CN104202280A - Hardware implementation for channel equalization of MIMO-OFDM (multiple-input multiple-output and orthogonal frequency division multiplexing) system - Google Patents

Abstract

The invention provides a hardware implementation method for channel equalization of an MIMO-OFDM (multiple-input multiple-output and orthogonal frequency division multiplexing) system. A Givens rotation-based QR decomposition method is provided and is applicable to a channel equalization module of the MIMO wireless system under the IEEE802.11ac protocol. A traditional matrix inversion hardware implementation method based on a block inversion idea involves mass second-order matrices and a calculation module of a complex multiplication and is only available to non-singular matrices. When the traditional QR decomposition method is used in matrix inversion, CORDIC (coordinated rotation digital computing) and SQUARE ROOT are involved, and many hardware resources are consumed. The invention provides a modified algorithm; matrix QR decomposition is achieved by a modified Givens rotation method to obtain a matrix inversion result. The hardware implementation method has the advantages that matrix inversion can be performed in a pipeline form, the use of coordinated rotation digital computing and square root is avoided, expenditure on the hardware resources is reduced, and a MIMO channel equalization process is implemented efficiently.

Description

The channel equalization hardware of MIMO-OFDM system is realized

Technical field

The present invention is applied to mimo wireless communication technical field, is specifically related to a kind of matrix inversion Hardware Implementation decomposing based on QR, for the channel equalization of MIMO receiver provides basis.

Background technology

MIMO (Multiple-InputMultiple-Output) system is used in the 4th third-generation mobile communication technical standard, has also been widely used in the standard agreement of high-speed radio local area network (LAN) (WLAN), such as IEEE802.11n/ac agreement etc. simultaneously.What accompanying drawing 1 was shown is the frame structure of IEEE802.11ac agreement, and wherein L-STF and L-LTF field are respectively used to receive frame synchronizing process and the symbol synchronization process of signal; After synchronous, signal data is transformed in frequency domain through FFT; The length that VHT-LTF field comprises n OFDM symbol (n representation space fluxion, n ≠ 3), for channel estimation process; At the receiving terminal of mimo system, channel estimating and channel equalization are very important modules.In the accuracy of the good and bad major embodiment channel matrix computing of MIMO receiver performance.Especially, OFDM wireless communication system for a 4 * 4MIMO, the channel equalization process of MIMO receiver has a lot of theoretical methods to realize, and MMSE equilibrium is exactly a kind of better and typical testing process, and this wherein needs to realize the process that quadravalence complex matrix is inverted.

FPGA (Filed-ProgrammableGateArray) is the most frequently used hardware development semi-custom circuit.Numerous associated auxiliary development products have also accelerated the step of the more new development of FPGA, and its Exploitation Scope is also more extensive.Utilize the PXI platform of NationalInstruments (NI) to carry out the FPGA exploitation of WLAN-OFDM system, broken the tradition that hardware program language carries out FPGA exploitation, the programming idea of the LabVIEW of NI based on graphic language makes hardware development more convenient, construction cycle shortens greatly, makes hardware development personnel more energy can be placed on to algorithm and realizes.This is also the advantage place of LabVIEW graphic language.

In mimo channel balancing procedure; hardware system is often difficult to reach the accuracy requirement of floating-point emulation; therefore some algorithms in past can be avoided matrix inversion process conventionally; this makes the design of channel equalization module become very complicated; not only the expression formula of equalizing signal is very loaded down with trivial details; and extremely consume hardware resource, the more important thing is that channel equalization performance is not high.Also can to channel matrix, carry out suitable minute block operations according to relevant matrix knowledge, for example quadravalence matrix can split into 4 second-order matrix, according to character such as matrix Applying Elementary Row Operations, the inversion process of quadravalence matrix is become to the inversion process of each piece self after piecemeal, this mode greatly reduces the design complexities of channel equalization module, and can effectively realize pile line operation mode.But the method can only realize the inversion process of invertible matrix, when running into singular matrix or approximate singular matrix, cannot obtain correct result.In practical communication environment, channel matrix can change along with the variation of environment, and system cannot guarantee the invertibity of channel matrix, and therefore, the method that high level matrix piecemeal is inverted is far from being enough for actual wireless communication system.Traditional QR decomposition method has been broken the partitioning of matrix requirement of method to matrix's reversibility of inverting, but for MIMO receiver, transmission rate is to weigh an important parameter of performance in wireless communication systems, the process that traditional Q R decomposition method realization matrix is inverted in this respect performance is not outstanding, thereby has limited the high-throughput transmission performance of mimo system.

Therefore, the channel equalization process need of MIMO receiver is considered the performances such as transmission rate of accuracy, integrality and the system of matrix inversion, also needs to consider hardware resource utilization simultaneously.

Summary of the invention

The technical problem running in the hardware of channel equalization is realized in order to overcome above-mentioned matrix inversion technique, for 4 * 4MIMO-OFDM system, the invention provides quadravalence complex matrix that a kind of complexity the is lower method of inverting, realize the channel equalization process of MIMO-OFDM system, comprise the following steps:

Step 1, obtain signal, and carry out synchronous;

Step 2, system is carried out to channel estimating, obtain channel estimate matrix A;

Step 3, channel estimate matrix A is carried out to Robin Givens rotation (GivensRotation) matrix decomposition;

Step 4, according to the inverse matrix of matrix A, carry out channel equalization.

Wherein, described step 1 is specially:

Step (1.1), transmitter encapsulate the baseband signal data that produced by the mode of accompanying drawing 1, by 4 transmit antennas, 4 road signals are sent, and signal sampling frequency is 130MHz; At receiver end, 4 signal process AD conversions that reception antenna receives, make continuous signal discretization, and down-sampled to 20MHz;

Step (1.2), the 4 road signals that receive are carried out synchronously, concrete steps are as follows:

The frame synchornization method of step (1.2a), employing delay-correlated obtains frame originating point information;

The symbol timing synchronization method of step (1.2b), employing difference delay-correlated obtains signal original position accurately, and by the original position of suitable delay acquisition VHT-LTF field;

Step (1.2c), 4 tunnels after synchronous are received to signals remove Cyclic Prefix, and use discrete Fourier transform (FFT) that signal is transformed in frequency domain.

Wherein, described step 2 is specially:

According to the VHT-LTF field that receives signal, calculate channel estimate matrix A, matrix A is quadravalence complex matrix, is defined as

$A=\left[\begin{array}{cccc}{a}_{11}& a_{12}& a_{13}& a_{14}\\ a_{21}& a_{22}& a_{23}& a_{24}\\ a_{31}& a_{32}& a_{33}& a_{34}\\ a_{41}& a_{42}& a_{43}& a_{44}\end{array}\right]\stackrel{\mathrm{\Δ}}{=}\left[\begin{array}{c}{r}_1\\ r_2\\ r_3\\ r_4\end{array}\right]$

Wherein, a _ij∈ C, row vector r _ithe i that represents matrix A is capable, r _ijrepresent r _ij element, i, j=1,2,3,4.

Wherein, described step 3 is specially:

By a series of Applying Elementary Row Operations, original matrix A is converted into the form of upper triangular matrix, this upper triangular matrix is designated as U, the transformation matrix T of equivalence and matrix A, U meet and are related to U=TA, wherein the effect factor of conversion is matrix T, if the factor obtained above is acted on quadravalence unit matrix E, E can be transformed into T, thereby therefore the effect factor of conversion be acted on to the upper result that can obtain transformation matrix T of quadravalence unit matrix E.

Wherein, described step 4 is specially:

Contrary the solving of step (4.1), matrix U:

$w_{\mathrm{ij}}=\left\{\begin{array}{cc}-\frac{1}{u_{\mathrm{jj}}}\underset{k=i}{\overset{j-1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ik}}{u}_{\mathrm{kj}}& i<j\\ \frac{1}{u_{\mathrm{jj}}}& i=j\\ 0& i>j\end{array}\right.---\left(3\right)$

According to formula (3), utilize the result U in step 3, thereby obtain the inverse matrix W of U;

Contrary the solving of step (4.2), matrix A:

The inverse matrix of original matrix A can be expressed as A ^-1=(T ^-1u) ^-1=U ^-1t.According to the T and the W that obtain in step 3 and step (4.1), the contrary product that can be expressed as W and T of matrix A, i.e. A ^-1=WT, matrix multiplication wherein, bottom computation structure adopts the complement of two's two's complement multiplier realization that contains 25 * 18;

Step (4.3), MIMO-OFDM system lower channel balancing procedure:

Suppose at t constantly, the reception vector that 4 tunnels receive the data field composition of signal is y=[y _1t; y _2t; y _3t; y _4t], according to the A obtaining in step (4.2) ^-1, channel equalization process can represent with following formula (4),

z＝A ^-1y

(4)

Wherein, z is the result after balancing procedure.

By matrix inversion technique provided by the invention, can realize inversion process with pipeline system, simultaneously can avoid using rotation of coordinate numerical calculation (CORDIC) and extracting operation (SQUAREROOT), reduce the expense of hardware resource and reduced the expense of quite a few hardware resource.

Accompanying drawing explanation

Fig. 1 is IEEE802.11ac agreement physical layer frame structure schematic diagram;

The QR decomposition process schematic diagram of Fig. 2 for revolving based on Robin Givens;

Fig. 3 is the constellation schematic diagram after 4 * 4MIMO-OFDM system channel equilibrium of preferred embodiment of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described.

According to preferred embodiment of the present invention, a kind ofly be applicable to the hard-wired QR based on Robin Givens rotation and decompose Hardware Implementation, the method can be applicable to the channel equalization module of MIMO wireless system under IEEE802.11ac agreement---and channel matrix is carried out to inversion process, thereby realize the channel equalization process of system receiver.Design comprises following steps:

Step 1, obtain signal, and carry out synchronous; Concrete steps are:

Step (1.1), transmitter become the frame format shown in accompanying drawing 1 by the baseband signal data encapsulation having produced, and by 4 transmit antennas, 4 road signals are sent, and signal sampling frequency is 130MHz; At receiver end, 4 signal process AD conversions that reception antenna receives, make continuous signal discretization, and down-sampled to 20MHz.

Step (1.2), the 4 road signals that receive are carried out synchronously, concrete steps are as follows:

The frame synchornization method of step (1.2a), employing delay-correlated obtains frame originating point information;

The symbol timing synchronization method of step (1.2b), employing difference delay-correlated obtains signal original position accurately, and by the original position of suitable delay acquisition VHT-LTF field.

Step (1.2c), 4 tunnels after synchronous are received to signals remove Cyclic Prefix, and use discrete Fourier transform (FFT) that signal is transformed in frequency domain.

Step 2, system is carried out to channel estimating, obtain channel estimate matrix A; Concrete steps are:

According to the reception calculated signals of VHT-LTF field, go out channel estimate matrix A, matrix A is quadravalence complex matrix, is defined as

$A=\left[\begin{array}{cccc}{a}_{11}& a_{12}& a_{13}& a_{14}\\ a_{21}& a_{22}& a_{23}& a_{24}\\ a_{31}& a_{32}& a_{33}& a_{34}\\ a_{41}& a_{42}& a_{43}& a_{44}\end{array}\right]\stackrel{\mathrm{\&Delta;}}{=}\left[\begin{array}{c}{r}_1\\ r_2\\ r_3\\ r_4\end{array}\right]$

Wherein, a _ij∈ C, row vector r _ithe i that represents matrix A is capable, r _ijrepresent r _ij element, i, j=1,2,3,4.

The GivensRotation matrix decomposition process of step 3, matrix A;

In QR decomposes, Robin Givens rotation (GivensRotation) method is that matrix is constantly carried out to Applying Elementary Row Operations in itself, progressively a common matrix is changed into the process of a upper triangular matrix.Because the element of channel matrix is normally plural, in traditional algorithm, this process relates to the operations such as plural delivery, extracting operation, and these operations can take more hardware resource, and this hardware designs to whole system is disadvantageous.Method provided by the invention can avoid using rotation of coordinate numerical calculation (CORDIC) and extracting operation (SQUAREROOT) in hardware implementation procedure, thereby reduces the expense of hardware resource.In the present invention, introduce zoom factor column vector k, k _ifor vectorial r _izoom factor, and be initialized as 1.

As shown in Figure 2, decomposable process comprises following concrete steps to concrete decomposition process:

Step (3.1), by a ₂₁, a ₃₁, a ₄₁correspondence position is transformed into 0 element, resolves into following concrete steps:

Step (3.1a), generation u ₁vector.

Step (3.1b), by a ₂₁the element of correspondence position is transformed into 0 element:

$\begin{array}{ccc}{\stackrel{\&OverBar;}{u}}_1=u_1+k_2{r}_{21}^{*}{r}_2& {\stackrel{\&OverBar;}{r}}_2=r_2-\left(\frac{r_{21}}{u_{11}}\right)u_1& {\stackrel{\&OverBar;}{k}}_2=k_2\left(\frac{u_{11}}{{\stackrel{\&OverBar;}{u}}_{11}}\right)\end{array}---\left(5\right)$

Wherein, vector with represent respectively vectorial u ₁and r ₂renewal, be still designated as u below ₁and r ₂.By such conversion, by r ₂₁be transformed into 0 element;

Step (3.1c), by a ₃₁and a ₄₁the element of correspondence position is transformed into 0 element.To u ₁and r ₃according to step (3.1b), convert, by r ₃₁be transformed into 0 element; To u ₁and r ₄according to same steps, convert, by r ₄₁be transformed into 0 element;

Step (3.2), by a ₃₂, a ₄₂evolution becomes 0 element, resolves into following concrete steps:

Step (3.2a), utilize the method for step (3.1a) to generate u ₂.

Step (3.2b), to u ₂and r ₃method according to formula (5) converts, by r ₃₂be transformed into 0; To u ₂and r ₄convert by the same way, by r ₄₂be transformed into 0;

Step (3.3), utilize identical method, the process of repeating step (3.2), can obtain u ₃and r ₄; According to the method for step (3.1a), utilize r ₄generate u ₄;

Step (3.4), step (3.1)～step (3.3) is drawn to the u of final updating _icombine and obtain matrix U, be expressed as following form:

U＝[u ₁；u ₂；u ₃；u ₄]

Matrix A is transformed into upper triangular matrix U by above-mentioned steps, and wherein the effect factor of conversion is matrix T.If it is upper that the factor obtained above is acted on to quadravalence unit matrix E, E can be transformed into T, therefore by above-mentioned by step (3.1)～step (3.3) thus result act on the upper result that can obtain transformation matrix T of quadravalence unit matrix E.

Step 4, according to the inverse matrix of matrix A, carry out channel equalization.Concrete steps are:

Contrary the solving of step (4.1), matrix U:

Known according to the character of upper triangular matrix, its inverse matrix W is upper triangular matrix equally.Element in matrix W meets following formula (6), i wherein, j ∈ { 1,2,3,4}, u _ijelement for the upper triangular matrix U that obtains in step 3.

$w_{\mathrm{ij}}=\left\{\begin{array}{cc}-\frac{1}{u_{\mathrm{jj}}}\underset{k=i}{\overset{j-1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ik}}{u}_{\mathrm{kj}}& i<j\\ \frac{1}{u_{\mathrm{jj}}}& i=j\\ 0& i>j\end{array}\right.---\left(6\right)$

According to formula (6), utilize the result U in step 3, thereby obtain the inverse matrix W of U.

Contrary the solving of step (4.2), matrix A:

The inverse matrix of original matrix A can be expressed as A ^-1=(T ^-1u) ^-1=U ^-1t.According to the T and the W that obtain in step 3 and step (4.1), the contrary product that can be expressed as W and T of matrix A, i.e. A ^-1=WT, matrix multiplication wherein, bottom computation structure adopts the complement of two's two's complement multiplier realization that contains 25 * 18;

Step (4.3), MIMO-OFDM system lower channel balancing procedure:

Suppose at t constantly, the reception vector that 4 tunnels receive the data field composition of signal is y=[y _1t; y _2t; y _3t; y _4t], according to the A obtaining in step (4.2) ^-1, channel equalization process can represent with following formula (7),

z＝A ^-1y

(7)

Wherein, z is the result after balancing procedure.Accompanying drawing 3 is for receiving the planisphere of signal.

Below in conjunction with actual hardware designs, further illustrate implementation method.In the present embodiment, the design of matrix inversion and channel equalization process be on Virtex-5 type FPGA, complete, carry out the control of radio-frequency antenna by NI5791R transceiver, the LabVIEW2012 of NI of take is hardware development instrument, and based on IEEE802.11ac agreement, system bandwidth is 20MHz.

Step 1, obtain signal, and carry out synchronous; Concrete steps are:

Step (1.1), transmitter become the frame format shown in accompanying drawing 1 by the baseband signal data encapsulation having produced, and by 4 transmit antennas, 4 road signals are sent, and signal sampling frequency is 130MHz; At receiver end, 4 signal process AD conversions that reception antenna receives, make continuous signal discretization, and down-sampled to 20MHz.

Step (1.2), the 4 road signals that receive are carried out synchronously, concrete steps are as follows:

The frame synchornization method of step (1.2a), employing delay-correlated obtains frame originating point information;

The symbol timing synchronization method of step (1.2b), employing difference delay-correlated obtains signal original position accurately, and by the original position of suitable delay acquisition VHT-LTF field.

Step (1.2c), 4 tunnels after synchronous are received to signals remove Cyclic Prefix, and use discrete Fourier transform (FFT) that signal is transformed in frequency domain, adopt the embedded IP core (FFT7.1) of Xilinx to realize.

Step 2, system is carried out to channel estimating, obtain channel estimate matrix A; Concrete steps are:

According to the reception calculated signals of VHT-LTF field, go out channel estimate matrix A, matrix A is quadravalence complex matrix, is defined as

$A=\left[\begin{array}{cccc}{a}_{11}& a_{12}& a_{13}& a_{14}\\ a_{21}& a_{22}& a_{23}& a_{24}\\ a_{31}& a_{32}& a_{33}& a_{34}\\ a_{41}& a_{42}& a_{43}& a_{44}\end{array}\right]\stackrel{\mathrm{\&Delta;}}{=}\left[\begin{array}{c}{r}_1\\ r_2\\ r_3\\ r_4\end{array}\right]$

Wherein, a _ij∈ C, row vector r _ithe i that represents matrix A is capable, r _ijrepresent r _ij element, i, j=1,2,3,4.

The GivensRotation matrix decomposition process of step 3, matrix A

In QR decomposes, Robin Givens rotation (GivensRotation) method is that matrix is constantly carried out to Applying Elementary Row Operations in itself, progressively a common matrix is changed into the process of a upper triangular matrix.Because the element of channel matrix is normally plural, in traditional algorithm, this process relates to the operations such as plural delivery, extracting operation, and these operations can take more hardware resource, and this hardware designs to whole system is disadvantageous.Method provided by the invention can avoid using rotation of coordinate numerical calculation (CORDIC) and extracting operation (SQUAREROOT) in hardware implementation procedure, thereby reduces the expense of hardware resource.In the present invention, introduce zoom factor column vector k, k _ifor vectorial r _izoom factor, and be initialized as 1.

As shown in Figure 2, decomposable process comprises the steps: concrete decomposition process

Step (3.1), by a ₂₁, a ₃₁, a ₄₁correspondence position is transformed into 0 element, resolves into following concrete steps:

Step (3.1a), generation u ₁vector.

Step (3.1b), by a ₂₁the element of correspondence position is transformed into 0 element:

$\begin{array}{ccc}{\stackrel{\&OverBar;}{u}}_1=u_1+k_2{r}_{21}^{*}{r}_2& {\stackrel{\&OverBar;}{r}}_2=r_2-\left(\frac{r_{21}}{u_{11}}\right)u_1& {\stackrel{\&OverBar;}{k}}_2=k_2\left(\frac{u_{11}}{{\stackrel{\&OverBar;}{u}}_{11}}\right)\end{array}---\left(8\right)$

Wherein, vector with represent respectively vectorial u ₁and r ₂renewal, be still designated as u below ₁and r ₂.By such conversion, by r ₂₁be transformed into 0 element;

Step (3.1c), by a ₃₁and a ₄₁the element of correspondence position is transformed into 0 element.To u ₁and r ₃according to step (3.1b), convert, by r ₃₁be transformed into 0 element; To u ₁and r ₄according to same steps, convert, by r ₄₁be transformed into 0 element;

Step (3.2), by a ₃₂, a ₄₂evolution becomes 0 element, resolves into following concrete steps:

Step (3.2a), utilize the method for step (3.1a) to generate u ₂.

Step (3.2b), to u ₂and r ₃method according to formula (8) converts, by r ₃₂be transformed into 0; To u ₂and r ₄convert by the same way, by r ₄₂be transformed into 0;

Step (3.3), utilize identical method, the process of repeating step (3.2), can obtain u ₃and r ₄; According to the method for step (3.1a), utilize r ₄generate u ₄;

Step (3.4) draws step (3.1)～step (3.3) u of final updating _icombine and obtain matrix U, be expressed as following form:

U＝[u ₁；u ₂；u ₃；u ₄]

Matrix A is transformed into upper triangular matrix U by above-mentioned steps, and wherein the effect factor of conversion is matrix T.If it is upper that the factor obtained above is acted on to quadravalence unit matrix E, E can be transformed into T, therefore by above-mentioned by step (3.1)～step (3.3) thus result act on the upper result that can obtain transformation matrix T of quadravalence unit matrix E.In actual applications, Robin Givens can be rotated to (GivensRotation) simplifies.First, renewal and the associative operation of zoom factor column vector k can save, and therefore, when the hardware inversion process of reality realization, the algorithm after simplification is compared the hardware resource that can save part with primal algorithm.

Step 4, according to the inverse matrix of matrix A, carry out channel equalization.Concrete steps are:

Contrary the solving of step (4.1), matrix U:

Known according to the character of upper triangular matrix, its inverse matrix W is upper triangular matrix equally.Element in matrix W meets following formula (9), i wherein, j ∈ { 1,2,3,4}, u _ijelement for the upper triangular matrix U that obtains in step 3.

$w_{\mathrm{ij}}=\left\{\begin{array}{cc}-\frac{1}{u_{\mathrm{jj}}}\underset{k=i}{\overset{j-1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ik}}{u}_{\mathrm{kj}}& i<j\\ \frac{1}{u_{\mathrm{jj}}}& i=j\\ 0& i>j\end{array}\right.---\left(9\right)$

According to formula (9), utilize the result U in step 3, thereby obtain the inverse matrix W of U.

Contrary the solving of step (4.2), matrix A:

The inverse matrix of original matrix A can be expressed as A ^-1=(T ^-1u) ^-1=U ^-1t.According to the T and the W that obtain in step 3 and step (4.1), the contrary product that can be expressed as W and T of matrix A, i.e. A ^-1=WT, matrix multiplication wherein, bottom computation structure adopts the complement of two's two's complement multiplier realization that contains 25 * 18;

Step (4.3), MIMO-OFDM system lower channel balancing procedure:

Suppose at t constantly, the reception vector that 4 tunnels receive the data field composition of signal is y=[y _1t; y _2t; y _3t; y _4t], according to the A obtaining in step (4.2) ^-1, channel equalization process can represent with following formula (10),

z＝A ^-1y

(10)

Wherein, z is the result after balancing procedure.Accompanying drawing 3 is for receiving the planisphere of signal.

Although illustrated and described embodiments of the invention, those having ordinary skill in the art will appreciate that: in the situation that not departing from principle of the present invention and aim, can carry out multiple variation, modification, replacement and modification to these embodiment, scope of the present invention is limited by claim and equivalent thereof.

Claims (5)

1. a channel equalization method for MIMO-OFDM system, take 4 * 4MIMO-OFDM system as example at this.It comprises the following steps:

Step 1, obtain signal, and carry out synchronous;

Step 2, system is carried out to channel estimating, obtain channel estimate matrix A;

Step 3, channel estimate matrix A is carried out to Robin Givens rotation (GivensRotation) matrix decomposition;

Step 4, according to the inverse matrix of matrix A, carry out channel equalization.

2. the channel equalization method of mimo system as claimed in claim 1, wherein said step 1 is specially:

Step (1.1), transmitter encapsulate the baseband signal data that produced, and by 4 transmit antennas, 4 road signals are sent, and signal sampling frequency is 130MHz; At receiver end, 4 signal process AD conversions that reception antenna receives, make continuous signal discretization, and down-sampled to 20MHz;

Step (1.2), the 4 road signals that receive are carried out synchronously, concrete steps are as follows:

The frame synchornization method of step (1.2a), employing delay-correlated obtains frame originating point information;

The symbol timing synchronization method of step (1.2b), employing difference delay-correlated obtains signal original position accurately, and by the original position of suitable delay acquisition VHT-LTF field;

Step (1.2c), 4 tunnels after synchronous are received to signals remove Cyclic Prefix, and use discrete Fourier transform (FFT) that signal is transformed in frequency domain.

3. the channel equalization method of mimo system as claimed in claim 1, wherein said step 2 is specially:

According to the VHT-LTF field that receives signal, calculate channel estimate matrix A, matrix A is quadravalence complex matrix, is defined as

$A=\left[\begin{array}{cccc}{a}_{11}& a_{12}& a_{13}& a_{14}\\ a_{21}& a_{22}& a_{23}& a_{24}\\ a_{31}& a_{32}& a_{33}& a_{34}\\ a_{41}& a_{42}& a_{43}& a_{44}\end{array}\right]\stackrel{\mathrm{\&Delta;}}{=}\left[\begin{array}{c}{r}_1\\ r_2\\ r_3\\ r_4\end{array}\right]$

Wherein, a _ij∈ C, row vector r _ithe i that represents matrix A is capable, r _ijrepresent r _ij element, i, j=1,2,3,4.

4. the channel equalization method of mimo system as claimed in claim 1, wherein said step 3 is specially:

By a series of Applying Elementary Row Operations, original matrix A is converted into the form of upper triangular matrix, this upper triangular matrix is designated as U, the transformation matrix T of equivalence and matrix A, U meet and are related to U=TA, wherein the effect factor of conversion is matrix T, if the factor obtained above is acted on quadravalence unit matrix E, E can be transformed into T, thereby therefore the effect factor of conversion be acted on to the upper result that can obtain transformation matrix T of quadravalence unit matrix E.

5. the channel equalization method of mimo system as claimed in claim 1, wherein said step 4 is specially:

Contrary the solving of step (4.1), matrix U:

$w_{\mathrm{ij}}=\left\{\begin{array}{cc}-\frac{1}{u_{\mathrm{jj}}}\underset{k=i}{\overset{j-1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ik}}{u}_{\mathrm{kj}}& i<j\\ \frac{1}{u_{\mathrm{jj}}}& i=j\\ 0& i>j\end{array}\right.---\left(1\right)$

According to formula (1), utilize the result U in step 3, thereby obtain the inverse matrix W of U;

Contrary the solving of step (4.2), matrix A:

The inverse matrix of original matrix A can be expressed as A ^-1=(T ^-1u) ^-1=U ^-1t.According to the T and the W that obtain in step 3 and step (4.1), the contrary product that can be expressed as W and T of matrix A, i.e. A ^-1=WT, matrix multiplication wherein, bottom computation structure adopts the complement of two's two's complement multiplier realization that contains 25 * 18;

Step (4.3), MIMO-OFDM system lower channel balancing procedure:

Suppose at t constantly, the reception vector that 4 tunnels receive the data field composition of signal is y=[y _1t; y _2t; y _3t; y _4t], according to the A obtaining in step (4.2) ^-1, channel equalization process can represent with following formula (2),

z＝A ^-1y (2)

Wherein, z is the result after balancing procedure.