Differential transmitter diversity technique for wireless communications

[0002] This application claims priority from U.S. Provisional Application No. 60/119,396 File Feb. 10, 1999.

[0003] This invention relates to wireless communication and, more particularly, to techniques for effective wireless communication in the presence of fading and other degradations.

[0004] Recently, some interesting approaches for transmitter diversity have been suggested. A delay diversity scheme was proposed by A. Wittneben in “Base Station Modulation Diversity for Digital SIMULCAST,” Proceeding of the 1991 IEEE Vehicular Technology Conference (VTC 41^st), PP. 848-853, May 1991, and in “A New Bandwidth Efficient Transmit Antenna Modulation Diversity Scheme For Linear Digital Modulation,” in Proceeding of the 1993 IEEE International Conference on Communications (IICC '93), PP. 1630-1634, May 1993. The proposal is for a base station to transmit a sequence of symbols through one antenna, and the same sequence of symbols—but delayed—through another antenna.

[0005] U.S. Pat. No. 5,479,448, issued to Nambirajan Seshadri on Dec. 26, 1995, discloses a similar arrangement where a sequence of codes is transmitted through two antennas. The sequence of codes is routed through a cycling switch that directs each code to the various antennas, in succession. Since copies of the same symbol are transmitted through multiple antennas at different times, both space and time diversity are achieved. A maximum likelihood sequence estimator (MLSE) or a minimum mean squared error (MMSE) equalizer is then used to resolve multipath distortion and provide diversity gain. See also N. Seshadri, J. H. Winters, “Two Signaling Schemes for Improving the Error Performance of FDD Transmission Systems Using Transmitter Antenna Diversity,” Proceeding of the1993 IEEE Vehicular Technology Conference(VTC 43rd), pp. 508-511, May 1993; and J. H. Winters, “The Diversity Gain of Transmit Diversity in Wireless Systems with Rayleigh Fading,” Proceeding of the1994 ICC/SUPERCOMM, New Orleans, Vol. 2, PP. 1121-1125, May 1994.

[0006] Still another interesting approach is disclosed by Tarokh, Seshadri, Calderbank and Naguib in U.S. application, Ser. No. 08/847635, filed Apr. 25, 1997 (based on a provisional application filed Nov. 7, 1996), where symbols are encoded according to the antennas through which they are simultaneously transmitted, and are decoded using a maximum likelihood decoder. More specifically, the process at the transmitter handles the information in blocks of M1 bits, where M1 is a multiple of M2, i.e., M1=k*M2. It converts each successive group of M2 bits into information symbols (generating thereby k information symbols), encodes each sequence of k information symbols into n channel codes, and applies each code of a group of codes to a different antenna.

[0007] When knowledge of the channel is available neither at the transmitter nor at the receiver, the above schemes require the transmission of pilot symbols. For one transmit antenna, differential detection schemes exist that neither require the knowledge of the channel nor employ pilot symbol transmission. These differential decoding schemes are used, for instance, in the IEEE IS-54 standard. This motivates the generalization of differential detection schemes for the case of multiple transmit antennas.

[0008] A partial solution to this problem was proposed in U.S. patent application Ser. No. 09/074,224 filed on May 7, 1998, where the detected sequence is used to estimate the channel at the receiver, and those estimates are fed back and used to detect the next transmitted set of symbols. Therefore, the '224 patent application disclosure can be thought of as a joint channel and data estimation.

[0009] Improvement in the art is realized by utilizing the fact that a space time encoding at the transmitter can be constructed where the symbols transmitted over a plurality of antennas in the time slots of a frame are orthogonal to each other. With this realization, in accordance with the principles of this disclosure, the inputs signals of each frame are mapped onto a coordinate system dictated by the symbols of the previous frame, and symbols from a constellation are selected based on the results of such mapping. Received signals are detected by preprocessing the signals detected at each antenna with signals detected by the antenna at the immediately previous frame, and then applied to a maximum likelihood detector circuit, followed by an inverse mapping circuit.

[0010] FIG. 1 is a block diagram of a transmitting unit and a receiving unit in conformance with the principles disclosed herein.

[0011] FIG. 1 depicts an arrangement where a transmitting unit 10 has two transmitting antennas, 11-1 and 11-2, and a receiving unit 20 has m receiving antenna, 21-1, 21-2, . . . , 21-m. At each time slot t, signals c_tⁱ, i=1,2 are transmitted simultaneously from the two transmit antennas. The coefficient α_i,j, is the path gain from transmit antenna i to receive antenna j. The path gains are modeled as samples of independent complex Gaussian random variables with variance 0.5 per real dimension. The wireless channel is assumed to be quasi-static, so that the path gains are constant over a frame of length l and vary, if at all, from one frame to another.

[0012] At time t the signal r_t^jthat is received at antenna j is given by $\begin{array}{cc}{r}_t^j=\sum _{i=1}^2{\alpha }_{i,j}{c}_t^i+{\eta }_t^j& \left(1\right)\end{array}$

[0013] where the noise samples η_t^jare independent samples of a zero-mean complex Gaussian random variable with variance 1/(2SNR) per complex dimension. The average energy of the symbols transmitted from each antenna is normalized to be 1/2, so that the average power of the received signal at each receive antenna is 1 and the signal to noise ratio is SNR.

[0014] Assuming coherent detection, the receiver computes the decision metric $\begin{array}{cc}\sum _{t=1}^2\sum _{j=1}^m{\left|r_t^j-\sum _{i=1}^2{\alpha }_{i,j}{c}_t^i\right|}^2& \left(2\right)\end{array}$

[0015] over all codewords

[0016] c₁¹c₁²c₂¹c₂². . . c₁¹c₁²,  (3)

[0017] and decides in favor of the codeword that minimizes the sum of equation (2).

[0018] In the FIG. 1 arrangement, the encoding matrix of transmitting unit 10 is $\begin{array}{cc}\delta =\left(\begin{array}{cc}{x}_1& x_2\\ -x_2^{*}& x_1^{*}\end{array}\right),& \left(4\right)\end{array}$

[0019] which means that as 2b bits arrive at the encoder at each frame, constellation signals s₁, and s₂are selected, and setting x_i=s_i, the first column of the matrix is transmitted in time slot t=1 and the second column of the matrix is transmitted in time slot 2.

[0020] Maximum likelihood detection amounts to minimizing the decision statistic $\begin{array}{cc}\sum _{j=1}^m\left({\left|r_1^j-{\alpha }_{1,j}{s}_1-{\alpha }_{2,j}{s}_2\right|}^2+{\left|r_2^j+{\alpha }_{1,j}{s}_2^{*}-{\alpha }_{2,j}{s}_1^{*}\right|}^2\right)& \left(5\right)\end{array}$

[0021] over all possible values of s₁and s₂. The minimizing values in equation (5) are the receiver estimates of s₁and s₂, respectively. Expanding the above metric and deleting the terms that are independent of the codewords, it can be observed that the above minimization is equivalent to minimizing $\begin{array}{cc}-\sum _{j=1}^m\left({\Psi }_1+{\Psi }_2\right)+\left({\left|s_1\right|}^2+{\left|s_2\right|}^2\right)\sum _{j=1}^m\sum _{i=1}^2{\left|{\alpha }_{i,j}\right|}^2,& \left(6\right)\end{array}$

[0022] where

[0023] Ψ₁=[r₁^jα* _1,js*₁+(r₁^j)*α_1,js₁+r₂^jα*_2,js*₁+(r₂^j)*α_2,js₁]  (7)

[0024] and

[0025] Ψ₂=[r₁^jα*_2,js*₂+(r₁^j)*α_2,js₂−r₂^jα*_1,js*₂−(r₂^j)*α_1,js₂].  (8)

[0026] The above metric decomposes into the two parts $\begin{array}{cc}-\sum _{j=1}^m{\Psi }_1+{\left|s_1\right|}^2\sum _{j=1}^m\sum _{i=1}^2{\left|{\alpha }_{i,j}\right|}^2& \left(9\right)\\ \mathrm{and}& \\ -\sum _{j=1}^m{\Psi }_2+{\left|s_2\right|}^2\sum _{j=1}^m\sum _{i=1}^2{\left|{\alpha }_{i,j}\right|}^2,& \left(10\right)\end{array}$

[0027] where equation (9) is only a function of s₁, and equation (10) is only a function of s₂. Thus, the minimization of equation (5), which is derived from equation (2), is achieved by minimizing equations (9) and (10) separately. This, in turn, is equivalent to minimizing the decision statistic $\begin{array}{cc}{\left|\sum _{j=1}^m\left(r_1^j{\alpha }_{1,j}^{*}+{\left(r_2^j\right)}^{*}{\alpha }_{2,j}\right)-s_1\right|}^2+\left(-1+\sum _{j=1}^m\sum _{i=1}^2{\left|{\alpha }_{i,j}\right|}^2\right){\left|s_1\right|}^2& \left(11\right)\end{array}$

[0028] for detecting s₁, and the decision statistic $\begin{array}{cc}{\left|\sum _{j=1}^m\left(r_1^j{\alpha }_{2,j}^{*}+{\left(r_2^j\right)}^{*}{\alpha }_{1,j}\right)-s_2\right|}^2+\left(-1+\sum _{j=1}^m\sum _{i=1}^2{\left|{\alpha }_{i,j}\right|}^2\right){\left|s_2\right|}^2& \left(12\right)\end{array}$

[0029] for decoding s₂.

[0030] From a careful look at the complex vectors that make up the matrix of equation (4) it can be observed that the pair of constellation symbols (x₁,x₂) and (−x*₂,x*₁) are orthogonal to each other (i.e., (x₁,x₂)(−x*₂,x*₁)^H=0), where the superscript H denotes transpose conjugate (Hermetian), and, therefore, they can constitute the two orthogonal coordinates of a coordinate system. Viewed in this manner, any pair of complex vectors, such as constellation symbols pair X=(x₃,x₄), can be mapped onto the coordinate system defined by (x₁,x₂) and (−x*₂,x*₁), and expressed in this coordinate system as a vector

[0031] P_x=(A_x,B_x).  (13)

[0032] That is,

[0033] (x₃,x₄)=A_x(x₁,x₂)+B_x(−x*₂,x*₁),  (14)

[0034] where A_xis the dot product of (x₃,x₄) and (x₁,x₂), and B_xis the dot product of (x₃,x₄) and (−x*₂,x*₁). This yields

[0035] A_x=x₃x*₁+x₄x*₂,  (15)

[0036] and

[0037] B_x=−x₃x₂+x₄x₁.  (16)

[0038] Defining V as the set of all vectors P_xfrom signal pairs belonging to constellation A, it can be shown that, if the constellation A is restricted to phase shift keying (where the constellation points lie along the (power) unit circle), V has the following properties:

[0039] It has 2^2belements corresponding to the pairs (x₃,x₄) of constellation symbols.

[0040] All elements of V have unit length.

[0041] For any two distinct elements X and Y of V,

[0042] ∥P_x−P_y∥−∥X−Y∥.  (17)

[0043] The minimum distance between any two distinct elements of V is equal to the minimum distance of the 2^b-PSK constellation.

[0044] Now, given a block B of 2b bits, the first b bits are mapped into a constellation symbol a₃and the second two bits are mapped into a constellation symbol a₄. Employing an arbitrary, fixed, starting pair of (a₁,a₂) that belongs to constellation α (for example $a_1=a_2=\frac{1}{\sqrt{2}}$

[0045] the complex vector pair (a₃,a₄) is mapped to the coordinate system defined by the orthogonal vectors (a₁,a₂) and (−a*₂,a₁), to yield the vector P_x=(A_x,B_x) for X=(a₃,a₄), or P(B)=(A(B),B(B)), where

[0046] A(B)=a₃a*₁+a₄a*₂,  (18)

[0047] and

[0048] B(B)=−a₃a₂+a₄a₁.  (19)

[0049] Conversely, given A(B), and B(B), the pair (a₃,a₄) is recovered in a receiver that knows the pair of (a₁,a₂) by

[0050] (a₃,a₄)=A(B)(a₁,a₂)+B(B)(−a*₂,a*₁).  (20)

[0051] The block B is then constructed by inverse mapping of a₃and a₄. Thus, there is a direct mapping from constellation symbol a₃and a₄to A(B), and B(B).

[0052] In accordance with the principles disclosed above transmitting unit 10 of FIG. 1 includes element 12 that receives the input signals and maps the input signal blocks with mapping operator M. Operator M corresponds to the mapping from bits of the input signal block directly to the complex vectors A(B), and B(B). The mapped signals are applied to symbol computation element 13, which with assistance with transmitted symbols from the previous two time intervals, computes symbols from constellation α corresponding to the mapped input signal block, and transmits them over antennas 11-1 and 11-2. Those symbols are then fed back to delay element 14 in preparation for the mapping of the next input signal block. Thus, based solely on a₁and a₂, the transmitter begins the transmission with the sending of arbitrary symbols s₁and s₂at time slot 1 and symbols −s*₂and s*₁at time slot 2. These transmissions do not convey any information, but they are fed back to element 12, where they are used in the mapping of the next set of inputs, in an inductive manner, as effectively described above.

[0053] To illustrate, suppose that during a frame q (frames having 2 time slots each), symbols s_2q−1and s_2qare transmitted. More specifically, suppose symbols s_2q−1and s_2qare respectively transmitted from antenna 11-1 and 11-2, and at time slot 1, and the symbols −s*_2qand s_2q−1are respectively transmitted from antenna 11-1 and 11-2 at time slot 2 of frame q. Suppose further that at frame q+1, a block of 2b bits B_q+1arrives at element 12. According to the above, element 12 uses the mapping of the form expressed in equations (18) and (19) to obtain A(B_q+1) and B(B_q+1), and element 12 computes the constellation points

[0054] (s_2(q+1)−1,s_2(q+1))=(s_2q+1,s_2q+2)=A(B_q+1)(s_2q−1,s_2q)+B(B_q−1)(−s*_2q,s*_2q−1).  (21)

[0055] Then, symbols s_2q+1and s_2q+2are transmitted from antennas 11-1 and 11-2, respectively at time slot 1, and symbols −s*_2q+2and s*_2q+1are transmitted from antennas 11-1 and 11-2, respectively, at time slot 2 of frame q+1. These signals are also sent to element 14 in preparation of the encoding of frame q+2. This process is inductively repeated until the end of the frame (or end of transmission).

[0056] The decoding of signals received by unit 20 is performed in detector elements 22-j, which are coupled to antennas j. Within element 22-j there is a delay element 221-j and dot product generators 222-j and 223-j. Dot product generator 222-j develops the dot product of (r_2q+1,r*_2q+2).(r_2q−1,r*_2q) for the signals received at antenna j, and dot product generator 223-j develops the dot product of (r_2q+1,r*_2q+2).(r_2q,−r*_2q−1) for the signals received at antenna j.

[0057] Considering the outputs of element 21-1, and simplifying the notation by employing r_tfor r_t¹, η_t, α₁for η_t¹, α₁for α_1,1, and α₂for α_2,1, it can be observed that the signal pairs (r_2q+1,r*_2q+2), (r_2q−1,r*_2q), and (r_{2q,−r*2q−1}) can be expressed by

[0058] (r_2q+1,r*_2q+2)=(s_2q+1,s_2q+2)Λ(α₁,α₂)+N_2q+1,  (22)

[0059] (r_2q−1,r*_2q)=(s_2q−1,s_2q)Λ(α₁,α₂)+N_2q−1,  (23)

[0060] and

[0061] (r_2q,−r*_2q−1)=(−s*_2q,s*_2q−1)Λ(α₁,α₂)+N_2q,  (24)

[0062] where r_2q−1, r_2q, r_2q+1, and r_2q+2are the received signals, $\begin{array}{cc}\Lambda \left({\alpha }_1,{\alpha }_2\right)=\left(\begin{array}{cc}{\alpha }_1& {\alpha }_2^{*}\\ {\alpha }_2& -{\alpha }_1^{*}\end{array}\right),& \left(25\right)\end{array}$

[0063] and

[0064] N_2q−1=(η_2q−1,η*_2q).  (26)

[0065] Hence, taking the dot product of equations (23) and (22) within element 222-1 results in

[0066] (r_2q+1,r*_2q+2).(r_2q−1,r*_2q)=(s_2q+1,s_2q+2)Λ(α₁,α₂)Λ*

[0067] (α₁,α₂)(s*_2q−1,s*_2q)+(s_2q+1,s_2q+2)Λ(α₁α₂)N*_2q−1+

[0068] N_2q+1,Λ*(α₁,α₂)(s_2q−1,s_2q)*+N_2q+1N*_2q−1.  (27)

[0069] Expanding equation (27) results in an output for element 221-1 that is

[0070] r_2q+1r*_2q−1+r*_2q+2r_2q=

[0071] (|α₁|²+|α₂|²)(s_2q+

[0072] 1s*_2q−1+s*_2q+2s_2q)+(s_2q+1,s_2q+2)Λ

[0073] (α₁,α₂)N*_2q−1+N_2q+1Λ*(α₁,α₂)(s_2q−1,s_2q)*+N_2q+1N*_2q−1,  (28)

[0074] which reduces to

[0075] R₁=(|α₁|²+|α₂|²)A(B_q−1)+N₁,  (29)

[0076] where

[0077] N₁=(s_2q+1,s_2q+2)Λ(α₁,α₂)N*_2q−1+N_2q+1Λ* (α₁,α₂)(s_2q−1,s_2q)*+N_2q−0N_2q−1.  (30)

[0078] Similarly, carrying out the mathematics of equations (27)-(30) for the dot product of (r_2q+1,r*_2q+2).(r_2q,−r_2q−1) within element 223-1 reveals that the output of element 223-1 corresponds to R₂, where

[0079] R₂=(|α₁|²+|α₂|²)B(B_q−1)+N₂.  (31)

[0080] The vector pair (R₁,R₂) at the output of detector 22-1 can then be expressed by

[0081] (R₁,R₂)=(|α₁|²+|α₂|²)(A(B_q−1),B(B_q−1))+(N₁,N₂).  (32)

[0082] The procedure disclosed above for antenna 12-1 is used for each of the j receive antennas, as depicted in FIG. 1, yielding a set of vectors R₁^jand R₂^j, j=1,2, . . . , m that is applied to element 25, wherein the closest vector of V to $\left(\sum _{j=1}^m{R}_1^j,\sum _{j=1}^m{R}_2^j\right)$

[0083] is computed, following the approach disclosed above in connection with equations (5)-(12). Subsequently, the transmitted bits are computed by applying the inverse mapping M⁻¹in element 26.

[0084] The above discloses the principles of this invention by way of analysis for a transmitter having two transmit antennas. It should be realized that various modifications are possible to be incorporated without departing from the spirit and scope of this invention and, in particular, it should be understood that extension to arrangements where there are more than two antennas is straight forward using, for example, the codes taught in U.S. patent application Ser. No. 09/186,908, titled “Generalized Orthogonal Designs for Space-Time Codes for Wireless Communication,” which has the effective filing date of Nov. 11, 1997.