CN1703034A - A MIMO-OFDM system based channel estimation method - Google Patents

Abstract

This invention provides one simple and effective training sequence design scheme in the MIMO-OFDM system with several antenna, which comprises front frequency structure and comb frequency structure to make each antenna frequency position crossed to remove the interference between each antenna, wherein, the front frequency structure scheme is suitable for slow declining channel and the comb frequency finding structure scheme is suitable form rapid declining channel. The invention also provides one MIMO-OFDM channel computation method based on the ML method to apply the method into the above two structures.

Description

A kind of channel estimation methods based on the MIMO-OFDM system

Technical field:

The present invention relates to OFDM (Orthogonal Frequency DivisionMultiplexing, OFDM) communication technology, particularly a kind of based on multiple-input and multiple-output (Multiple-Input Multiple-Output, MIMO) channel estimation methods of the system that combines with the OFDM technology of channel.

Background technology:

Along with the fast development of Digital Signal Processing and integrated circuit technique, (OrthogonalFrequency Division Multiplexing, OFDM) technology is just becoming one of research focus of communication to OFDM.OFDM has very high message transmission rate, and high bandwidth efficiency, and the premium properties of antagonism multipath fading have obtained application more and more widely in wireless multimedia communication, be (B3G, 4G) core technology that will adopt of system behind the 3G.And, be applied to multiple-input and multiple-output (Multiple-Input Multiple-Output, MIMO) Space Time Coding in the channel is a kind of technology of having very much potentiality, the Space Time Coding system has been included among the standard (IMT-2000 standard) of 3G.On the other hand, OFDM is the good application platform of Space Time Coding on frequency selective fading channels, and Space Time Coding and the OFDM technology that is applied in the mimo channel combined the comprehensively advantage of the two, and diversity gain is provided on frequency-selective channel.But in broadband wireless communications wireless channel be frequency selectivity with the time become, the dynamic estimation of carrying out channel before therefore signal carries out demodulation in to OFDM and MIMO-OFDM system is very important.

Good channel estimation method is one of key that realizes the OFDM premium properties.In ofdm system, for the channel estimating based on pilot tone, can be divided into preceding pilot channel estimation (promptly towards decision method) and comb type pilot channel estimation from the pilot tone insertion position (be the pilot assisted modulation method, PSAM).The former is suitable for slow fading channel, and the latter is suitable for fast-changing channel.In channel estimating,, can adopt least mean-square error (MMSE), least square (LS), maximal possibility estimation (MLE) etc. for the estimation of the channel condition information at pilot sub-carrier place based on comb type pilot tone.Though MMSE method performance is fine, more complicated, amount of calculation is bigger, and the LS method is the simplest, but performance is generally undesirable.

Equally, the dynamic estimation that signal carries out carrying out before the demodulation channel in to the MIMO-OFDM system also is very important.In ofdm system, can easily realize channel estimating by on subcarrier, inserting pilot tone, yet in the multiple transmit antennas ofdm system, received signal on each subcarrier comes from the superposition of the independent fading signal of a plurality of transmitting antennas, and therefore the OFDM channel estimation methods at single antenna can not identify a plurality of channel fading coefficients.

At the problems referred to above, a kind of simple and effective training sequence design need to be proposed, the phase mutual interference when removing channel estimating between antenna, and propose a kind of simply based on the MIMO-OFDM channel estimation method of maximum likelihood (ML) method.

Summary of the invention:

Technical problem to be solved by this invention is in having a plurality of MIMO-OFDM systems that transmit and receive antenna, a kind of simple and effective training sequence design is provided, pilot configuration and comb type pilot configuration before comprising, make the pilot frequency locations on each antenna stagger mutually, phase mutual interference during with the removal channel estimating between each antenna pilot, wherein preceding pilot tone organization plan is applicable to slow fading channel, and comb type pilot configuration scheme is applicable to fast fading channel.A kind of new MIMO-OFDM channel estimation method based on maximum likelihood (ML) method is then proposed, and with this algorithm application in these two kinds of pilot configurations.

For solving the problems of the technologies described above, the present invention is achieved in that

The invention provides a kind of channel estimation methods, comprise the steps: based on the MIMO-OFDM system

(1) training sequence design, phase mutual interference when making it to avoid channel estimating between each antenna pilot, described training sequence design is that the pilot frequency locations on each transmitting antenna is staggered mutually, if certain subcarrier is as the pilot tone of certain transmitting antenna, then other antenna sends zero-signal in this sub-carrier positions, thereby avoids the phase mutual interference between each antenna pilot when channel estimating;

(2) adopt a kind of channel estimation method based on maximum likelihood, this likelihood function of algorithm definition and distance function estimate channel response and make the distance function minimum, thereby make the likelihood function maximum.

Has n for one _TThe MIMO-OFDM system of individual transmitting antenna, the total number of sub-carriers that is used to insert frequency pilot sign is N _p〉=n _TL, L represent the channel multi-path number, and frequency pilot sign should evenly expand to whole subcarrier zone, thereby evaluated error is distributed on the whole system bandwidth.

If the OFDM symbol contains N _sIndividual subcarrier, if pilot configuration before adopting, with certain cycle with all subcarriers of an OFDM symbol pilot tone as each antenna, be applicable to slow fading channel, with the whole subcarrier K={0 of described OFDM symbol, 1 ..., N _s-1} is divided into n _TGroup: K _i={ mn _T+ i-1, m=0,1 ..., N _p/ n _T-1}, i=1,2 ... n _T, N here _p=N _s, i antenna is at k _i∈ K _iSubcarrier place pilot signal transmitted, and send zero-signal at other subcarrier place.

If adopt comb type pilot configuration, all will insert pilot tone on each OFDM symbol, each OFDM symbol comprises frequency pilot sign and data symbol, and the mode that this pilot tone is inserted has been followed the tracks of the variation of channel preferably, be applicable to fast-changing channel, with the N of OFDM symbol _pIndividual pilot sub-carrier is divided into n _TGroup: $K_i=\{{\mathrm{mn}}_T\frac{N_s}{N_p}+i-1,m=\mathrm{0,1},\·\·\·,N_p/n_T-1\},,i=\mathrm{1,2},\·\·\·,n_T,$ For i antenna, at k _i∈ K _iSubcarrier place pilot signal transmitted, at k _i∈ K _j, j=1,2 ..., n _TAnd the subcarrier place of j ≠ i sends zero-signal, and sends data-signal at other subcarrier place.

At the given channel impulse response h of hypothesis from i transmitting antenna to reception antenna _iThe time, the definition likelihood function:

$\stackrel{^\·}{f}\left(R\right|h_i)=\frac{1}{{\left(2\mathrm{\π}{\mathrm{\σ}}_N^2\right)}^{N_s}}\exp \{-\frac{D\left(h_i\right)}{2{\mathrm{\σ}}_N^2}\}$

Wherein R is a received signal vector, σ _N ²Be the variance of channel additive white Gaussian noise, function D (h _i) be called distance function, be defined as:

$D\left(h_i\right)=\underset{j=0}{\overset{N_p{n}_T}{\mathrm{\Σ}}}|R\left(k_i^j\right)-X_i\left(k_i^j\right)\underset{i=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\α}}_{i,l}\exp {(-j2\mathrm{\π}\frac{k_i^{j}l}{N_s})|}^2$

X wherein _i(k _i ^j), R (k _i ^j) be respectively i transmitting antenna k _i ^jTransmission signal and received signal on the individual subcarrier, k _i ^jBe j pilot sub-carrier on each and every one transmitting antenna of i, L represents multipath number, α _{I, l}Be the multiple fading coefficients of l paths, work as X from antenna i to reception antenna _iWhen known, obtain channel impulse response h _i, so that likelihood function f is (R|h _i) maximum, just make distance function D (h _i) minimum.

Has n at one _TIn the MIMO-OFDM system of an individual transmitting antenna and a reception antenna, the number of sub carrier wave of OFDM is N _s, each antenna be used to insert frequency pilot sign subcarrier add up to N _p, N here _p=N _s, the channel multi-path number is L, the implementation of this algorithm may further comprise the steps in this system:

Pilot configuration before 1. if a. adopts, with whole subcarrier K={0,1 ..., N _s-1} is divided into n _TGroup: K _i={ mn _T+ i-1, m=0,1 ..., N _p/ n _T-1}, i=1,2 ... n _T, i antenna is at k _i∈ K _iSubcarrier place pilot signal transmitted, and send zero-signal at other subcarrier place;

If b. adopt comb type pilot configuration, with N _pIndividual pilot sub-carrier is divided into n _TGroup:

$K_i=\{{\mathrm{mn}}_T\frac{N_s}{N_p}+i-1,m=\mathrm{0,1},\·\·\·,N_p/n_T-1\},,i=\mathrm{1,2},\·\·\·,n_T,$ For i antenna, at k _i∈ K _iSubcarrier place pilot signal transmitted, at k _i∈ K _j, j=1,2 ..., n _TAnd the subcarrier place of j ≠ i sends zero-signal, and sends data-signal at other subcarrier place;

2. for i antenna, by formula

S _i(k _i ^j)=|X _i(k _i ^j)| ²，Z _i(k _i ^j)=X _i ^*(k _i ^j)R(k _i ^j)j＝0，1，…，N _p/n _T-1

Try to achieve { S _i(k _i ^j) g and { Z _i(k _i ^j), k _i ^j∈ K _i, j=0,1 ..., N _p/ n _T-1;

3. ask sequence { S respectively _i(k _i ^j) and { Z _i(k _i ^j) N _p/ n _TPoint IFFT obtains { s _iAnd { z (p) } _i(p) }, p=0,1 ..., N _p/ n _T-1;

4. ask sequence { s respectively _iAnd { z (p) } _i(p) } L point FFT obtains sequence { S _i ^(L)(k _i ^j) and { Z _i ^(L)(k _i ^j);

5. ask sequential element Z one by one _i ^(L)(k _i ^j) and S _i ^(L)(k _i ^j) the merchant, i.e. Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j);

6. ask sequence { Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j) L point IFFT, obtain sequence

L=0,1 ..., L-1;

7. will Adopt N behind the zero padding _sPoint FFT transforms to frequency domain, the frequency response coefficient { H on just can obtaining from i transmitting antenna to each subcarrier of reception antenna _i(k) } estimated value

$\left\{{\hat{H}}_i\left(k\right)\right\},k=\mathrm{0,1},\·\·\·,N_s-1.$

The advantage that the present invention has is:

(1) adopts a kind of simple and effective training sequence design, pilot configuration and comb type pilot configuration before comprising, make the pilot tone arrangement position on each antenna stagger mutually, the phase mutual interference during with the removal channel estimating between each antenna pilot, the accuracy of raising channel estimating.The present invention can adapt to the characteristics of MIMO-OFDM system multiple transmit antennas.

(2) channel estimation method that adopts this training sequence design to be proposed can be applied in the MIMO-OFDM system.

(3) it is less that this channel estimation method is compared general MMSE method amount of calculation, and channel estimating performance is better than general LS algorithm.

(4) in slow fading channel, adopt designed preceding pilot tone organization plan, this channel estimation method can be obtained effect preferably.

(5) in fast-changing channel, adopt designed comb type pilot configuration scheme to upgrade channel condition information in time along with channel variation, this channel estimation method can be obtained effect preferably.

Description of drawings:

Fig. 1 is a MIMO-OFDM system configuration schematic diagram of the present invention.

Fig. 2 is that training sequence of the present invention is the schematic diagram of preceding pilot configuration.

Fig. 3 is that training sequence of the present invention is the schematic diagram of comb type pilot configuration.

Fig. 4 is to the amended structural representation of comb type pilot configuration shown in Figure 3.

Fig. 5 is a maximum likelihood channel estimation method schematic diagram of the present invention.

Fig. 6 is the channel estimating performance schematic diagram that the present invention adopts preceding pilot configuration and comb type pilot configuration training sequence.

Embodiment:

The present invention is in having a plurality of MIMO-OFDM systems that transmit and receive antenna, adopt a kind of simple and effective training sequence design, pilot configuration and comb type pilot configuration before comprising, make the pilot frequency locations on each antenna stagger mutually, phase mutual interference during with the removal channel estimating between each antenna pilot, wherein preceding pilot tone organization plan is applicable to slow fading channel, and comb type pilot configuration scheme is applicable to fast fading channel.A kind of new MIMO-OFDM channel estimation method based on maximum likelihood (ML) method is then proposed, and with this algorithm application in these two kinds of pilot configurations.

Concrete scheme is as follows:

(1) system model

MIMO-OFDM system with nT transmitting antenna and a reception antenna as shown in Figure 1.Suppose that intercarrier is synchronous, does not consider Doppler frequency shift.In n OFDM mark space, data block b (k, n), k=0,1 ..., N _sThe space-time encoded one-tenth nT of-1} (wherein Ns is the number of sub carrier wave of OFDM) different symbolic blocks { X _i(k, n), k=0,1 ..., N _s-1}, i=1 ..., n _TEach symbolic blocks is modulated via OFDM, promptly carry out quick inverse-Fourier transform (IFFT) after, at the N of corresponding antenna _sLaunch on the individual subcarrier.That is to say, between each transmitting antenna and reception antenna, exist the communication link of setting up by OFDM.Therefore, in demodulation, promptly carry out fast Fourier transform (FFT) after, the received signal on each subcarrier is the superposition that nT distortion sends signal, then can be expressed as:

$R(k,n)=\underset{i=1}{\overset{n_T}{\mathrm{\Σ}}}{X}_i(k,n)H_i\left(k,n\right)+W(k,n)$ k＝0，1，…，N _s-1 (1)

X wherein _i(k, n), (k n) is transmission signal and received signal on k subcarrier of n OFDM symbol respectively to R; H _i(k n) is from i transmitting antenna to reception antenna the channel frequency response coefficient on k subcarrier during n the OFDM symbol.(k n) is additive white Gaussian noise on k subcarrier during n the OFDM symbol to W, and its average is 0, and variance is σ _N ²In order to obtain to send diversity gain and detection of transmitted signals, must estimate the state information H of channel _i(k, n), i=1 ..., n _TAnd channel adopts the tapped delay line model, and its corresponding discrete channel impulse response can be expressed as:

$h_i(n,\mathrm{\τ})=\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\α}}_{i,l}\mathrm{\δ}(\mathrm{\τ}-{\mathrm{\τ}}_{i,l})$ i＝1，…，n _T (2)

Wherein L represents multipath number, α _{I, l}Being the multiple fading coefficients of the l paths from antenna i to reception antenna, is that average is 0, and variance is σ _l ²Multiple Gaussian process.Channel response is carried out normalization, then have $\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\σ}}_l^2=1.$ Channel is the Rayleigh multidiameter fading channel, and hypothesis is constant in an OFDM mark space Ts, and changes to another OFDM symbol from an OFDM symbol.H _i(k n) can be expressed as:

$H_i(k,n)=\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\α}}_{i,l}\exp (-j2\mathrm{\π}\frac{\mathrm{kl}}{N_s})$ k＝0，1，…，N _s-1；i＝1，…，n _T (3)

Because channel carries out normalization, then have E (| H _i(k, n) | ²)=1.Has equal ENERGY E if send signal _s, the variance of channel additive white Gaussian noise is σ _N ², then average received symbol signal to noise ratio is

$\mathrm{SNR}=n_T{E}_s/{\mathrm{\σ}}_N^2.$

(2) design of training sequence

The present invention adopts the channel estimating based on pilot tone.The general training sequence can be by constructed in two ways: preceding pilot configuration and comb type pilot configuration.The former inserts pilot tone with certain cycle on all subcarriers of OFDM symbol, be applicable to slow fading channel; The latter will insert pilot tone on each OFDM symbol, the OFDM symbol comprises frequency pilot sign and data symbol, and the mode that this pilot tone is inserted has been followed the tracks of the variation of channel preferably.Be applicable to fast-changing channel.

If on a certain subcarrier k, have only i antenna transmission signal, then formula (1) just can be reduced to:

R(k，n)＝X _i(k，n)H _i(k，n)+W(k，n) (4)

Can construct training sequence in view of the above.To transmitting antenna i, pilot symbol transmitted on subcarrier k, then other all transmitting antennas send zero-signal on subcarrier k.To next subcarrier k+1, antenna i+1 can send a frequency pilot sign, and other all transmitting antennas send zero-signal.We can further be divided into pilot sub-carrier the nT group, and transmitting antenna i sends zero-signal only at the pilot frequency locations place pilot signal transmitted that belongs to the i group at other pilot frequency locations place.Receiving terminal just can estimate { H from transmitting antenna i so _i(k, n), k=0,1 ..., N _s-1, i=1 ..., n _TIn the frequency response coefficient of i group subcarrier, and { H _i(k, n), k=0,1 ..., N _sThe frequency response coefficient of other position can be obtained by interpolating method among the-1}.

For the selection and the insertion of pilot tone, adopt the described optimal spatial pilot tone of document [1].Has n for one _TThe MIMO-OFDM system of individual transmitting antenna, the total number of sub-carriers that is used to insert frequency pilot sign is N _p〉=n _TL.And frequency pilot sign should evenly expand to whole subcarrier zone, so that evaluated error is distributed on the whole system bandwidth.

If pilot configuration before adopting, as shown in Figure 2, with whole subcarrier K={0,1 ..., N _s-1} is divided into n _TGroup: K _i={ mn _T+ i-1, m=0,1 ..., N _p/ n _T-1}, i=1,2 ..., n _T, N here _p=N _sI antenna is at k _i∈ K _iSubcarrier place pilot signal transmitted, and send zero-signal at other subcarrier place.

If adopt comb type pilot configuration, as shown in Figure 3.With N _pIndividual pilot sub-carrier is divided into n _TGroup:

$K_i=\{{\mathrm{mn}}_T\frac{N_s}{N_p}+i-1,m=\mathrm{0,1},\·\·\·,N_p/n_T-1\},,\mathrm{1,2},\·\·\·,n_T.$ For i antenna, at k _i∈ K _iSubcarrier place pilot signal transmitted, at k _i∈ K _j, j=1,2 ..., n _TAnd the subcarrier place of j ≠ i sends zero-signal, and sends data-signal at other subcarrier place.

In order to improve efficiency of transmission, for slow fading channel, can the comb type pilot configuration of Fig. 3 be improved, to improve the utilance of frequency spectrum, as shown in Figure 4.

(3) maximum likelihood channel estimation method

To n _TThe MIMO-OFDM system of individual transmitting antenna is with N _pIndividual pilot sub-carrier is divided into n _TGroup K _i, i=1,2 ..., n _TAfter, concerning antenna i, make k _i ^j∈ K _i, j=0,1 ..., N _p/ n _T-1, then have:

$R\left(k_i^j\right)=X_i\left(k_i^j\right)H_i\left(k_i^j\right)+W\left(k_i^j\right)$

$=X_i\left(k_i^j\right)\left(\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\α}}_{i,l}\exp (-j2\mathrm{\π}\frac{k_i^{j}l}{N_s})\right)+W\left(k_i^j\right)$

i＝1，2，…，n _T；j＝0，1，…，N _p/n _T-1 (5)

Here remove OFDM notation index n for simplicity.

At given channel impulse response h _iThe time, the definition likelihood function:

$f\left(R\right|h_i)=\frac{1}{{\left(2\mathrm{\π}{\mathrm{\σ}}_N^2\right)}^{N_s}}\exp \{-\frac{D\left(h_i\right)}{2{\mathrm{\σ}}_N^2}\}$ (6)

Function D (h wherein _i) be called distance function, be defined as:

$D\left(h_i\right)=\underset{j=0}{\overset{N_p{n}_T}{\mathrm{\Σ}}}|R\left(k_i^j\right)-X_i\left(k_i^j\right)\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\α}}_{i,j}\exp (-j2\mathrm{\π}\frac{k_i^{j}l}{N_s}){|}^2$ (7)

Work as X _iWhen known, demand goes out channel impulse response h _i, so that likelihood function f is (R|h _i) maximum, just make distance function D (h _i) minimum.

Make α _{I, l}=a _Il+ jb _Il, l=0,1 ..., L-1 is for obtaining α _{I, l}, with D (h _l) its real part and imaginary part are asked partial derivative, have:

$\frac{\∂D\left(h_i\right)}{\∂a_{\mathrm{il}}}{|}_{h_i{\hat{h}}_i}=0,\frac{\∂D\left(h_i\right)}{{\∂b}_{\mathrm{il}}}{|}_{h_i={\hat{h}}_i}=0$ (8)

Wherein Be h _iEstimated value.Can get thus:

$\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\hat{a}}_{\mathrm{il}}\mathrm{Re}\left(s_i(p-l)\right)-\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\hat{b}}_{\mathrm{ij}}\mathrm{Im}\left(s_i(p-l)\right)\mathrm{Re}\left(z_i\left(p\right)\right)$ (9)

$\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\hat{a}}_{\mathrm{il}}\mathrm{Im}\left(s_i(p-l)\right)+\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\hat{b}}_{\mathrm{ij}}\mathrm{Re}\left(s_i(p-l)\right)=\mathrm{Im}\left(z_i\left(p\right)\right)$ (10)

Wherein, p=0,1 ..., L-1, _Il,  _IlBe respectively a _Il, b _IlEstimated value; Re (), Im () are respectively and get real part and get imaginary part; s _i(p) and z _i(p) be respectively S _i(k _i ^j) and Z _i(k _i ^j) the IFFT conversion:

$S_i\left(k_i^j\right)={\left|X_i\left(k_i^j\right)\right|}^2,z_i\left(k_i^j\right)=X_i^{*}\left(k_i^j\right)R\left(k_i^j\right)$ (11)

j＝0，1，…，N _p/n _T-1

By formula (9) and (10), have of equal valuely:

$\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\widehat{\mathrm{\α}}}_{i,l}{s}_i(p-l)=z_i\left(p\right)$ p＝0，1，…，L-1，i＝1，2，…，n _T (12)

L point FFT is got on formula (12) both sides, then has:

${\hat{H}}_i^{\left(L\right)}\left(k_i^j\right)S_i^{\left(L\right)}\left(k_i^j\right)=Z_i^{\left(L\right)}\left(k_i^j\right)$ j＝0，1，…，L-1，i＝1，2，…，n _T (13)

Wherein subscript (L) is represented counting that FFT (IFFT) adopted, to be different from previous FFT and IFFT.

Be H _i(k _i ^j) estimated value, H _i(k _i ^j) be from i transmitting antenna to reception antenna at k _i ^jFrequency response coefficient on the individual subcarrier.Can get thus:

${\widehat{\mathrm{\α}}}_{i,l}^{\left(L\right)}=\mathrm{IFFT}\left\{\frac{Z_i^{\left(L\right)}\left(k_i^j\right)}{S_i^{\left(L\right)}\left(k_i^j\right)}\right\}$ l＝0，1，…，L-1 (14)

Will

Adopt N behind the zero padding _sPoint FFT transforms to frequency domain, the frequency response coefficient H on just can obtaining from i transmitting antenna to each subcarrier of reception antenna _i(k) estimated value

K=0,1 ..., N _s-1.

In sum, in a MIMO-OFDM system with nT transmitting antenna and a reception antenna, the number of sub carrier wave of OFDM is N _s, each antenna be used to insert frequency pilot sign subcarrier add up to N _p, N here _p=N _s, adopting the optimal spatial pilot configuration, the channel multi-path number is L.The concrete implementation of this maximum likelihood channel estimation method is as follows in this system:

Pilot configuration before 1. if a. adopts, as shown in Figure 2, with whole subcarrier K={0,1 ..., N _s-1) is divided into n _TGroup: K _i={ mn _T+ i-1, m=0,1 ..., N _p/ n _T-1}, i=1,2 ... n _TI antenna is at k _i∈ K _iSubcarrier place pilot signal transmitted, and send zero-signal at other subcarrier place.

If b. adopt comb type pilot configuration, as shown in Figure 3.With N _pIndividual pilot sub-carrier is divided into n _TGroup:

$K_i=\{{\mathrm{mn}}_T\frac{N_s}{N_p}+i-1,m=\mathrm{0,1},\·\·\·,N_p/n_T-1\},,i=\mathrm{1,2},\·\·\·,n_T.$ For i antenna, at k _i∈ K _iSubcarrier place pilot signal transmitted, at k _i∈ K _j, j=1,2 ..., n _TAnd the subcarrier place of j ≠ i sends zero-signal, and sends data-signal at other subcarrier place.

2. for i antenna, try to achieve { S by formula (11) _i(k _i ^j) and { Z _i(k _i ^j), k _i ^j∈ K _i, j=0,1 ..., N _p/ n _T-1.

3. ask sequence { S respectively _i(k _i ^j) and { Z _i(k _i ^j) N _p/ n _TPoint IFFT obtains { s _iAnd { z (p) } _i(p) }, p=0,1 ..., N _p/ n _T-1.

4. ask sequence { s respectively _iAnd { z (p) } _i(p) } L point FFT obtains sequence { S _i(L) (k _i ^j) and { Z _i ^(L)(k _i ^j).

5. ask sequential element Z one by one _i ^(L)(k _i ^j) and S _i ^(L)(k _i ^j) the merchant, i.e. Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j).

6. ask sequence { Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j) L point IFFT, obtain sequence

L=0,1 ..., L-1.

7. will Adopt N behind the zero padding _sPoint FFT transforms to frequency domain, the frequency response coefficient { H on just can obtaining from i transmitting antenna to each subcarrier of reception antenna _i(k) } estimated value

, k=0,1 ..., N _s-1.

This algorithm flow as shown in Figure 5.

Performance for estimating channel can be weighed by its mean square error (MSE):

$\mathrm{MSE}=\underset{i=1}{\overset{n_T}{\mathrm{\Σ}}}E\left\{{|{\hat{H}}_i\left(k\right)-H_i\left(k\right)|}^2\right\}/n_T$ (15)

(4) simulation result

To have two transmitting antennas and a reception antenna MIMO-OFDM system is an example, adopt the QPSK modulation, bandwidth is 10MHz, is divided into 1024 subchannels, with normalized protection of sample time interval N _g=226.Channel has 5 multipaths, and maximum delay spread is L=136.If pilot configuration before adopting as shown in Figure 2, is the cycle with 5 OFDM symbols, insert a preceding frequency pilot sign, remaining OFDM symbol is used to transmit data, and channel status supposes it is constant in a frame (5 OFDM mark spaces), and changes to another frame from a frame.If adopt comb type pilot configuration, as shown in Figure 3, the pilot sub-carrier number of two antennas is N _p=512, be divided into n _T=2 groupings.

The channel estimating performance that adopts these two kinds of training sequences as shown in Figure 6.For relatively, adopt same comb type pilot configuration also to be applied to this system in conjunction with the channel estimation method of time domain interpolation the LS method in the document [2], the result of gained also is shown among this figure.As seen from Figure 6, with this employing equally the former algorithm of comb type pilot configuration compare, channel estimating performance of the present invention will be got well, this method has suppressed The noise effectively.And adopt leading frequency ratio to adopt the channel estimating performance of comb type pilot tone to get well.But, for fast-changing channel,, but, can bring bigger error up to the arrival of next frequency pilot sign because pilot tone can not be upgraded channel condition information in time along with channel variation before adopting, it generally only is applicable to slow fading channel.

Being preferred embodiment of the present invention only in sum, is not to be used for limiting practical range of the present invention.Be that all equivalences of doing according to the content of the present patent application claim change and modification, all should be technology category of the present invention.

Claims (7)

1, a kind of channel estimation methods based on the MIMO-OFDM system is characterized in that: comprise the steps:

(1) training sequence design, phase mutual interference when making it to avoid channel estimating between each antenna pilot, described training sequence design is that the pilot frequency locations on each transmitting antenna is staggered mutually, if certain subcarrier is as the pilot tone of certain transmitting antenna, then other antenna sends zero-signal in this sub-carrier positions, thereby avoids the phase mutual interference between each antenna pilot when channel estimating

(2) adopt a kind of channel estimation method based on maximum likelihood, this likelihood function of algorithm definition and distance function estimate channel response and make the distance function minimum, thereby make the likelihood function maximum.

2, the channel estimation methods based on the MIMO-OFDM system according to claim 1 is characterized in that: have n for one _TThe MIMO-OFDM system of individual transmitting antenna, the total number of sub-carriers that is used to insert frequency pilot sign is N _p〉=n _TL, L represent the channel multi-path number, and frequency pilot sign should evenly expand to whole subcarrier zone, thereby evaluated error is distributed on the whole system bandwidth.

3, the channel estimation methods based on the MIMO-OFDM system according to claim 1 and 2 is characterized in that: establish the OFDM symbol and contain N _sIndividual subcarrier, if pilot configuration before adopting, with certain cycle with all subcarriers of an OFDM symbol pilot tone as each antenna, be applicable to slow fading channel, with the whole subcarrier K={0 of described OFDM symbol, 1 ..., N _s-1} is divided into n _TGroup: K _i={ mn _T+ i-1, m=0,1 ..., N _p/ n _T-1}, i=1,2 ... n _THere N _p=N _s, i antenna is at k _i∈ K _iSubcarrier place pilot signal transmitted, and send zero-signal at other subcarrier place.

4, the channel estimation methods based on the MIMO-OFDM system according to claim 1 and 2, it is characterized in that: if adopt comb type pilot configuration, on each OFDM symbol, all to insert pilot tone, each OFDM symbol comprises frequency pilot sign and data symbol, the mode that this pilot tone is inserted has been followed the tracks of the variation of channel preferably, be applicable to fast-changing channel, with the N of OFDM symbol _pIndividual pilot sub-carrier is divided into n _TGroup: $K_i=\{{\mathrm{mn}}_T\frac{N_s}{N_p}+i-1,m=\mathrm{0,1},\·\·\·{,N}_p/n_T-1\},,i=\mathrm{1,2}\·\·\·,n_T,$ For i antenna, at k _i∈ K _iSubcarrier place pilot signal transmitted, at k _i∈ K _j, j=1,2 ..., n _TAnd the subcarrier place of j ≠ i sends zero-signal, and sends data-signal at other subcarrier place.

5, the channel estimation methods based on the MIMO-OFDM system according to claim 1 is characterized in that: at the given channel impulse response h from i transmitting antenna to reception antenna of hypothesis _iThe time, the definition likelihood function:

$f=\left(R\right|h_i)=\frac{1}{{\left(2\mathrm{\π}{\mathrm{\σ}}_N^2\right)}^{N_s}}\exp \{-\frac{D\left(h_i\right)}{2{\mathrm{\σ}}_N^2}\}$

Wherein R is a received signal vector, σ _N ²Be the variance of channel additive white Gaussian noise, function D (h _i) be called distance function, be defined as:

$D\left(h_i\right)=\underset{j=0}{\overset{N_p/n_T-1}{\mathrm{\Σ}}}|R\left(k_i^j\right)-X_i\left(k_i^j\right)\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{\mathrm{\α}}_{i.l}\exp (-j2\mathrm{\π}\frac{k_i^j}{N_s}){|}^2$

X wherein _i(k _i ^j), R (k _i ^j) be respectively i transmitting antenna k _i ^jTransmission signal and received signal on the individual subcarrier, k _i ^jBe j pilot sub-carrier on each and every one transmitting antenna of i, L represents multipath number, α _{I, j}Be the multiple fading coefficients of l paths, work as X from antenna i to reception antenna _iWhen known, obtain channel impulse response h _i, so that likelihood function f is (R|h _i) maximum, just make distance function D (h _i) minimum.

6, according to claim 1 or 5 described channel estimation methods, it is characterized in that based on the MIMO-OFDM system:

Has n at one _TIn the MIMO-OFDM system of an individual transmitting antenna and a reception antenna, the number of sub carrier wave of OFDM is N _s, each antenna be used to insert frequency pilot sign subcarrier add up to N _p, N here _p=N _s, the channel multi-path number is L, the implementation of this algorithm may further comprise the steps in this system:

Pilot configuration before 1. if a. adopts, with whole subcarrier K={0,1 ..., N _s-1} is divided into n _TGroup: K _i={ mn _T+ i-1, m=0,1 ..., N _p/ n _T-1}, i=1,2 ... n _T, i antenna is at k _i∈ K _iSubcarrier place pilot signal transmitted, and send zero-signal at other subcarrier place

If b. adopt comb type pilot configuration, with N _pIndividual pilot sub-carrier is divided into n _TGroup: $K_i=\{{\mathrm{mn}}_T\frac{N_s}{N_p}+i-1,m=\mathrm{0,1},\·\·\·,N_p/n_T-1\},,i=\mathrm{1,2},\·\·\·,n_T,$ For i antenna, at k _i∈ K _iSubcarrier place pilot signal transmitted, at k _i∈ K _j, j=1,2 ..., n _TAnd the subcarrier place of j ≠ i sends zero-signal, and sends data-signal at other subcarrier place

2. for i antenna, by formula

$S_i\left(k_i^j\right)={\left|X_i\left(k_i^j\right)\right|}^2,Z_i\left(k_i^j\right)=X_i^{*}\left(k_i^j\right)R\left(k_i^j\right)---j=\mathrm{0,1},\·\·\·,N_p/n_T-1$

Try to achieve { S _i(k _i ^j) and { Z _i(k _i ^j), k _i ^j∈ K _i, j=0,1 ..., N _p/ n _T-1

3. ask sequence { S respectively _i(k _i ^j) and { Z _i(k _i ^j) N _p/ n _TPoint IFFT obtains { s _iAnd { z (p) } _i(p) }, p=0,1 ..., N _p/ n _T-1

4. ask sequence { s respectively _iAnd { z (p) } _i(p) } L point FFT obtains sequence { S _i ^(L)(k _i ^j) and { Z _i ^(L)(k _i ^j)

5. ask sequential element Z one by one _i ^(L)(k _i ^j) and S _i ^(L)(k _i ^j) the merchant, i.e. Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j)

6. ask sequence { Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j) L point IFFT, obtain sequence L=0,1 ..., L-1

7. will

Adopt N behind the zero padding _sPoint FFT transforms to frequency domain, the frequency response coefficient { H on just can obtaining from i transmitting antenna to each subcarrier of reception antenna _i(k) } estimated value

K=0,1 ..., N _s-1.

7, based on a kind of maximum likelihood channel estimation methods of MIMO-OFDM system, it is characterized in that:

Has n at one _TIn the MIMO-OFDM system of an individual transmitting antenna and a reception antenna, the number of sub carrier wave of OFDM is N _s, each antenna be used to insert frequency pilot sign subcarrier add up to N _p, N here _p=N _s, the channel multi-path number is L, the implementation of this algorithm may further comprise the steps in this system:

1. adopt the training sequence that is adapted to the MIMO-OFDM system, can avoid the phase mutual interference of each antenna pilot

2. for i antenna, by formula

$S_i\left(k_i^j\right)={\left|X_i\left(k_i^j\right)\right|}^2,Z_i\left(k_i^j\right)=X_i^{*}\left(k_i^j\right)R\left(k_i^j\right),j=\mathrm{0,1},\·\·\·,N_p/n_T-1$

Try to achieve { S _i(k _i ^j) and { Z _i(k _i ^j), k _i ^j∈ K _i, j=0,1 ..., N _p/ n _T-1, k _i ^jBe j pilot sub-carrier on i the transmitting antenna, K _iBe the sets of pilot sub-carriers on i the transmitting antenna, suppose that here each antenna has identical pilot tone and counts N _p/ n _T, X _i(k _i ^j), R (k _i ^j) be respectively i transmitting antenna k _i ^jTransmission signal and received signal on the individual subcarrier

3. ask sequence { S respectively _i(k _i ^j) and { Z _i(k _i ^j) N _p/ n _TPoint IFFT obtains { s _iAnd { z (p) } _i(p) }, p=0,1 ..., N _p/ n _T-1

4. ask sequence { s respectively _iAnd { z (p) } _i(p) } L point FFT obtains sequence { S _i ^(L)(k _i ^j) and { Z _i ^(L)(k _i ^j)

5. ask sequential element Z one by one _i ^(L)(k _i ^j) and S _i ^(L)(k _i ^j) the merchant, i.e. Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j)

6. ask sequence { Z _i ^(L)(k _i ^j)/S _i ^(L)(k _i ^j) L point IFFT, obtain sequence L=0,1 ..., L-1

7. will Adopt N behind the zero padding _sPoint FFT transforms to frequency domain, the frequency response coefficient { H on just can obtaining from i transmitting antenna to each subcarrier of reception antenna _i(k) } estimated value K=0,1 ..., N _s-1.

Images