responses corresponding to the Nt transmission channels respectively separating one of the transmit antennas (TX1 to TXNt) from the receive antenna RXj concerned, the data symbols and the pilot symbols being frequency-modulated by a multiplexing and modulation module using NFFT orthogonal functions (MX1 to MXNt) to form orthogonal symbols that are transmitted by the transmit antennas in the form of a multiple-carrier signal using NFFT carriers with Np pilot carriers, characterized in that it includes means for computing a matrix A constructed in the form of blocks from the training sequences and from the Fourier matrix of dimensions NFFT×NFFT, the number of blocks of which is equal to the product of the number of orthogonal symbols of the basic training sequence by the sum of the values Ki, a block being determined by the product of a diagonal matrix formed of the pilot symbols contained in an orthogonal symbol of the training sequence associated with the block multiplied by the Fourier matrix, and, for a receive antenna (RXj) concerned: means for computing Ki estimates of at least one of the Nt impulse responses in the time domain by multiplying Np pilot symbols extracted from a frequency-domain signal (Rj(n)) obtained after demodulation of a time-domain signal received by the receive antenna (RXj) concerned by means of an FFT of size NFFT by the pseudo-inverse matrix of the matrix A; andmeans for computing the average of the Ki estimates of the impulse response concerned.

15. A receiver for a multiple-antenna system which includes at least one module according to claim 14 for estimating transmission channels.

16. A multiple-antenna system which includes at least one receiver according to claim 15.

17. A computer program on an information medium, said program including program instructions adapted to execute a transmission method according claim 1 when said program is loaded in and executed in a transmitter.

18. An information medium including program instructions adapted to execute a transmission method according to claim 1 when said method is loaded in and executed in a transmitter.

19. A computer program on an information medium, said program including program instructions adapted to execute a channel estimation method according to claim 9 when said program is loaded in and executed in a receiver.

20. An information medium including program instructions adapted to execute a channel estimation method according to claim 9 when said program is loaded in and executed in a receiver.