Adaptive receiving system and method for MIMO

Adaptive receiving system and method for MIMO
RE44219

Disclosed is an adaptive receiving mimo (multi input and multi output) system and method which decides a symbol detecting order so as to estimate the symbol having the minimum summation of weights of least square errors at the time of estimating the symbol for respective equalizers provided in parallel by the number of transmit antennas, and updates filter tap coefficients based on the RLS algorithm according to the detecting orders. Therefore, the filter tap coefficients are directly updated without tracking channels in the time-varying channel environment, and accordingly, detection performance very similar to those of the channel tracking and conventional V-BLAST scheme is provided with reduced complexity.

PTO Wrapper PDF
Dossier Espace Google

Patent RE44219
Priority Oct 21 2003
Filed Mar 18 2010
Issued May 14 2013
Expiry Apr 21 2024
Inventors Lee, Yong-…
Assg.orig Electronic…
Assg.curr Electronic…
Entity Small
Referenced by 0
References 4
Maint.: EXPIRED

BRIEF DESCRIPTION OF…
DETAILED DESCRIPTION…

0. 31. A wireless communication method through a communication channel, comprising:

canceling interference from receive symbol vectors and extracting transmit symbol vectors ; and

canceling signal interference from the detected signals and the receive symbol vectors according to a filter tap coefficient and a symbol detecting order, wherein a cross correlation vector is obtained by crossing a first value and a second value, the first value obtained when the receive symbol vector is passed through a feedforward filter, and the second value obtained when the detected signal is fed back through a feedback filter.

0. 30. A wireless communication device through a communication channel, comprising:

a linear equalizer configured to cancel interference from receive symbol vectors and extract transmit symbol vectors; and

a feedback equalizer configured to cancel signal interference from the signals detected by the linear equalizer and the receive symbol vectors according to a filter tap coefficient and a symbol detecting order, wherein a cross correlation vector is obtained by crossing a first value and a second value, the first value obtained when the receive symbol vector is passed through a feedforward filter, and the second value obtained when the detected signal is fed back through a feedback filter.

0. 25. A wireless communication method for transmitting and receiving transmit and receive symbol vectors through a communication channel, comprising:

(a) canceling interference from the receive symbol vectors and extracting transmit symbol vectors by performing a detection; and

(b) canceling signal interference from the detected signals and the receive symbol vectors according to a filter tap coefficient and a symbol detecting order updated for each predetermined period, and detecting the transmit symbol vectors, wherein a cross correlation vector is obtained by crossing a first value and a second value to define a cross correlation matrix, the first value obtained when the receive symbol vector is passed through a feedforward filter, and the second value obtained when the detected signal is fed back through a feedback filter.

0. 18. A wireless communication device for transmitting and receiving transmit and receive symbol vectors through a communication channel, comprising:

a linear equalizer for canceling interference from the receive symbol vectors and extracting transmit symbol vectors by performing a detection: and

a plurality of feedback equalizers for canceling signal interference from the signals detected by the linear equalizer and the receive symbol vectors according to a filter tap coefficient and a symbol detecting order updated for each predetermined period, and detecting the transmit symbol vectors, wherein a cross correlation vector is obtained by crossing a first value and a second value to define a cross correlation matrix, the first value obtained when the receive symbol vector is passed through a feedforward filter, and the second value obtained when the detected signal is fed back through a feedback filter.

0. 19. An adaptive receiving multi input and multi output method for allowing an equalizer to detect transmit symbol vectors when the transmit symbol vectors transmitted from multiple transmit antennas are input as receive symbol vectors through multiple receive antennas, comprising:

(a) the equalizer detecting an error signal by allowing the initial receive symbol vector to be passed through a feedforward filter, and detecting the transmit symbol vectors, when the receive symbol vectors are input; and

(b) allowing the equalizer to apply an optimal filter tap coefficient to the feedforward filter and a feedback filter, canceling interference from the receive symbol vectors and detected signals according to a predetermined symbol detecting order, and detecting the transmit symbol vectors when (a) is finished, and using a cross correlation vector obtained by crossing a first value and a second value to define a cross correlation matrix g(n), the first value being obtained when the receive symbol vector is passed through the feedforward filter, and the second value being obtained when the detected signal is fed back through the feedback filter.

7. An adaptive receiving method in an adaptive receiving system for a mimo (multi input and multi output) for allowing an equalizer to detect transmit symbol vectors when the transmit symbol vectors transmitted from multiple transmit antennas are input as receive symbol vectors through multiple receive antennas, comprising:

(a) the equalizer detecting an error signal by allowing the initial receive symbol vector to be passed through a feedforward filter, and detecting the transmit symbol vectors according to a predetermined decision reference, when the receive symbol vectors are input; and

1. An adaptive receiving mimo (multi input and multi output) system for transmitting and receiving transmit and receive symbol vectors through channels between multiple transmit and receive antennas, comprising:

a linear equalizer for canceling interference added when passed through the channel from the receive symbol vectors and extracting transmit symbol vectors by performing an initial detection;

a plurality of parallel decision feedback equalizers for canceling signal interference from the signals detected by the linear equalizer and the receive symbol vectors according to a filter tap coefficient and a symbol detecting order updated for each predetermined period, and detecting the transmit symbol vectors, wherein a cross correlation vector is obtained by crossing a first value and a second value to define a cross correlation matrix, the first value obtained when the receive symbol vector is passed through the feedforward filter, and the second value obtained when the detected signal is fed back through the feedback filter; and

an adaptive block for deciding the symbol detecting orders of the linear equalizer and the parallel decision feedback equalizers, and updating the filter tap coefficients based on the RLS (recursive least square) algorithm according to the decided symbol detecting orders.

2. The adaptive receiving system of claim 1, further comprising a reordering unit for reordering the signals sequentially detected through the equalizer according to the symbol detecting order on a plurality of receive symbol vectors decided by the adaptive block.

3. The adaptive receiving system of claim 1, wherein the linear equalizer comprises:

a feedforward filter for receiving the first receive symbol vector from among the receive symbol vectors received through the receive antennas, and filtering error signals; and

a decision unit for applying the optimal filter tap coefficient to signals output by the feedforward filter according to a predetermined decision reference, and outputting a decision value.

4. The adaptive receiving system of claim 1, wherein the parallel decision feedback equalizer comprises:

a feedforward filter for receiving the receive symbol vector from the receive antennas, and performing filtering by using a feedforward filter tap coefficient decided by the adaptive block;

a feedback filter for receiving detected signals from among the receive symbol vectors, and performing filtering by using a feedback filter tap coefficient decided by the adaptive block;

an adder for adding the respective signals output by the feedforward filter and the feedback filter, and outputting added values; and

a decision unit for applying the optimal filter tap coefficient to the added values output by the adder according to a predetermined decision reference, and outputting decision values.

5. The adaptive receiving system of claim 4, wherein the feedback filter increases by an order by one when a detection on the receive symbol vectors from the second symbol to the last symbol is repeated.

6. The adaptive receiving system of claim 1, wherein the adaptive block defines the reference for deciding the tap coefficient by errors of between the transmit symbol vector transmitted by the transmit antenna and the output of the equalizer, and defines the optimal filter tap coefficient as a filter coefficient for minimizing the errors.

8. The adaptive receiving method of claim 7, wherein (b)comprises: allowing the feedforward filter and the feedback filter to update the optimal filter tap coefficient based on the RLS (recursive least square) algorithm, and deciding the symbol detecting order on the equalizer so that a symbol for minimizing the summation of weights of square errors may be estimated.

9. The adaptive receiving method of claim 8, wherein the optimal filter tap coefficient is a filter coefficient for minimizing error between the transmit symbol vector transmitted by the transmit antenna and the estimated transmit symbol vector output by the equalizer.

10. The adaptive receiving method of claim 8, wherein the optimal filter tap coefficient w_tj(n), i=1, . . . , M is recursively found by using subsequent equations:

q₁(n)=Φ_i⁻¹(n−1)y_t,i(n)

k_{i} (n) = \frac{λ^{- 1} q_{i} (n)}{1 + λ^{- 1} y_{t, i}^{H} (n) q_{i} (n)}

Φ_i⁻¹(n)=λ⁻¹Φ_i⁻¹(n−1)−λ⁻¹k₁(n)q_l^H(n)

w_t,l(n)+w_t,l(n−1)+k_l(n)ξ_l*(n)

where ξ_l(n) is an a priori estimation error which is given as

ξ_l(n)={circumflex over (d)}_k_i(n)−w_l,t^H(n−1)y_t,l(n).

11. The adaptive receiving method of claim 8, wherein the symbol detecting order k_iof the equalizer is decided to detect the symbol for minimizing the summation of weights of square errors:

k_{i} = \arg \min_{j} ɛ_{i, j} (n) and ɛ_{i, j} (n) = \sum_{l = 1}^{n} λ^{n - 1} {\langle d_{j} (l) - w_{t, i}^{H} (l) y_{t, i} (l) \rangle}^{2}

where w_t,i(n) is an integrated filter coefficient vector of the equalizer, and y_t,i(n) is an integrated input signal vector, and d(n) is a transmit symbol vector.

12. The adaptive receiving method of claim 7, wherein (a) comprises deciding a value {circumflex over (d)}(n) on the first transmit symbol vector by using a value {tilde over (d)}_k₁(n) generated when the receive symbol vector y_ti(n) is input to the feedforward filter and is then output therefrom:

{tilde over (d)}_k₁(n)=w_l,t^H(n−1)y_t,i(n)

{circumflex over (d)}_k₁(n)=decision{{tilde over (d)}_k₁(n)}

y_t,i+1(n)=[y^t(n),{circumflex over (d)}_k₁(n), . . . , {circumflex over (d)}_k₁(n)]^t.

13. The adaptive receiving method of claim 7, wherein (b) comprises:

(i) the cross correlation vector is obtained by an equation:

g(n)=λG(n−1)+[y^T(n),{circumflex over (d)}(n)]^T{circumflex over (d)}^H(n);

(ii) calculating the optimal filter tap coefficient of {v_1,j(n), j=1,2, . . . , M} applied to the feedforward filter and the feedback filter when (i) is finished;

q₁(n)=Φ₁⁻¹(n−1)y(n)

k_{1} (n) = \frac{λ^{- 1} q_{1} (n)}{1 + λ^{- 1} y^{H} (n) q_{1} (n)}

Φ_i⁻¹(n)=λ⁻¹Φ_i⁻¹(n−1)−λ⁻¹k₁(n)q_l^H(n)

v_1,l(n)=v_1,l(n−1)+k₁(n)(d_l(n)−v_1,l^H(n−1)y(n))

(iii) deciding the symbol detecting orders of the equalizer, and updating the filter tap coefficients when (ii) is finished; and

(iv) applying the symbol detecting order decided in (iii) and the filter tap coefficient to the next receive symbol vector, detecting the transmit symbol vector, and repeating (iv).

14. The adaptive receiving method of claim 13, wherein (iii) comprises obtaining the cross correlation vector z_ij(n), calculating the summation ε_ij(n) of weights of square errors, and deciding the symbol detecting order 30

k₁=arg

ε_i,j(n)

of each equalizer:

z_i,j(n)=[g_l,j(n), . . . g_N,j(n),g_N+k_lj(n)]^T

ε_i,j(n)=g_N+J,J(n)z_i,j(n)

w_l,l(n)=v_i,k_l(n)

v_{i + 1, j} (n) = [\begin{matrix} v_{i, j} (n) \\ 0 \end{matrix}] + \frac{g_{N + k_{i}, j} (n) - w_{t, i}^{H} (n) z_{i, j} (n)}{ɛ_{i, k_{i}} (n)} [\begin{matrix} - w_{t, i} (n) \\ 1 \end{matrix}]

where w_t,i(n) is an integrated filter tap coefficient.

15. The adaptive receiving method of claim 7, wherein (b) comprises deciding the symbol detecting order of the equalizer at intervals of a constant γ.

16. The adaptive receiving method of claim 15, wherein when inputting the n^threceive symbol vector, (i) when ‘n’ is a multiple of the constant y, obtaining the cross correlation vector z_i,j(n), calculating the summation ε_ij(n) of weights of square errors, and deciding the symbol detecting order 30

k₁=arg

ε_i,j(n)

of each equalizer:

z_i,j(n)=[g_l,j(n), . . . g_N,j(n),g_N+k_lj(n_n)]^T

ε_i,j(n)=g_N+J,J(n)z_i,j(n)

w_l,l(n)=v_i,k₁(n)

v_{i + 1, j} (n) = [\begin{matrix} v_{i, j} (n) \\ 0 \end{matrix}] + \frac{g_{N + k_{i}, j} (n) - w_{t, i}^{H} (n) z_{i, j} (n)}{ɛ_{i, k_{i}} (n)} [\begin{matrix} - w_{t, i} (n) \\ 1 \end{matrix}]

where w_t,i(n) is an integrated filter tap coefficient; (ii) when ‘n’ is not a multiple of the constant γ, updating the filter tap coefficient by use of subsequent equations, and using the previous (n−1) symbol detecting order for the symbol detecting order of each equalizer:

q₁(n)=Φ₁⁻¹(n−1)y(n)

q_{i + 1} (n) = [\begin{matrix} q_{i} (n) \\ 0 \end{matrix}] + c_{i} (n - 1) ɛ_{i} (n) [\begin{matrix} - w_{t, i} (n - 1) \\ 1 \end{matrix}]

k_{i} (n) = \frac{λ^{- 1} q_{i} (n)}{1 + λ^{- 1} y_{t, i}^{H} (n) q_{i} (n)}

w_t,j(n)=w_t,j(n−1)+k₁(n)({circumflex over (d)}_k₁(n)−w_t,j^H(n−1)y_t,j(n))

17. The adaptive receiving method of claim 7, wherein (b) comprises updating the filter tap coefficient and deciding the symbol detecting order for each symbol time according to the speed of the channel varying during a predetermined period, and deciding the symbol detecting order once with the interval of the constant γ, and updating the filter tap coefficient and maintaining the symbol detecting order during a residual period.

0. 20. The wireless communication device of claim 18, wherein the transmission and reception of the transmit and receive symbol vectors are performed between multiple transmit and receive antennas.

0. 21. The wireless communication device of claim 20, further comprising:

an adaptive block for deciding the symbol detecting orders of the linear equalizer and the feedback equalizers, and updating the filter tap coefficients according to the decided symbol detecting orders.

0. 22. The wireless communication device of claim 21, wherein the adaptive block updates the filter tap coefficients based on the recursive least square (RLS) algorithm.

0. 23. The wireless communication device of claim 22, further comprising a reordering unit for reordering the signals sequentially detected through the linear equalizer according to the symbol detecting order on a plurality of receive symbol vectors decided by the adaptive block.

0. 24. The adaptive receiving multi input and multi output method of claim 19, wherein the equalizer detecting an error signal comprises:

using the cross correlation vector obtained by crossing a first value and a second value to define a cross correlation matrix g(n).

0. 26. The wireless communication method of claim 25, wherein the transmission and reception of the transmit and receive symbol vectors are performed between multiple transmit and receive antennas.

0. 27. The wireless communication method of claim 26, further comprising:

deciding the symbol detecting orders of the linear equalizer and the feedback equalizers, and updating the filter tap coefficients according to the decided symbol detecting orders.

0. 28. The wireless communication method of claim 27, wherein the updating of the filter tap coefficients is performed based on the recursive least square (RLS) algorithm.

0. 29. The wireless communication method of claim 28, further comprising:

reordering the signals sequentially detected according to the symbol detecting order on a plurality of receive symbol vectors.

of each equalizer; and

(ii) when ‘n’ is not a multiple of the constant γ, updating the filter tap coefficient by use of subsequent equations, and using the previous (n−1) symbol detecting order for the symbol detecting sequence of each equalizer.

The step (b) comprises updating the filter tap coefficient and deciding the symbol detecting order for each symbol time according to the speed of the channel varying during a predetermined period, and deciding the symbol detecting order once with the interval of the constant γ, and updating the filter tap coefficient and maintaining the symbol detecting order during a residual period.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the description, serve to explain the principles of the invention:

FIG. 1 shows a block diagram of an adaptive receiving system for the MIMO according to a preferred embodiment of the present invention;

FIG. 2 shows a detailed schematic diagram for an equalizer included in the configuration of FIG. 1;

FIG. 3 shows a process for updating optimal filter tap coefficients in an adaptive receiving method for the MIMO according to a preferred embodiment of the present invention;

FIG. 4 shows a simplified process for updating optimal filter tap coefficients compared to FIG. 3; and

FIGS. 5 to 7 show graphs for comparing BER performance between the preferred embodiment of the present invention and the existing V-BLAST scheme.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

In the following detailed description, only the preferred embodiment of the invention has been shown and described, simply by way of illustration of the best mode contemplated by the inventor(s) of carrying out the invention. As will be realized, the invention is capable of modification in various obvious respects, all without departing from the invention. Accordingly, the drawings and description are to be regarded as illustrative in nature, and not restrictive.

A MIMO adaptive receiving system according to a preferred embodiment of the present invention will be described in detail with reference to FIG. 1.

FIG. 1 shows a block diagram of an adaptive receiving system for the MIMO according to a preferred embodiment of the present invention.

As shown, the adaptive receiving system comprises an equalizer 100, an adaptive block 200, and a reordering unit 300. The equalizer 100 is classified as a linear equalizer 150, shown in FIG. 2, for detecting the initial transmit symbol vector, and a parallel equalizer for detecting residual transmit symbol vectors.

The equalizer 100 is coupled in parallel by the number of the transmit antennas, and is classified as a linear equalizer and parallel equalizers for ease of description.

The i^thequalizer comprises an N^th-order feedforward filter 110, an (i−1)^th-order feedback filter 120, an adder 130, and a decision unit 140.

The feedforward filter 110 receives (N×1) receive symbol vectors through N receive antennas and filters error signals, and the feedback filter 120 filters the error signals from the detected signals from among (N×1) receive symbol vectors and feeds results back.

The adder 130 adds respective signals output by the feedforward filter 110 and the feedback filter 120 and outputs an added value, and the decision unit 140 applies the optimal filter tap coefficient to the added value according to a predetermined decision reference and outputs a decision value, that is, a transmit symbol vector.

The adaptive block 200 updates the tap coefficient and the symbol detecting order of the equalizer 100 based on the RLS (recursive least square) algorithm for each symbol, and decides the symbol detecting sequence of the equalizer 100 by using a summation of square errors.

The reordering unit 300 reorders transmit symbol vectors sequentially detected through the equalizer 100 according to a symbol detecting order decided by the adaptive block 200.

The configuration of the adaptive receiving system for the MIMO will now be described.

When using M transmit antennas and N receive antennas, an (M×1) transmit symbol vector d(n)=[d₁(n), d₂(n), . . . , d^M(n)]^Tis transmitted through the transmit antennas, and an (N×1) receive symbol vector y(n)=[y₁(n), y₂(n), . . . , y_N(n)]^Tis received through the receive antennas.

When the transmit symbol vector is passed through an (N×M) channel matrix H(n) caused by channel gains between the respective transmit and receive antennas, and the noise vector u(n)=[u₁(n), u₂(n), . . . , u_N(n)]^Tis added thereto, the receive symbol vector is given as Equation 1.
y(n)=H(n)d(n)+u(n) Equation 1

The receive symbol vector is input to the equalizer 100, is passed through the feedforward filter 110 which corresponds to the nulling vector of the V-BLAST scheme, interference is cancelled from the receive symbol vector by using detected signals and the feedback filter 120, and is detected.

The symbol detecting order and the filter tap coefficient vector of the equalizer 100 are updated for each symbol time, and the operation for updating the filter tap coefficient is referred to as a time update.

FIG. 2 shows a detailed schematic diagram for an equalizer included in the configuration of FIG. 1.

As shown, the linear equalizer 150 comprises a feedforward filter 110 and a decision unit 140, and the parallel equalizer 100 comprises a feedforward filter 110, a feedback filter 120, a correlator 130, and a decision unit 140.

The first parallel equalizer provided after the linear equalizer 150 has a first-order feedback filter using a detected {circumflex over (d)}_k₁, the second parallel equalizer has a second-order feedback filter using {circumflex over (d)}_k₁and {circumflex over (d)}_k₂, and the last equalizer has an (M−1)^th-order feedback filter.

Referring to FIG. 2, {w_f,i(n)} is an N^th-order feedforward filter tap coefficient vector, and {w_h,i(n)} is an (i−1)^thorder feedback filter tap coefficient vector.

The concurrently transmitted M transmit symbol vectors are sequentially detected through the equalizer 100, and in further detail, they are detected by the linear equalizer 150 in the first stage, and then are detected by the parallel decision feedback equalizers 100 which increase the order of the feedback filter 120 by one as the detection is repeated.

A set for indicating the symbol detecting order of the transmit symbol vectors d(n) is defined as S={k₁, k₂, . . . , k_M} where s is a set generated by changing the sequence of elements in {1, 2, . . . , M}.

The parallel decision feedback equalizers 100 have a superimposed configuration of equalizers in parallel, the number of the equalizers corresponding to the number of transmit antennas, and the equalizers starting from the top equalizer to the bottom equalizer sequentially detect the transmit symbol vectors according to the symbol detecting order.

The result of detecting the transmit symbol vectors by the decision unit 140, that is, the decision unit is defined to be {circumflex over (d)}(n)=[{circumflex over (d)}_k₁(n), {circumflex over (d)}_k₁(n), . . . , {circumflex over (d)}_k_i-1(n)]^T.

The adaptive block 200 updates coefficients of the feedforward filter 110 and the feedback filter 120 for each symbol time based on the RLS algorithm, calculates summation of weights of square errors for the respective equalizers 100, and decides the symbol detecting order.

The integrated filter coefficient vector w_t,i(n) and the integrated input signal vector y_t,i(n) are defined in Equations 2 and 3.

$\begin{matrix} w_{t, i} (n) = {\begin{matrix} w_{f, i} (n), & i = 1 \\ {[w_{f, i}^{T} (n), w_{b, i}^{T} (n)]}^{T}, & i = 2, \dots, M \end{matrix} & Equation 2 \\ y_{t, i} (n) = {\begin{matrix} y (n), & i = 1 \\ {[y^{T} (n), {\hat{d}}_{k_{1}}, \dots, {\hat{d}}_{k_{i} - 1}]}^{T}, & i = 2, \dots, M \end{matrix} & Equation 3 \end{matrix}$

An operation of the MIMO adaptive receiving system according to the preferred embodiment of the present invention will be described with reference to a drawing.

FIG. 3 shows a process for updating optimal filter tap coefficients in an adaptive receiving method for the MIMO according to a preferred embodiment of the present invention.

As shown, it is required to find the optimal filter tap coefficient of the equalizer 100 when the M transmit symbol vectors are input in the adaptive receiving method for the MIMO.

For this, the adaptive block 200 defines a decision reference as an error of between the transmit symbol vector transmitted from the transmit antenna and the equalizer output of the receiver, and defines the filter coefficient for minimizing the error as the optimal filter tap coefficient.

To define the optimal filter tap coefficient, the outputs of the i^thequalizer is defined as Equation 4, and the cost function is given in Equation 5.
{tilde over (d)}_k_i(n)=w_t,i^H(n−1)y_t,i(n) Equation 4

$\begin{matrix} J_{i} (n) = \sum_{l = 1}^{n} λ^{n - l} {\langle {\hat{d}}_{k_{i}} (l) - w_{t, i}^{H} (l) y_{t, i} (l) \rangle}^{2} & Equation 5 \end{matrix}$

The optimal filter tap coefficient for minimizing J_i(n) is given as Equation 6.
w_t,i(n)=Φ_i⁻¹(n)z_i,k_i(n) Equation 6

where Φ_i(n) is an autocorrelation matrix, and z_i,k_i(n) is a cross correlation vector, respectively given as Equations 7 and 8.

$\begin{matrix} Φ_{i} (n) = \sum_{l = 1}^{n} λ^{n - l} y_{t, i} (l) y_{t, i}^{H} (l) & Equation 7 \\ z_{i, j} (n) = \sum_{l = 1}^{n} λ^{n - l} y_{t, i} (l) {\hat{d}}_{j}^{*} (l) & Equation 8 \end{matrix}$

The optimal filter tap coefficient is found recursively by using the RLS algorithm as given in Equation 9.

$\begin{matrix} q_{i} (n) = Φ_{i}^{- 1} (n - 1) y_{t, i} (n) & Equation 9 \\ k_{i} (n) = \frac{λ^{- 1} q_{i} (n)}{1 + λ^{- 1} y_{t, i}^{H} (n) q_{i} (n)} \\ Φ_{i}^{- 1} (n) = λ^{- 1} Φ_{i}^{- 1} (n - 1) - λ^{- 1} k_{i} (n) q_{i}^{H} (n) \\ w_{t, i} (n) = w_{t, i} (n - 1) + k_{i} (n) ξ_{i}^{*} (n) \end{matrix}$

where ξ_i(n) is an a priori estimation error which is given in Equation 10.
ξ_i(n)={circumflex over (d)}_k_i(n)−w_t,i^H(n−1)y_t,i(n)| Equation 10

A large amount of calculation of the optimal filter tap coefficient using the RLS algorithm is needed since the RLS algorithm is independently applied to the M equalizer filter tap coefficients w_t,i(n), i=1, . . . , M.

When the autocorrelation matrix Φ_i+1(n) is defined as Equation 11, Φ₂⁻¹(n), . . . , Φ_M⁻¹(n) are recursively calculated by using the autocorrelation matrix through Equation 12.

$\begin{matrix} Φ_{i + 1} (n) = [\begin{matrix} Φ_{i} (n) & z_{i, k_{i}} (n) \\ z_{i, k_{i}}^{H} (n) & a_{k_{i}} (n) \end{matrix}] & Equation 11 \\ Φ_{i + 1}^{- 1} (n) = [\begin{matrix} Φ_{i}^{- 1} (n) + c_{i} (n) w_{t, i} (n) w_{t, i}^{H} & - c_{i} (n) w_{t, i} (n) \\ - c_{i} (n) w_{t, i}^{H} (n) & c_{i} (n) \end{matrix}] & Equation 12 \end{matrix}$

where it is given that α_j

$α_{j} (n) = \sum_{l = 1}^{n} λ^{n - l} {\langle {\hat{d}}_{j} (l) \rangle}^{2} in Equation 11, and$ $c_{i} (n) = \frac{1}{α_{k_{i}} (n) - z_{i, k_{i}}^{H} (n) w_{t, i} (n)} in Equation 12.$

Since it is given that y_t,i+1(n)=[y_t,i^T(n), {circumflex over (d)}_k_i(n)]^Tfrom Equations 11 and 12, q_i+1(n) is recursively calculated as expressed in Equation 13 when substituting Equation 12 and an expansion of y_t,i+1(n) for q_i(n) of Equation 9. Here, note that q₁(n)=Φ₁⁻¹(n−1)y(n).

$\begin{matrix} q_{i + 1} (n) = [\begin{matrix} q_{i} (n) \\ 0 \end{matrix}] + c_{i} (n - 1) ξ_{i} (n) [\begin{matrix} - w_{t, i} (n - 1) \\ 1 \end{matrix}] & Equation 13 \end{matrix}$

Equation 13 is applicable when the symbol detecting order is known.

The symbol detecting order is to be decided after the optimal filter coefficient of the equalizer is calculated.

In general, it is known as the symbol detecting sequence of minimizing the symbol error probability in the V-BLAST scheme to detect the signal which has the largest SNR (signal-to-noise ratio) in each stage. An equivalent SNR is found by using the summation of weights of square errors given as Equation 14 in the RLS algorithm.

$\begin{matrix} ɛ_{i, j} (n) = \sum_{l = 1}^{n} λ^{n - l} {\langle d_{j} (l) - w_{t, i}^{H} (l) y_{t, i} (l) \rangle}^{2} & Equation 14 \end{matrix}$

The equalizer 100 detects the symbol which minimizes the summation of weights of square errors given as Equation 14 in each detecting stage as given in Equation 15.
k_i=arg custom character ε_i,j(n) Equation 15

When it is defined that v_i,j(n)=Φ_i⁻¹(n)z_i,j(n), v_i,jcorresponds to Equation 6 considering that k_iis changed to j in w_i,j(n) of Equation 6. The summation of the square error is given as Equation 16.
ε_i,j(n)=α_j(n)−v_i,j^H(n)z_i,j(n) Equation 16

For example, when the number of transmit antennas is four, and the symbol detecting sequence at the receiver is given as S={4, 3, 2, 1}, Φ_i⁻¹is recursively found using Equation 12, the optimal symbol detecting sequence is decided using Equation 15, and the optimal filter tap coefficient v_i,j(n) is found as shown in FIG. 3.

FIG. 4 shows a simplified process for updating the optimal filter tap coefficient compared to FIG. 3.

The method for updating the optimal filter tap coefficient shown in FIG. 4 proposes a method for finding v_i,j(n) without updating Φ_i⁻¹since the method for updating the optimal filter tap coefficient shown in FIG. 3 requires a relatively large amount of calculation for recursively finding Φ_i⁻¹and v_i,j(n).

Assuming that the calculation on v_i,j(n)=Φ_i⁻¹(n)z_i,j(n) is finished, z_i,j(n) is defined as Equation 17 since Equation 12 for calculating Φ_i+1⁻¹and an update equation of z_i+1,j(n) are needed to calculate v_i+1,j(n)=Φ_i+1⁻¹(n)z_i+1,j(n).
z_i,j(n)=[g_1,j(n), . . . , g_N,j(n), g_N+k₁_,j(n), . . . , g_N+k_i−1_,j(n)]^T Equation 17
Noting that

$g_{i, j} (n) = {\begin{matrix} \sum_{l = 1}^{n} λ^{n - l} y_{i} (n) d_{j}^{*} (n), & 1 \leq i \leq N \\ \sum_{l = 1}^{n} λ^{n - l} d_{i - N} (n) d_{j}^{*} (n), & N + 1 \leq i \leq N + M \end{matrix},$

z_i+1,j(n) can be given as an update equation of Equation 18.
z_i+1,j(n)=[z_i,j^T(n), g_N+k_i_,j(n)]^T Equation 18

Since α_j(n) corresponds to g_N+j,j(n), the summation of square error can be given as Equation 19.
ε_i,j(n)=g_N+j,j(n)−v_i,j^H(n)z_i,j(n) Equation 19

Also, an (N+M)×N cross correlation matrix is defined in Equation 20 by using Equation 8 which has defined z_i,j(n).

$\begin{matrix} G (n) = \sum_{l = 1}^{n} {λ^{n - l} [y^{T} (n), {\hat{d}}^{T} (n)]}^{T} {\hat{d}}^{H} (n) & Equation 20 \end{matrix}$

Noting from Equation 20 that g_i,j(n) is the (i,j)^thelement of G(n), z_i,j(n) in Equation 18 is obtained through G(n). In addition, a detected transmit symbol vector {circumflex over (d)}(n) is required so as to obtain G(n).

The above-noted (n−1) filter tap coefficient and the symbol detecting order are used as shown in Equations 21 and 22 for the transmit symbol vector {circumflex over (d)}(n).
{tilde over (d)}_k_i(n)=w_t,i^H(n−1)y_t,i(n) Equation 21
{circumflex over (d)}_k_i(n)=decision {{tilde over (d)}_k_i(n)} Equation 22

When G(n) is given, Equations 12 and 18 are used to recursively find v_i+1,j(n) as Equation 23.

$\begin{matrix} \begin{matrix} v_{i + 1, j} (n) = Φ_{i + 1}^{- 1} (n) z_{i + 1, j} (n) \\ = [\begin{matrix} v_{i, j} (n) \\ 0 \end{matrix}] + \frac{g_{N + k_{i}, j} (n) - w_{i, j}^{H} (n) z_{i, j} (n)}{ɛ_{i, k_{i}} (n)} \\ [\begin{matrix} - w_{t, i} (n) \\ 1 \end{matrix}] \end{matrix} & Equation 23 \end{matrix}$

In Equation 23, note that jεS_i+1=S_i−{k_i}, S₁={1, . . . , M}, and v_1,j(n)=Φ₁⁻¹(n)z_1,j(n), j=1,2, . . . , M is calculated according to the existing method.

The detecting process on the adaptive receiving method for the MIMO is summarized as follows.

In the first stage, all the parameters are initialized so as to start the detection.
n=1
k_i=i, for all i
Φ₁⁻¹(0)=δ⁻¹I, G(0)=0
w_f,i(0)=1, w_b,i(0)=0, v_1,i(0)=0, for all i

where δ is a small positive constant.

In the second stage, the receive symbol vector input through a receive antenna is output through the feedforward filter 110 and the feedback filter 120, and the decision unit 140 outputs a decision value, that is, a transmit symbol vector where i=1, 2, . . . , M.
{tilde over (d)}_k_i(n)=w_t,i^H(n−1)y_t,i(n)
{circumflex over (d)}_k_i(n)=decision{{tilde over (d)}_k_i(n)}
y_t,i+1(n)=[y^T(n), {circumflex over (d)}_k_i(n), . . . , {circumflex over (d)}_k_i(n)]^T

In the third stage, the adder 130 uses output values of the feedforward filter 110 and the feedback filter 120 to update the cross correlation matrix.
G(n)=λG(n−1)+[y^T(n), {circumflex over (d)}^T(n)]^T{circumflex over (d)}^H(n)

In the fourth stage, the adaptive block 200 updates the filter tap coefficient and the detecting order for each symbol time.

(I) Calculation of {v_1,j(n), j=1, 2, . . . , M} by using a time-update of filter tap coefficients
q₁(n)=Φ₁⁻¹(n−1)y(n)

$k_{1} (n) = \frac{λ^{- 1} q_{1} (n)}{1 + λ^{- 1} y^{H} (n) q_{1} (n)}$
Φ₁⁻¹(n)=λ⁻¹Φ₁⁻¹(n−1)−λ⁻¹k₁(n)q₁^H(n)
v_1,j(n)=v_1,j(n−1)+k₁(n)(d_j(n)−v_1,j^H(n−1)y(n))*

(II) Decision of symbol detecting sequences and update of filter tap coefficients

i) Obtainment of z_i,j(n)=[g_1,j(n), . . . , g_N,j(n), g_N+k₁_,j(n), . . . , g_N+k_i−1,j(n)]^T

ii) Calculation of the summation of the square errors
ε_i,j(n)=g_N+j,i(n)−v_i,j^H(n)z_i,j(n)

iii) Decision of the symbol detecting sequence
k_i=arg custom character ε_i,j(n).
w_t,i(n)=v_i,k_i(n)

$v_{i + 1, j} (n) = [\begin{matrix} v_{i, j} (n) \\ 0 \end{matrix}] + \frac{g_{N + k_{i}, j} (n) - w_{t, i}^{H} (n) z_{i, j} (n)}{ɛ_{i, k_{i}} (n)} [\begin{matrix} - w_{t, i} (n) \\ 1 \end{matrix}]$

In the fifth stage (n=n+1), the stages starting from the second stage are repeated until all the detections on the M transmit symbol vectors transmitted from the transmit antenna are finished.

In algorithm 1, the second stage can be omitted since previously known data are transmitted during the initial period. When the terminal moves slowly and the channel is varying slowly, it frequently occurs that the symbol detecting order is not changed for each symbol time, and the symbol detecting order used during detecting d(n−1) is not varied while detecting d(n).

In algorithm 2, a constant γ is defined, the symbol detecting order is established for each time with the interval of γ, and the predefined symbol detecting order is maintained in another time without updating the symbol detecting order for each symbol time as shown in algorithm 1.

Similar to algorithm 1, but (II) of the fourth stage is modified as below.

In (II) for updating the filter tap coefficients of the fourth stage, (II) is performed when ‘n’ is a multiple of γ, and w_t,i(n) is calculated by using Equation 24 when ‘n’ is not a multiple of γ.
q₁(n)=Φ₁⁻¹(n−1)y(n) Equation 24

$q_{i + 1} (n) = [\begin{matrix} q_{i} (n) \\ 0 \end{matrix}] + c_{i} (n - 1) ξ_{i} (n) [\begin{matrix} - w_{t, i} (n - 1) \\ 1 \end{matrix}]$ $k_{i} (n) = \frac{λ^{- 1} q_{i} (n)}{1 + λ^{- 1} y_{t, i}^{H} (n) q_{i} (n)}$
w_t,i(n)=w_t,i(n−1)+k_i(n)({circumflex over (d)}_k_i(n)−w_t,i^H(n−1)y_t,i(n))*

It will be described that how much of the amount of calculation in the adaptive receiving system and method for the MIMO according to the preferred embodiment of the present invention is reduced in the time-varying channel environment compared to the existing V-BLAST method.

It is assumed that the V-BLAST method uses the method of “Modified Decorrelating Decision-Feedback Detection of BLAST Space-Time System,” in Proc. ICC 2002 by Wei Zha and Steven D. Blostein which provides simple calculation, and channel estimation results are tracked by using the conventional RLS method.

Assuming that the number of transmit antennas and that of receive antennas are respectively M, the number of times of multiplication and addition of complex numbers is given as follows according to whether the symbol detecting order is varied.

The number of times of multiplication and addition of complex numbers is given as Table 1 when the symbol detecting order is varied while detecting d(n) and d(n+1).

TABLE 1

	Multiplication of	Addition of
	complex numbers	complex numbers

V-BLAST + RLS channel tracking	7/3 M³	5/3 M³
Proposed MIMO-DFE	4/3 M³	4/3 M³

The number of times of multiplication and addition of complex numbers is given as Table 2 when the symbol detecting order is the same while detecting d(n) and d(n+1).

TABLE 2

	Multiplication of	Addition of
	complex numbers	complex numbers

V-BLAST + RLS channel tracking	⅚ M³+ 5.5 M²	⅚ M³+ 2 M²
Proposed MIMO-DFE	7.5 M²	4 M²

Table 3 shows operations of the multiplication and addition of complex numbers as number values according to the number of antennas, and shows corresponding decrease information.

TABLE 3

	Operations	M = N = 4	M = N = 8	M = N = 12


V-BLAST +	Multiplication	1139	6646	19656
RLS	of complex
channel	numbers
tracking	Addition of	704	4736	14976
	complex
	numbers
V-BLAST +	Multiplication	926	4043	9864
RLS	of complex
channel	numbers
tracking	Addition of	534	2475	6336
	complex
	numbers
Decrease	Multiplication	18.7%	39.2%	49.8%
rates	of complex
	numbers
	Addition of	24.2%	47.7%	57.5%
	complex
	numbers

FIGS. 5 and 7 show graphs of performance through simulations according to the MIMO adaptive receiving system and method.

A channel model by W. C. Jakes is used for the simulation, and it is assumed that no correlation is provided between the respective transmit and receive antennas, and the number of transmit and receive antennas is four.

A frame includes 160 transmit symbol vectors, 32 top transmit symbol vectors thereof are allocated to the training period to have the signals of the transmit antennas cross each other, and the signals modulated by the QPSK are transmitted to the receive axis (note that λ=0.95).

FIG. 5 shows a graph on the performance of algorithm 1 when the normalized Doppler frequency which is a product of the Doppler frequency fd and the symbol length T, the Doppler frequency being a barometer for indicating varied degrees of channels.

As shown, algorithm 1 with a simpler amount of calculation shows the same performance as that of the V-BLAST scheme.

When comparing the performance of algorithm 1 with the result of applying the V-BLAST scheme assuming that the channel of the lowest performance is completely known in FIG. 5, the performance is reduced by about 2 dB in the case of BER=10-3.

The method for inserting a mid-amble by the ratio of 1/4, and tracking the channel by blocks produces performance lower than that of the preferred embodiment.

FIG. 6 shows performance variations depending on fdT when the ratio of bit power and noise power, which indicating that the above-described trend is not varied when the degrees of channel variation is changed.

FIG. 7 shows performance of algorithm 2 according to the preferred embodiment of the present invention.

As shown, the simulation environment by algorithm 2 corresponds to that of FIG. 5, and the symbol detecting sequence follows variations of the update period γ. The performance is lowered by less than about 1 dB when γ is 12 or 24 in the environment that f_dT is 0.0005.

As described, the MIMO adaptive receiving system and method decides the optimal filter tap coefficients and the symbol detecting order in the time-varying channel environment to reduce complexity, provide easy realization, and produce similar performance compared to the conventional V-BLAST scheme, and further, it is applicable to other types of MIMO receivers.

While this invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

INVENTORS:

Lee, Yong-Hoon, Yu, Hee-jung, Choi, Ji-hoon

THIS PATENT IS REFERENCED BY THESE PATENTS:

Patent

Priority

Assignee

Title

THIS PATENT REFERENCES THESE PATENTS:

Patent	Priority	Assignee	Title
6097771,	Jul 01 1996	THE CHASE MANHATTAN BANK, AS COLLATERAL AGENT	Wireless communications system having a layered space-time architecture employing multi-element antennas
7072392,	Nov 13 2000	ENTROPIC COMMUNICATIONS, INC ; Entropic Communications, LLC	Equalizer for time domain signal processing
7113540,	Sep 18 2001	AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED	Fast computation of multi-input-multi-output decision feedback equalizer coefficients
KR19980013075,

ASSIGNMENT RECORDS Assignment records on the USPTO

Executed on	Assignor	Assignee	Conveyance	Frame	Reel	Doc
Mar 18 2010		Electronics and Telecommunications Research Institute	(assignment on the face of the patent)

MAINTENANCE FEES AND DATES: Maintenance records on the USPTO

Date	Maintenance Fee Events
Jul 28 2015	ASPN: Payor Number Assigned.
Sep 02 2015	M2552: Payment of Maintenance Fee, 8th Yr, Small Entity.
Nov 04 2019	REM: Maintenance Fee Reminder Mailed.
Apr 20 2020	EXP: Patent Expired for Failure to Pay Maintenance Fees.

Date	Maintenance Schedule
May 14 2016	4 years fee payment window open
Nov 14 2016	6 months grace period start (w surcharge)
May 14 2017	patent expiry (for year 4)
May 14 2019	2 years to revive unintentionally abandoned end. (for year 4)
May 14 2020	8 years fee payment window open
Nov 14 2020	6 months grace period start (w surcharge)
May 14 2021	patent expiry (for year 8)
May 14 2023	2 years to revive unintentionally abandoned end. (for year 8)
May 14 2024	12 years fee payment window open
Nov 14 2024	6 months grace period start (w surcharge)
May 14 2025	patent expiry (for year 12)
May 14 2027	2 years to revive unintentionally abandoned end. (for year 12)