Title of Invention  "METHOD FOR THE MULTIANTENNAE EMISSION OF A SIGNAL BY UNITARY SPACETIME CODES" 

Abstract  The present invention relates to a method for sending a signal formed by successive vectors each comprising N symbols to be sent, and implementing at least two transmitter antennas. According to the invention, a distinct submatrix is associated with each of said antennas, said submatrices being obtained by subdivision of a unitary square matrix, and each of said antennas sends subvectors, obtained by subdivision of said vectors, respectively multiplied by said submatrices so as to form, as seen from a receiver, a single combined signal representing the multiplication of said vectors by said unitary matrix. Figure 1 
Full Text  Method for the multiantennae emission of a signal by unitary spacetime codes, receiving method, and corresponding signal 1. Field of the invention The field of the invention is that of wireless digital communications. More specifically, the invention relates to the sending/receiving of a signal using a new type of spacetime block code in the context of a MIMO ("Multiple Input Multiple Output ") or MISO ("Multiple Input Single Output") type multipleantenna system. The invention can thus be applied especially to transmission systems implementing a plurality of antennas (at least two of them) at sending and/or at reception. The invention is therefore well suited to receivers for nonorthogonal spacetime codes with Nt transmitter antennas and Nr receiver antennas. The invention can be applied in the field of radio communications, especially for systems of the third, fourth and following generations. 2. Prior art solutions There already exist several known sending/receiving systems comprising several transmitter antennas and using spacetime codes. The earliest systems proposed all used orthogonal spacetime block codes. Thus, Alamouti in "A Simple Transmit Diversity Technique for Wireless Communications", IEEE J. Sel. Areas Comm., 1998, 16, (8), pp. 14511458, presented the first system using a rateone orthogonal spacetime block code (where rate is defined as the ratio between the number N of symbols sent and the number L of symbol times or periods during which they are sent), for two transmitter antennas. Tarokh et al. ("Spacetime block codes from orthogonal designs", IEEE Trans, on Information Theory, 1999, 45, (5), pp. 14561467) then generalized the orthogonal spacetime block codes to systems comprising three or four transmitter antennas. However the rate values R=N/L obtained were only rate 1/2 or rate 3/4. The next studies led to envisaging the use of nonorthogonal spacetime block codes. Thus Jafarkhani ("A QuasiOrthogonal SpaceTime Block Code", IEEE Trans. Comm., 2001, 49, (1), pp 14) and Tirkonnen et al. ("Minimal nonorthogonality rate one spacetime block code for 3+ Tx antennas", ISSSTA, 2000, pp 429432) have found rateone nonorthogonal spacetime block codes for a fourantenna system. Subsequently, Damen et al. ("Diagonal Algebraic SpaceTime Block Codes", IEEE Trans. Inf. Theory, 2002, 48, (3), pp 628626) envisaged the use of nonorthogonal spacetime codes based on a Hadamard construction and other rotations with a number of transmitter antennas equal to the size of the code matrix. Xin et al., in "Time ConstellationRotating Codes Maximizing Diversity and Coding Gains", GLOBECOM, San Antonio, 2001, pp 455459, subsequently presented other rotationbased spacetime codes. 3. The drawbacks of the prior art One drawback of Alamouti's or Tarokh's orthogonal spacetime codes is that they require the channels to be constant for the duration L, corresponding to the number of symbol periods during which the symbols are sent. Such codes therefore place heavy constraints on the sending/receiving systems, and cannot be used to exploit the diversity of the channel. One drawback of the nonorthogonal spacetime codes proposed by Jafarkhani, Tirkonnen, Damen or Xin is that they require the channel to be constant for a period L=Nt, where Nt is the number of antennas at transmission. This is particularly true for the Damen and Xin codes. In other words, a major drawback of all the spacetime codes proposed in the literature is that they require the solution to be placed in the context of a quasistatic channel. This is particularly restrictive and does not permit the diversity of the channels to be exploited. Furthermore, the Jafarkhani and Tirkonnen codes dictate a maximum likelihood (ML) decoding whose complexity increases exponentially with the order of modulation and the code size. Finally, another drawback of the Damen algebraic spacetime codes, which rely on a Hadamard construction, is that they have to be sent in a particular matrix form. They therefore cannot be used to obtain a choice of encoding that is flexible according to the variations of the channel. 4. The goals of the invention It is a goal of the invention especially to mitigate these drawbacks of the prior art. More specifically, it is a goal of the invention to provide a signalsending technique that implements spacetime codes with improved performance as compared with prior art spacetime codes . It is yet another goal of the invention to implement a technique of this kind that lays down no special conditions as regards the nonvariation of the channel over a finite duration or given number of symbol periods. It is yet another goal of the invention to provide a technique of this kind that is adapted to MIMO and MISO type antenna systems. More particularly, it is a goal of the invention to provide a technique of this kind that gives a constant encoding rate of one, whatever the number of antennas. It is also a goal of the invention to implement a technique of this kind whose binary error rate performance at high signaltonoise ratios is superior to that of the prior art. It is yet another goal of the invention to provide a technique of this kind that can be easily adapted to different types of configuration, such as an increase in the number of antennas or the size of the code used. It is yet another goal of the invention to implement a technique of this kind enabling channel diversity to be better exploited than in prior art techniques and enabling flexible encoding. 5. Essential characteristics of the invention These goals, as well as others there should appear here below, are achieved by means of a method for sending a signal formed by successive vectors each comprising N symbols to be sent, and implementing at least two transmitter antennas. According to the invention, a distinct submatrix is associated with each of said antennas, said submatrices being obtained by subdivision of a unitary square matrix, and each of said antennas sends subvectors, obtained by subdivision of said vectors, respectively multiplied by said submatrices, so as to form, as seen from a receiver, a single combined signal representing the multiplication of said vectors by said unitary matrix. Thus, the invention relies on a wholly novel and inventive approach to the sending of a signal implementing a spacetime code in a multipleantenna system. The technique of the invention is particularly advantageous since it imposes no conditions on the channel: unlike the prior art techniques, the proposed encoding does not require the channel to be constant for the duration of the code. The particular building of spacetime codes proposed by the invention is totally novel, and relies on an orthogonal or unitary matrix construction for each antenna. Indeed, the use of these matrices enables a separation of the signals sent by each antenna. In the invention, therefore, the system constraints are less limited than with the techniques of the prior art, and the channel diversity can be better exploited. The binary error rate performance values obtained at high signaltonoise ratios are superior to those given in the literature. Furthermore, the method of the invention can very easily be extended to a larger number of antennas because it can be obtained directly by means of the basic unitary or orthogonal matrix. Whatever the number of antennas used, the encoding rate remains constant. Advantageously, with such a sending method implementing Nt antennes, each of said submatrices has a size of (N/Nt) x N. The method of the invention can indeed be easily adapted to different configurations, especially to an increase in the number Nt of antennas. The different matrices sized (N/Nt) x N are obtained from a main matrix sized N x N that is subdivided into Nt different matrices. Preferably, N/Nt is greater than or equal to 2. Advantageously, said unitary matrix is full. In other words, each of the elements of the matrix is nonzero. Preferably, said unitary matrix belongs to the group comprising:  the real Hadamard matrices;  the complex Hadamard matrices;  the Fourier matrices;  the real rotation matrices; the complex rotation matrices. The different matrices bear no relation with one another. However, they all have the common characteristic of being either orthogonal in the case of a real matrix, or unitary in the case of a complex matrix. According to a first advantageous variant of the invention, such a method implements two transmitter antennas, and said submatrices are equal to [ 1 1] and [I 1]. According to a second advantageous variant of the invention, such a method implements two transmitter antennas and said submatrices have a value (Figure Removed) In this preferred embodiment of the invention, therefore, several codes are used for each user; in other words, for each user, each submatrix has at least two rows. According a third advantageous variant of the invention, such a method implements four transmitter antennas and said submatrices have a value of [1 1 1 1], [I 1 1 !],[! I 1 l]and[l 1 1 1]. The invention also relates to a method for the reception of a signal sent according to the sending method described here above, implementing at least one receiver antenna, that receives said single combined signal on each of said receiver antennas, and decodes said single combined signal by means of the decoding matrix corresponding to a conjugate transpose matrix of said unitary matrix. Preferably, a maximum likelihood decoding is applied to the data coming from the multiplication by said conjugate transpose matrix. It is also possible to use other less complex decoders, and thus achieve a sphere decoding or a QR decomposition decoding for example. It may be recalled that any hermitian matrix may indeed be decomposed in the form QR, where Q is a unitary matrix and R is an upper triangular matrix. A QR decomposition of this kind has a O3 complexity, which is therefore lower than the maximum likelihood decoding which has an OL complexity. The invention therefore relates to a signal sent according to the transmission method described here above, corresponding to the combination of the contributions of each of said transmitter antennas, a distinct submatrix being associated with each of said antennas, said submatrices being obtained by subdivision of a unitary square matrix. Each of said antennas sends subvectors, obtained by subdivision of said vectors, respectively multiplied by said submatrices. Seen from a receiver, such a signal forms a single combined signal representing the multiplication of said vectors by said unitary matrix. 6. List of figures Other features and advantages of the invention shall appear more clearly from the following description of a preferred embodiment, given by way of a simple illustrative and nonexhaustive example and from the appended drawings, of which: Figure 1 is a flow chart showing the different steps implemented in transmission and in reception for a signal encoded by means of the spacetime codes of the invention; Figure 2 illustrates a twoantenna system implementing a 2x2 matrix spacetime code according to the invention; Figure 3 illustrates a twoantenna system similar to that of figure 2 but implementing a 4x4 matrix spacetime code; Figure 4 describes a fourantenna system implementing a 4x4 matrix spacetime code; Figure 5 presents the comparative performance values of the different rate 1 spacetime codes, in the prior art and in the present invention, for two transmitter antennas and one receiver antenna; Figure 6 presents the comparative performance values of the different rate 1 spacetime codes, in the prior art and in the present invention, for four transmitter antennas and one receiver antenna. 7. Description of an embodiment of the invention The genera! principle of the invention relies on a novel type of spacetime code for a multipleantenna system. The particular building of these codes relies on an orthogonal or unitary matrix construction for each transmitter antenna, the use of these matrices enabling a separation of the signals sent by each antenna. Figure 1 presents the different steps implemented during the sending and reception of a signal according to the invention. At encoding, X is considered to be a vector sized N containing the N symbols to be sent. A system comprising a number Nt of transmitter antennas is also considered. The novel spacetime codes proposed by the invention are conceived as follow: In a first step 1, the vector X containing the symbols to be sent is divided into Nt subvectors sized N/Nt. Then, in a step 2, each of the subvectors sized N/Nt is divided by a different matrix sized (N/Nt)xN. These matrices are obtained from a real Hadamard matrix, a complex Hadamard or Fourier matrix, as well as from any real or complex rotation matrix. Although they have no relationship with one another, these different matrices have common characteristics. Indeed, each of these matrices is either orthogonal if it is a real matrix or unitary if it is a complex matrix. Furthermore, each of these matrices has to be full, i.e. each of these elements has to be nonzero. Thus, the different matrices sized (N/Nt)xN are obtained from a main matrix sized NxN which is subdivided into Nt different matrices. Then, in the step 3, the different subvectors encoded on each of the Nt During the step referenced 2, each of the subvectors sized 5 is multiplied by a different matrix. A known matrix sized 10x10 is the Fourier matrix. This matrix sized 10x10 is divided into two matrices sized 5x10. Therefore, each of the subvectors sized 5 is multiplied by one of the two matrices 5x10. After this operation, the two subvectors corresponding to the two antennas corresponds to the space time code used for sending. During the step referenced 3, the two different subvectors encoded on each of the two transmitter antennas are sent. After propagation 7 via the transmission channel and reception 4, , an equalisation step 5 is performed at decoding. This equalisation step 5 is associated with the reverse decoding of the recombined transmission matrix. This matrix is the conjugate transpose of the 10x10 transmission Fourier matrix. In a step referenced 6, a maximum resemblance decoding, or a less complex decoding of the sphere type, or a QR decomposition decoding is performed. With a maximum likelihood decoding, if M corresponds to the modulation alphabet, an exhaustive search has to be made for the signal sent by performing M'° comparisons. Referring now to figure 2, an embodiment of the invention is presented in the case of a twoantenna (10, 11) transmission system using 2x2 matrix spacetime codes. This figure 2 illustrates the transmission system of the invention with Hadamard code sequences with a length 2 per antenna. This code has an encoding rate of 1/2 or a spacetime encoding rate of 1 if the rate R is considered to be equal to the number of symbols N divided by the number of symbol periods L during which they are sent. In figure 2, x(, x2 represent the symbols to be transmitted, h,, h2, h3, h4are, for example, flat Rayleigh propagation channels, associated with the two transmitter antennas 10, 11 and y, and y2 are the equalized symbols retrieved during the step referenced 5 of figure 1. After multiplication of the subvector by the submatrix [l ij associated with the transmitter antenna 10, this antenna sends the subvector [i, jc,J. Similarly, the transmitter antenna 11 sends the subvector [2 i,J, obtained by multiplication of x, J by the submatrix [l ij. In considering an adapted interleaving and a decorrelation between the antennas, the channel varies at every symbol period. At reception, the contribution of the first antenna 10 as well as that of the second antenna 11 affected by the corresponding channel is received at the receiver antenna 12 : this contribution is written in matrix form: \f\x\ + h2x2 //3x, h4x2\. A decoding and equalization step then consists in applying the transconjugate \ 2t of the transmitted unitary matrix, while the same time performing an equalization. Consequently, the different channels having contributed to the sending step are considered. An ML (or maximum likelihood) decoding can then be performed. This decoding will seek the word that was most probably sent. To do this, the metric min presented in figure 2 is computed for all the vectors (x},x2) possible, in order to determine the most probable code word sent. Figure 3 shows a twoantenna transmission system 10, 1 1 similar to that of figure 2, using 4x4 matrix spacetime codes. The system of figure 3 more specifically Hadamard code sequences with a length 4 per antenna. It is possible to consider increasing the size of the Hadamard matrices and thus obtain codes with a length L for two antennas. Again, x,, x2 x,, x4 are the symbols to be transmitted and h,, h2, h3, h4 hs, h6, h7, hH are the flat Rayleigh propagation channels associated with the two transmitter antennas 10, 11. The oddnumber indices referring to the first transmitter antenna 10, and the evennumber indices referring to the second antenna 1 1 . y,, y2 y,, y4are the equalized symbols retrieved at the receiver antenna 1 2 at the end of the step 5 of figure 1 . The transmitter antenna 10 sends the subvector obtained by multiplication of the subvector [v, ,v2] by the submatrix _L . Similarly, the V2 1 1 1 I] transmitter antenna 11 sends out the subvector, obtained by multiplication of , jc4 J by the submatrix _ At reception, the contribution of the first antenna 10 as well as that of the second antenna 11 affected by the corresponding channel ([/", r2 r3 r4 J) are received at the receiver antenna 12. A step for the decoding and equalization of the received signal then consists in applying the transconjugate of the transmission unitary matrix while at the same time  hi performing an equalization. Then, the metric min presented in figure 3 for all the vectors (jc,,j?2,jc3,.c4) possible in order to have the most probable code word sent is computed. The system of the invention is not limited in terms of number of transmitter antennas. As illustrated in figure 4, it is possible to create spacetime codes with four transmitter antennas 10, 11, 13 and 14 with a minimum matrix size L=4. In figure 4, x,, x2 x3, x4are the symbols to be transmitted, h,, h2, h3, h4 h5, h6, h7, h8 h,,, h,0, hu, h,2 h,3, h14, h,s, h^are the flat Rayleigh propagation channels, associated with each of the four antennas 10, 11, 13 and 14 as illustrated in figure 4 and y,, y2 y,, y4 are the symbols equalized after reception by the receiver antenna 12. Then the metric min illustrated in figure 4 is computed for all the possible vectors in order to determine the most probable code word sent. The principle of the sending of the subvectors by each of the transmitter antennas is similar to the one presented here above with reference to figures 10, 11, 13 and 14 and hence, for the sake of simplicity, it shall not be described in greater detail herein. It will be noted that, in the examples illustrated here above with reference to figures 2 to 4, the spacetime codes considered had been created by using simple Hadarnard matrices. However, it is possible to use any unitary, complex Hadamard or Fourier matrix. More generally, any unitary matrix can be used in the transmitter system of the invention. Figures 5 and 6 present the performance values obtained, according to the invention, in decoding the spacetime codes with an ML (maximum resemblance) decoder. Figure 5 illustrates the compared performance values of different rateone space time codes for 2 transmitter antennas and one receiver antenna in the context of a QPSK modulation. More specifically, figure 5 presents the performance of the spacetime codes of the invention for an encoding matrix size L=2, L=4 and L=8, as well as the performance of the Alamouti code. As shown in figure 5, the performance values of the codes of the invention are good at high signaltonoise ratio, when the size of the encoding matrix L increases. Indeed, for two transmitter antennas and one receiver antenna, when an encoding matrix L=8 is taken, the performance values of the codes of the invention exceed the performance values of the reference curve of the Alamouti encoding ("A Simple Transmit Diversity Technique for Wireless Communications", IEEE J. Sel. Areas Comm., 1998, 16, (8), pp. 14511458) for a bit/energy noise ratio Eb/N0>10 dB. More specifically, it can be realised that, the greater the increase in the size of the matrix of the codes, the more the signals are detected with a high order of channel diversity. This is expressed by the slope of the performance curves of figure 5: the more accentuated the slope, the greater the increase in the order of We Claim 1. A method for sending a signal implementing Nt transmit antennas (10, 11), with Nt > 2, wherein the method implements the following steps, for at least one vector comprising N symbols to be sent: dividing (1) said vector into Nt subvectors; multiplying (2) each of the Nt subvectors by a distinct submatrice, each submatrix being associated with one of the transmit antennas (10, 11), and said submatrices being obtained by subdivision of a unitary square matrix; and sending (3), from the Nt transmit antennas (10, 11), the Nt sub vectors resulting from the multiplying step (2). 2. A method as claimed in claim 1, implementing Nt antennas, wherein each of said submatrices has a size of (N/Nt) x N. 3. A method as claimed in claim 1, wherein N/Nt is greater than or equal to 2. 4. A method as claimed claims 1 to 3, wherein said unitary matrix is full. 5. A method as claimed in claims 1 to 4, wherein said unitary matrix belongs to the group comprising: real Hadamard matrices, complex Hadamard matrices, Fourier matrices, real rotation matrices, complex rotation matrices. 6. A method as claimed in claims 1 to 5, wherein it implements two transmitter antennas and in that said submatrices having a value of [1 1] and [11]. 7. A method as claimed in claims 1 to 5, wherein it implements two transmitter antennas and in that said submatrices having a value of l/√2. 8. A method as claimed in claims 1 to 5, wherein it implements four transmitter antennas and in that said submatrices have a value of [ 1 1 1 1], [11 1 1], [1 111] and [111 1]. 9. A method as claimed in claim 1 ,for reception of a signal corresponding to a combination of contributions of Nt transmit antennas , with Nt >= 2, wherein for at least one vector comprising N symbols to be sent, the signal is generated by dividing said vector into Nt subvectors, multiplying each of the Nt sub vectors by a distinct submatrice, each submatrix being associated with one of the Nt subvectors by a distinct submatrice , each submatrix being associated with one of the transmit antennas , and said submatrices being obtained by subdivision of a unitary square matrix, and sending , from the Nt transmit antennas , the Nt subvectors resulting from the multiplying step , wherein the signal forms, seen from a receiver , a single combined signal representing the multiplication , wherein the method of reception comprises: Implementing at least one receiver antenna; Receiving said single combined signal on each of said receiver antennas; and decoding said single combined signal by a decoding matrix corresponding to a matrix that is the conjugate transpose of said unitary square matrix. 10. A reception method as claimed in claim 9, wherein a maximum likelihood decoding is applied to the data coming from the multiplication by said conjugate transpose matrix. 11. A method as claimed in claims 1 to 8, wherein sent signal corresponds to the combination of the contributions of each of said transmitter antennas, a distinct submatrix being associated with each of said antennas, said submatrices being obtained by subdivision of a unitary square matrix and in that each of said antennas sends subvectors, obtained by subdivision of said vectors, respectively multiplied by said submatrices, and in that it forms, seen from a receiver, a single combined signal representing the multiplication of said vectors by said unitary matrix. 

1726DELNP2006Abstract(27112009).pdf
1726DELNP2006Claims(27112009).pdf
1726delnp2006correspondenceothers 1.pdf
1726DELNP2006CorrespondenceOthers(27112009).pdf
1726delnp2006correspondenceothers.pdf
1726delnp2006description (complete).pdf
1726DELNP2006Drawings(27112009).pdf
1726DELNP2006Form1(27112009).pdf
1726DELNP2006Form3(27112009).pdf
1726DELNP2006GPA(01062006).pdf
Patent Number  241019  

Indian Patent Application Number  1726/DELNP/2006  
PG Journal Number  25/2010  
Publication Date  18Jun2010  
Grant Date  15Jun2010  
Date of Filing  29Mar2006  
Name of Patentee  FRANCE TELECOM, a corporation organized and existing under the laws of France, of 6, Place d Alleray 75015 Paris, France  
Applicant Address  6, PLACE D'ALLERAY 75015 PARIS,FRANCE.  
Inventors:


PCT International Classification Number  H04B 7/06  
PCT International Application Number  PCT/FR2004/002444  
PCT International Filing date  20040927  
PCT Conventions:
