CN101188447B - A method and device for carrier frequency deviation estimation - Google Patents

Abstract

The embodiment of the invention discloses a carrier frequency offset estimation method, and includes: the transmitting signal of an antenna is received and then matched to be filtered, a sampling time is confirmed after synchronization, the filtered result is sampled to obtain the received data serial, and the received data serial of the receiving antenna forms a received data vector; according to the received data vector, the practice matrix which is formed by the practice serial, the related properties of MIMO channel and the noise features of the channel, an average log-likelihood function between the received data vector and the practice matrix is constituted; according to the average log-likelihood function, the carrier offset estimation of the MIMO channel is confirmed. The method can estimate the carrier frequency offset in the related MIMO channel with accuracy, and improve the signal detection performance of the MIMO system. The embodiment of the invention also discloses a carrier frequency offset estimation device.

Description

A kind of method and apparatus of Nonlinear Transformation in Frequency Offset Estimation

Technical field

The present invention relates to the signal detection technique of multiple-input, multiple-output (MIMO) system, the method and apparatus of Nonlinear Transformation in Frequency Offset Estimation under particularly a kind of relevant mimo channel.

Background technology

In recent years, the MIMO technology has obtained people's extensive concern.In mimo system, make a start and receiving end all adopt a plurality of antennas that wireless signal is transmitted and received, can make full use of the space diversity of wireless channel, thereby obviously improve the spectrum efficiency of system.

Fig. 1 is the structural representation of mimo system.The same with other wireless communication systems, in mimo system, because the relative motion between transmitter and the receiver and the frequency error of transmitter and receiver oscillator inevitably can produce carrier wave frequency deviation (CFO), and the existence of carrier wave frequency deviation can make the systematic function severe exacerbation.Therefore before detecting, must carry out Nonlinear Transformation in Frequency Offset Estimation accurately, and frequency deviation is compensated.

The basic thought of frequency offset estimating is: when carrying out frequency offset estimating, adopt specific training sequence to carry out, this training sequence all has preservation at transmitting-receiving two-end, make a start training sequence modulation back is sent out by transmitting antenna, receiver carries out demodulation to it after utilizing reception antenna to receive the receiving sequence of training sequence correspondence, and by certain algorithm, compare with the training sequence of self preserving, thereby determine carrier wave frequency deviation.

At present, when studying the frequency offset estimating problem in the mimo system, suppose that all each bar branch road of mimo channel is separate,, carry out frequency offset estimating as prerequisite.Such frequency offset estimating is not in fact considered the correlation between each branch road of mimo channel, therefore has certain limitation in actual applications.

In the practical MIMO channel, each bar branch road usually has certain correlation, and in this type of mimo system, utilizing separate with mimo channel is that the frequency deviation estimating method of prerequisite is when carrying out frequency offset estimating, its accuracy of frequency offset estimation can reduce, thereby further reduces the signal detection performance of system.

Summary of the invention

In view of this, the embodiment of the invention provides a kind of method of Nonlinear Transformation in Frequency Offset Estimation, can accurately estimate the carrier wave frequency deviation under the relevant mimo channel, thereby improves the signal detection performance of system.

The embodiment of the invention also provides a kind of device of Nonlinear Transformation in Frequency Offset Estimation, utilizes this device can accurately estimate relevant mimo channel carrier wave frequency deviation down, thus the signal detection performance of raising system.

For achieving the above object, the embodiment of the invention adopts following technical scheme:

A kind of method of Nonlinear Transformation in Frequency Offset Estimation, this method comprises:

Reception antenna receives and to transmit, and this is transmitted carries out matched filtering, determines sampling instant through back synchronously, and the filtering result is sampled obtains receiving data sequence, and with the receiving data sequence composition reception data vector of reception antenna;

According to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix;

According to average log-likelihood function, determine the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel.

A kind of device of Nonlinear Transformation in Frequency Offset Estimation, this device comprises: receiver module, likelihood function module and frequency deviation estimating modules, wherein,

Described receiver module, be used for receiving and transmit, and this is transmitted carry out matched filtering, after synchronously, determine sampling instant, the filtering result sampled obtain receiving data sequence, and the receiving data sequence of reception antenna formed receive data vector, should receive data vector and send to described likelihood function module;

Described likelihood function module, be used to receive the reception data vector that described receiver module sends, and according to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix, should send to described frequency deviation estimating modules by average log-likelihood function;

Described frequency deviation estimating modules is used for determining the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel according to the average log-likelihood function that receives.

By technique scheme as can be seen, the receiving data sequence after the embodiment of the invention will be carried out matched filtering to the signal that receives on the reception antenna and be sampled is combined to form the reception data vector; According to this training matrix, the correlation properties of mimo channel and noise characteristic of channel that receives data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix; Last according to average log-likelihood function, determine the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel.Because the embodiment of the invention has used the correlation properties of mimo channel when the average log-likelihood function of structure, therefore serve as according to the Nonlinear Transformation in Frequency Offset Estimation value of determining with this average log-likelihood function, can be applied in the mimo channel that each branch road has correlation, be implemented in and accurately estimate carrier wave frequency deviation in the relevant mimo channel, improve system signal and detect performance.

Description of drawings

Fig. 1 is the structural representation of mimo system.

Fig. 2 is the overview flow chart of carrier frequency bias estimation under the relevant mimo channel of the embodiment of the invention.

Fig. 3 is the overall construction drawing of carrier wave frequency deviation estimation device under the relevant mimo channel of the embodiment of the invention.

The process that Fig. 4 modulates and launches training sequence for transmitting terminal in the embodiment of the invention one.

Fig. 5 is the particular flow sheet of carrier frequency bias estimation under the relevant mimo channel in the embodiment of the invention one.

Fig. 6 is the concrete structure figure of carrier wave frequency deviation estimation device under the relevant mimo channel in the embodiment of the invention two.

Embodiment

For making purpose of the present invention, technological means and advantage clearer,, the present invention is described in further details below in conjunction with the accompanying drawing embodiment that develops simultaneously.

Fig. 2 is the overview flow chart of carrier frequency bias estimation under the relevant mimo channel of the embodiment of the invention.As shown in Figure 2, this method comprises:

Step 201, reception antenna receive and to transmit, and this is transmitted carry out matched filtering, determine sampling instant through back synchronously, and the filtering result is sampled obtains receiving data sequence, and with the receiving data sequence composition reception data vector of reception antenna.

Step 202, according to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix.

Step 203 according to average log-likelihood function, is determined the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel.

Fig. 3 is the overall construction drawing of carrier wave frequency deviation estimation device under the relevant mimo channel of the embodiment of the invention.As shown in Figure 3, this device 300 comprises: receiver module 310, likelihood function module 320 and frequency deviation estimating modules 330, wherein,

In this device 300, receiver module 310, be used for receiving and transmit, and this is transmitted carry out matched filtering, after synchronously, determine sampling instant, the filtering result sampled obtains receiving data sequence, and the receiving data sequence of all reception antennas formed receives data vector, should receive data vector and send to likelihood function module 320.

Likelihood function module 320, be used to receive the reception data vector that described receiver module sends, and according to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix, should send to frequency deviation estimating modules 330 by average log-likelihood function.

Frequency deviation estimating modules 330 is used for determining the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel according to the average log-likelihood function that receives.

The above-mentioned overview that is Nonlinear Transformation in Frequency Offset Estimation under the mimo channel of the embodiment of the invention is below by specific embodiment explanation the specific embodiment of the present invention.

Embodiment one:

In the present embodiment, make a start and receiving end all adopts centralized antenna, and make a start and receiving end all adopts the mimo system of same oscillator.Like this, the time delay and the frequency deviation of making a start between each reception antenna of each transmitting antenna and receiving end are all identical, and promptly whole M IMO system only has a time delay and a frequency deviation.System channel is the flat fading mimo channel, can have coefficient correlation arbitrarily between its each branch road, but the coefficient correlation between each branch road can not be 1 entirely, also is that the characteristic of channel of each branch road is incomplete same.In real system, the identical probability of the characteristic of channel of each branch road is very little.

Suppose at M _TTransmit antennas, M _RIn the mimo system of root reception antenna, utilize training sequence that carrier wave frequency deviation is estimated.

The process that Fig. 4 modulates and launches training sequence for transmitting terminal in the embodiment of the invention one.As shown in Figure 4, this process comprises:

(1), is one group of training sequence: a of each emitting antenna selecting at transmitting terminal _l[n], l=1,2 ..., M _TN=1,2 ..., N, wherein N is a training sequence length, l is the index of transmitting antenna, a _l[n] can be arbitrary sequence, as long as the sequence that keeps transmitting-receiving two-end to preserve is identical.

(2) with the training sequence a on each transmitting antenna _l[n] by pulse shaping filter, multiplies each other with impulse waveform φ (t) respectively, obtains M _TTransmitting on the individual transmitting antenna:

$s_l\left(t\right)=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{a}_l\left[n\right]\mathrm{\φ}(t-\mathrm{nT}),l=\mathrm{1,2},...,M_T$

Wherein T is a symbol duration.

(3) with M _TS emission signal s on the individual transmitting antenna _l(t) pass through M respectively _TIndividual transmission antennas transmit is gone out.

At receiving terminal, receive through the transmitting of relevant mimo channel, thereby and this signal handled obtain frequency offset estimating.Particularly, receiving end processing to received signal comprises front-end processing and two parts of frequency offset estimating.

Fig. 5 is the particular flow sheet of carrier frequency bias estimation under the relevant mimo channel in the embodiment of the invention one.As shown in Figure 5, this method comprises:

Step 501, each reception antenna received signal is also carried out matched filtering.

In this step, all reception antennas to signal receive, the process of matched filtering is identical.Here be that example illustrates this process with k reception antenna.

If the time domain specification of receiving filter is φ (t), be matched with the φ (t) that makes a start.Obtaining k the baseband receiving signals on the reception antenna after the filtering is

$r_k\left(t\right)=\underset{l=1}{\overset{M_T}{\mathrm{\Σ}}}{h}_{\mathrm{kl}}{e}^{\mathrm{j\ω}(t-\mathrm{\τ})}\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{a}_l\left[n\right]g(t-\mathrm{nT}-\mathrm{\τ})+n_k(t-\mathrm{\τ}),k=\mathrm{1,2},...,M_R$

Wherein, M _RBe the reception antenna number, g (t)=φ (t) * φ (t), * represents convolution, h _KlRepresent the channel gain between l transmitting antenna and k the reception antenna, τ and ω be respectively receiving end and make a start between time delay and (angle) frequency deviation, n _k(t) the bilateral power spectral density of expression is N ₀/ 2 zero-mean additive white Gaussian noise (AWGN).

Step 502, receiver the deadline synchronously after, determine sampling instant, and the filtering result sampled.

In this step, the baseband signal after the matched filtering on each reception antenna is sampled with the speed of 1/T, obtain receiving data sequence.With k reception antenna is example, to being sampled as by reception antenna:

$r_k\left(\mathrm{mT}\right)=r_k(\mathrm{mT}+\mathrm{\τ})$

$=\underset{l=1}{\overset{M_T}{\mathrm{\Σ}}}{h}_{\mathrm{kl}}{e}^{\mathrm{jm\ωT}}\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{a}_l\left[n\right]g(\mathrm{mT}-\mathrm{nT})+n_k\left(\mathrm{mT}\right)---\left(1\right)$

Definition ε=ω T is normalization (angle) frequency deviation, also is the object of Nonlinear Transformation in Frequency Offset Estimation.Simultaneously, adopt the transmitted waveform that satisfies Nyquist criterion, promptly g (t) satisfies:

$g\left(\mathrm{nT}\right)=\left\{\begin{array}{cc}1& n=0\\ 0& n\≠0\end{array}\right.---\left(2\right)$

(2) formula substitution (1) promptly can be obtained k the receiving data sequence on the reception antenna:

$r_k\left[m\right]=\underset{l=1}{\overset{M_T}{\mathrm{\Σ}}}{h}_{\mathrm{kl}}{e}^{\mathrm{jm\ϵ}}{a}_l\left[m\right]+n_k\left[m\right],m=\mathrm{1,2},...,N$

Step 503, obtain receiving data sequence on all reception antennas according to the mode of step 501 and step 502 after, with this M _RReceiving data sequence r on the individual reception antenna _k[m] is arranged in following reception data vector:

$r={[r_1^T,r_2^T,...,r_{M_R}^T]}^T---\left(3\right)$

Wherein, r _k=[r _k[1], r _k[2] ..., r _k[N]] ^TRepresent k the receiving data sequence on the reception antenna, r _kI symbol on k reception antenna of [i] expression in the receiving data sequence.

Step 504, according to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix.

In this step, the concrete mode that structure receives the average log-likelihood function between data vector and training matrix is:

Consider Rayleigh flat fading channel model, then channel gain h _KlFor having zero-mean circulation symmetry multiple Gauss (ZMCSCG) stochastic variable, the i.e. h of unit variance _Kl～CN (0,1).Simultaneously the embodiment of the invention has also been considered the correlation between the channel gain of each bar branch road of mimo channel, and its cross-correlation matrix is defined as R _h=E{hh ^H, wherein $h={[h_1^T,h_2^T,...,h_{M_R}^T]}^T.$ In addition, suppose that the noise on each reception antenna is separate, then n _kZMCSCG sequence of random variables for separate has $E\left\{{\mathrm{nn}}^H\right\}=N_0{I}_{N{M}_R},$ Wherein, $n={[n_1^T,n_2^T,...,n_{M_R}^T]}^T,$ N ₀One-sided power spectrum density for white Gaussian noise.

The conditional likelihood function of received signal under unknown parameter ε and h can be expressed as

$p\left(r\right|\mathrm{\ϵ},h)={\left(\mathrm{\π}{N}_0\right)}^{-N{M}_R}\exp \{-\frac{1}{N_0}\underset{k=1}{\overset{M_R}{\mathrm{\Σ}}}{\left|\right|r_k-\mathrm{\Γ}\left(\mathrm{\ϵ}\right)A{h}_k\left|\right|}^2\}---\left(4\right)$

Wherein, $r={[r_1^T,r_2^T,...,r_{M_R}^T]}^T$ Be the r shown in the formula (1), Г (ε)=diag{e ^{J ε}, e ^{J2 ε}..., e ^{JN ε}.Order $Z\left(\mathrm{\ϵ}\right)=\mathrm{\Γ}\left(\mathrm{\ϵ}\right)A\⊗I_{M_R},$ Wherein

The Kronecker product of representing matrix, M _RBe the number of reception antenna in the mimo system, $A=[a_1,a_2,...,a_{M_T}]$ Be training matrix, $a_l={\left[a_l\right[1],a_l[2],...,a_l[N\left]\right]}^T$ It is the training sequence that the l transmit antennas uses.Then formula (4) becomes

$p\left(r\right|\mathrm{\ϵ},h)={\left(\mathrm{\π}{N}_0\right)}^{-N{M}_R}\exp \{-\frac{1}{N_0}{\left|\right|r-Z\left(\mathrm{\ϵ}\right)h\left|\right|}^2\}$

$={\left(\mathrm{\π}{N}_0\right)}^{-{\mathrm{NM}}_R}\exp \{-\frac{1}{N_0}[\mathrm{const}-2\mathrm{Re}\left\{r^{H}Z\left(\mathrm{\ϵ}\right)h\right\}+h^H{Z}^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)h\left]\right\}---\left(5\right)$

Wherein, () ^HThe He Mite conjugation is got in expression.From formula (5), as can be seen, except parameter ε to be estimated, also comprise unknown parameter h in the conditional likelihood function.In the present embodiment, suppose that channel is a Rayleigh flat fading channel model, then likelihood function is asked statistical average about h, obtain average log-likelihood function.

In order to construct average log-likelihood function, present embodiment at first needs the conditional likelihood function table is shown as real number form.Order

$\hat{r}={\left[\mathrm{Re}\right\{r^{H}Z\left(\mathrm{\ϵ}\right)\},-\mathrm{Im}\{r^{H}Z\left(\mathrm{\ϵ}\right)\}]}^T$

$\hat{h}={[\mathrm{Re}{\left\{h\right\}}^T,\mathrm{Im}{\left\{h\right\}}^T]}^T$

$\hat{Z}=\frac{1}{2}\left[\begin{array}{cc}\mathrm{Re}\left\{Z^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)\right\}& -\mathrm{Im}\left\{Z^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)\right\}\\ \mathrm{Im}\left\{Z^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)\right\}& \mathrm{Re}\left\{Z^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)\right\}\end{array}\right]$

Then have $2\mathrm{Re}\left\{r^{H}Z\left(\mathrm{\ϵ}\right)h\right\}=2{\hat{r}}^T\hat{h},$ $h^H{Z}^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)h=2{\hat{h}}^T\hat{Z}\hat{h},$ Thereby the conditional likelihood function becomes

$p\left(\hat{r}\right|\mathrm{\ϵ},\hat{h})={\left(\mathrm{\π}{N}_0\right)}^{-N{M}_R}\exp \{-\frac{1}{N_0}[\mathrm{const}-2{\hat{r}}^T\hat{h}+2{\hat{h}}^T\hat{Z}\hat{h}\left]\right\}$

Under the Rayleigh flat fading channel, the joint probability density function of channel gain h (PDF) is $p\left(h\right)={\mathrm{\π}}^{-M_R{M}_T}\exp \{-h^H{R}_h^{-1}h\},$ Adopt and top same method order ${\hat{R}}_h^{-1}=\frac{1}{2}\left[\begin{array}{cc}\mathrm{Re}\left\{R_h^{-1}\right\}& -\mathrm{Im}\left\{R_h^{-1}\right\}\\ \mathrm{Im}\left\{R_h^{-1}\right\}& \mathrm{Re}\left\{R_h^{-1}\right\}\end{array}\right],$ Then PDF can be expressed as real number form: $p\left(\hat{h}\right)={\mathrm{\π}}^{-M_R{M}_T}\exp \{-2{\hat{h}}^T{\hat{R}}_h^{-1}\hat{h}\}.$

Pass through the conditional likelihood function then

Right

Ask statistical average, obtain average log-likelihood function:

$p\left(\hat{r}\right|\mathrm{\ϵ})=\∫p\left(\hat{r}\right|\mathrm{\ϵ},\hat{h})p\left(\hat{h}\right)d\hat{h}$

$=\mathrm{const}\×\exp \left\{\frac{1}{N_0}{\hat{r}}^T{(2\hat{Z}+2N_0{\hat{R}}_h^{-1})}^{-1}\hat{r}\right\}$

Above average log-likelihood function is taken the logarithm, just obtains log-likelihood function:

$\ln p\left(\hat{r}\right|\mathrm{\ϵ})=\mathrm{const}+\frac{1}{N_0}{\hat{r}}^T{(2\hat{Z}+2N_0{\hat{R}}_h^{-1})}^{-1}\hat{r}$

Utilize the relation between real number matrix and the complex matrix, can the average log-likelihood function that obtains be reduced to plural form:

$\ln p\left(r\right|\mathrm{\ϵ})=\mathrm{const}+\frac{1}{N_0}{r}^{H}Z\left(\mathrm{\ϵ}\right){[Z^H\left(\mathrm{\ϵ}\right)Z\left(\mathrm{\ϵ}\right)+N_0{R}_h^{-1}]}^{-1}{Z}^H\left(\mathrm{\ϵ}\right)r---\left(6\right)$

At last, formula (6) is carried out some arrangements, just obtains the average log-likelihood function of constructing:

$f\left(\mathrm{\ϵ}\right)=\ln p\left(r\right|\mathrm{\ϵ})=r^H[\mathrm{\Γ}\left(\mathrm{\ϵ}\right)\⊗I_{M_R}]G[{\mathrm{\Γ}}^H\left(\mathrm{\ϵ}\right)\⊗I_{M_R}]r---\left(7\right),$

Wherein,

$G=(A\⊗I_{M_R}){(A^{H}A\⊗I_{M_R}+N_0{R}_h^{-1})}^{-1}(A^H\⊗I_{M_R}).$

Also promptly, the average log-likelihood function of constructing in this step is $f\left(\mathrm{\ϵ}\right)=\ln p\left(r\right|\mathrm{\ϵ})=r^H[\mathrm{\Γ}\left(\mathrm{\ϵ}\right)\⊗I_{M_R}](A\⊗I_{M_R}){(A^{H}A\⊗I_{M_R}+N_0{R}_h^{-1})}^{-1}(A^H\⊗I_{M_R})[{\mathrm{\Γ}}^H\left(\mathrm{\ϵ}\right)\⊗I_{M_R}]r.$ By as can be seen above-mentioned, the average log-likelihood function of constructing in the embodiment of the invention is that the their cross correlation according to mimo channel carries out, so its likelihood function is applicable to relevant mimo channel.

Step 505 according to average log-likelihood function, is determined the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel.

In this step, determine that the concrete mode of the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel is, the value of carrier wave frequency deviation when obtaining average log-likelihood function and getting maximum, and with this value as the Nonlinear Transformation in Frequency Offset Estimation value.

So far, the method flow of the embodiment of the invention one finishes.

By above-mentioned formula (7) as can be seen, in fact, constructing average log-likelihood function and can be divided into for two steps and carry out, at first is the structure of G matrix, then is the structure of average log-likelihood function in the formula (7).This two step all carries out after forming the reception data vector in embodiment one.

In fact, the structure of G matrix also can carry out before the reception data vector forms.Because the structure of G matrix only needs the their cross correlation of channel, the real time data that does not need system, therefore can in system, preserve the G matrix, thereby reduce overhead, after the data that receive the training sequence correspondence, receive the formation of data vector again and according to the process of the average log-likelihood function of G matrix construction.

By the Nonlinear Transformation in Frequency Offset Estimation that above-mentioned flow process obtains,, serve as promptly can be applicable to relevant mimo channel therefore according to the frequency offset estimating of carrying out with this likelihood function because the structure of its average log-likelihood function carries out according to relevant mimo channel.

Embodiment two:

Fig. 6 is the concrete structure figure of carrier wave frequency deviation estimation device under the relevant mimo channel in the embodiment of the invention two.As shown in Figure 6, this device 600 comprises: receiver module 610, likelihood function module 620 and frequency deviation estimating modules 630.Wherein, receiver module 610 comprises reception submodule 611, matched filtering submodule 612, synchronous submodule 613 and sampling submodule 614.

In the receiver module 610 of this device 600, receive submodule 611, be used for reception and transmit, and this signal is sent to matched filtering submodule 612 and synchronous submodule 613.Matched filtering submodule 612 is used for the signal that receives submodule 611 forwardings is carried out matched filtering, and filtered result is sent to synchronous submodule 613.

Synchronous submodule 613, be used to receive the signal of submodule 611 forwardings and the filtering result that matched filtering submodule 612 sends, and definite sampling instant, the matched filtering result that the corresponding sampling instant on each reception antenna is begun sends to sampling submodule 614.Sampling submodule 614, be used for the matched filtering result who receives sampled and obtain receiving data sequence, and the receiving data sequence of all reception antennas formed receive data vector, send to the average log-likelihood function constructor module 622 in the likelihood function module 620.

In likelihood function module 620, G matrix construction submodule 621, be used for training matrix, the correlation properties of mimo channel and the noise characteristic of channel, construct the G matrix, and the G matrix of structure is sent to average log-likelihood function submodule 622 according to the training sequence formation.Average log-likelihood function constructor module 622, be used for receiving the reception data vector of receiver module 610 sampling modules 614 transmissions and the G matrix that G matrix construction submodule 621 sends, and, will construct the result and send to frequency deviation estimating modules 630 according to receiving data vector and the average log-likelihood function of G matrix construction.

Frequency deviation estimating modules 630 is used for determining the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel according to the average log-likelihood function that receives.

By above-mentioned two embodiment as seen, the receiving data sequence after the embodiment of the invention will be carried out matched filtering to the signal that receives on each reception antenna and be sampled is combined to form the reception data vector; According to this training matrix, the correlation properties of mimo channel and noise characteristic of channel that receives data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix; Last according to average log-likelihood function, determine the Nonlinear Transformation in Frequency Offset Estimation value of relevant mimo channel.Because the present invention has used the correlation properties of mimo channel when the average log-likelihood function of structure, therefore serve as according to the Nonlinear Transformation in Frequency Offset Estimation value of determining with this average log-likelihood function, can be applied in the mimo channel that each branch road has correlation, be implemented in and accurately estimate carrier wave frequency deviation in the relevant mimo channel, improve the purpose that system signal detects performance.

Simultaneously, since the embodiment of the invention based on average maximum likelihood (ML) do not need channel information, only need channel ASSOCIATE STATISTICS information, therefore when carrying out the average log-likelihood function of actual configuration, can at first in system, preserve the G matrix, the average log-likelihood function of G matrix construction according to receiving data and preservation has improved the processing capability in real time when carrying out carrier wave frequency deviation.

Being preferred embodiment of the present invention only below, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. the method for a Nonlinear Transformation in Frequency Offset Estimation is characterized in that, this method comprises:

Reception antenna receives and to transmit, and this is transmitted carries out matched filtering, determines sampling instant through back synchronously, and the filtering result is sampled obtains receiving data sequence, and with the receiving data sequence composition reception data vector of reception antenna;

According to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix;

According to average log-likelihood function, determine the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel.

2. method according to claim 1 is characterized in that, described matched filtering is that the filter that the pulse shaping filtering of utilization and transmitting terminal is complementary carries out filtering to the received signal.

3. method according to claim 1 is characterized in that, described receiving data sequence with reception antenna is formed the reception data vector and is:

Wherein, r _k=[r _k[1], r _k[2] ..., r _k[N]] ^T, be the reception data vector of k root reception antenna, M _RBe the number of reception antenna in the mimo system, N is a training sequence length.

4. method according to claim 1 is characterized in that, the training matrix that described training sequence constitutes is: the training sequence that adopts with every transmit antennas is a column vector, with the sequence arrangement of column vector according to transmitting antenna, composing training matrix.

5. method according to claim 1 is characterized in that, the average log-likelihood function that described structure receives between data vector and training matrix comprises,

Structure G matrix,

According to the G matrix, construct average log-likelihood function

Wherein, Γ (ε)=diag{e ^{J ε}, e ^{J2 ε}..., e ^{JN ε}, The Kronecker product of representing matrix, () ^HThe He Mite conjugation is got in expression,

Be unit matrix, M _RBe the number of reception antenna in the mimo system,

Be training matrix, a _l=[a _l[1], a _l[2] ..., a _l[N]] ^TBe the training sequence that the l transmit antennas uses, R _hBe the correlation matrix of mimo channel, N ₀Be the one-sided power spectrum density of white Gaussian noise, r is the reception data vector that the receiving data sequence of reception antenna is formed,

r _k=[r _k[1], r _k[2] ..., r _k[N]] ^T, be the reception data vector of k root reception antenna, M _TBe the number of transmitting antenna in the mimo system, N is a training sequence length.

6. method according to claim 5 is characterized in that, described structure G matrix carries out before forming the reception data vector, or carries out after forming the reception data vector.

7. according to any described method in the claim 1 to 6, it is characterized in that, the Nonlinear Transformation in Frequency Offset Estimation value of described definite mimo channel is: when obtaining average log-likelihood function and getting maximum, the value of carrier wave frequency deviation, and with this value as the Nonlinear Transformation in Frequency Offset Estimation value.

8. the device of a Nonlinear Transformation in Frequency Offset Estimation is characterized in that, this device comprises: receiver module, likelihood function module and frequency deviation estimating modules, wherein,

Described receiver module, be used for receiving and transmit, and this is transmitted carry out matched filtering, after synchronously, determine sampling instant, the filtering result sampled obtain receiving data sequence, and the receiving data sequence of reception antenna formed receive data vector, should receive data vector and send to described likelihood function module;

Described likelihood function module, be used to receive the reception data vector that described receiver module sends, and according to the training matrix, the correlation properties of mimo channel and the noise characteristic of channel that receive data vector, training sequence formation, structure receives the average log-likelihood function between data vector and training matrix, should send to described frequency deviation estimating modules by average log-likelihood function;

Described frequency deviation estimating modules is used for determining the Nonlinear Transformation in Frequency Offset Estimation value of mimo channel according to the average log-likelihood function that receives.

9. device according to claim 8 is characterized in that, described receiver module comprises reception submodule, matched filtering submodule, synchronous submodule and sampling submodule, wherein,

Described reception submodule is used for reception and transmits, and this signal is sent to described matched filtering submodule and described synchronous submodule;

Described matched filtering submodule is used for the signal that described reception submodule is transmitted is carried out matched filtering, and filtered result is sent to described synchronous submodule;

Described synchronous submodule is used to receive the signal of described reception submodule forwarding and the filtering result that described matched filtering submodule sends, and definite sampling instant, and the filtering result that corresponding sampling instant is begun sends to described sampling submodule;

Described sampling submodule, being used for the filtering result who receives sampled obtains receiving data sequence, and the receiving data sequence of reception antenna formed receives data vector, sends to described likelihood function module.

10. according to Claim 8 or 9 described devices, it is characterized in that described likelihood function module comprises G matrix construction submodule and average log-likelihood function constructor module, wherein,

Described G matrix construction submodule is used for training matrix, the correlation properties of mimo channel and the noise characteristic of channel according to the training sequence formation, structure G matrix

And with the structure the G matrix send to described average log-likelihood function constructor module;

Described average log-likelihood function constructor module is used to receive the reception data vector of described receiver module transmission and the G matrix that described G matrix construction submodule sends, and according to receiving data vector and the average log-likelihood function of G matrix construction To construct the result and send to described frequency deviation estimating modules,

Wherein, Γ (ε)=diag{e ^{J ε}, e ^{J2 ε}..., e ^{JN ε},

The Kronecker product of representing matrix,

The He Mite conjugation is got in expression,

Be unit matrix, M _RBe the number of reception antenna in the mimo system,

Be training matrix, a _l=[a _l[1], a _l[2] ..., a _l[N]] ^TBe the training sequence that the l transmit antennas uses, R _hBe the correlation matrix of mimo channel, N ₀Be the one-sided power spectrum density of white Gaussian noise, r is the reception data vector that the receiving data sequence of reception antenna is formed,

r _k=[r _k[1], r _k[2] ..., r _k[N]] ^T, be the reception data vector of k root reception antenna, M _TBe the number of transmitting antenna in the mimo system, N is a training sequence length.

Images