CN1684379A - Method and device for evaluating channels - Google Patents

Abstract

This invention discloses a channel evaluation method including carrying out sliding window process and noise evaluation to the de-perturbed and de-expanded symbols of the public pilot channel for each path from airborne interface data to generate the optimum evaluation value by combing the primary evaluation value and the noise evaluation value of each path with the delay information applying the principle of the smallest square error evaluation. A device is also disclosed including: more than one de-perturbed de-expanded module, a smoothing process module, a noise evaluation module and a perturbed suppressing module based on the smallest square error evaluation. Every de-perturbed and de-expanded module delivers output data to the corresponding noise evaluation module and smoothing process module, which output the data to the perturbed suppressing module.

Description

A kind of method and apparatus of channel estimating

Technical field

The present invention relates to the data processing technique in the radio communication, be meant a kind of method and apparatus of channel estimating especially.

Background technology

In the signal receiving course of radio communication, there are polytype interference and distorted signals, mainly can be divided into following a few major types: (1) multiple access disturbs, though each user's spreading code is a quadrature in spread spectrum communication system, but because the existence of incomplete synchronous and a plurality of transmission paths of each user makes each user all can be subjected to other user's interference; (2) interference between multipath is because non-whole chip (chip) sampling of the nonorthogonality of scrambler and filter brings the interference between multipath; (3) interference between intersymbol and chip; (4) interference of other cell signal; (5) other type interference of noise.

More than the existence of these interference, can influence in the system receiver to the accurate reception of signal, and then the capacity of restriction wireless communication system carrying signal, therefore in the signal receiving course, at first need to estimate these interference, when restituted signal, suppress these again and disturb, to improve the performance and the power system capacity of receiver.

The channel estimation scheme of prior art is that level and smooth channel estimating is carried out in every footpath in the channel respectively, and the structure of its channel estimating is referring to shown in Figure 1.In the Traffic Channel of supposing to be received the L paths is arranged, then in channel estimating apparatus 4, includes and every footpath L descrambling and de-spreading module 11,21 one to one ..., L1, L smoothing processing module 12,22 ..., L2 and L interpolation processing module 13,23 ..., L3.

Carry out in the signal receiving course, air interface data is sent into channel estimating apparatus 4 after receiver radio frequency, automatic gain control (AGC, Automatic Gain Control).The data handling procedure that channel estimating apparatus 4 carries out is as follows:

Step 101, each descrambling and de-spreading module 11,21, ..., L1 receive respectively its air interface data in corresponding footpath, and according to the footpath deferred message in the every footpath that obtains from multipath, each descrambling and de-spreading module 11,21, ..., L1 adjusts the position of air interface data respectively to pairing footpath separately, again Common Pilot Channel (the CPICH that produces with this locality, Common Pilot Channel) scrambler, spreading code is relevant, obtain the value of symbol (Common Pilot Channel descrambling and de-spreading symbol) behind the Common Pilot Channel descrambling and de-spreading, send into the pairing smoothing processing module 12,22 in every footpath respectively, ..., L2.

Step 102, smoothing processing module 12,22 ..., L2 is slided window to the Common Pilot Channel descrambling and de-spreading symbol of sending into respectively and is handled, to weaken Gaussian noise, thereby obtain the preliminary channel estimation value in every footpath, preliminary channel estimation value is sent into the interpolation processing module 13,23 of every footpath correspondence respectively, ..., L3.

Wherein, n Common Pilot Channel descrambling and de-spreading symbol of n Common Pilot Channel descrambling and de-spreading symbol and back added the result after the Common Pilot Channel descrambling and de-spreading symbol of current location is total to 2n+1 symbol simple average before generally the result of each Common Pilot Channel descrambling and de-spreading symbol after sliding window is handled was this symbol position, wherein the value of n can suitably be selected as required, as: get n=5.

The concrete computational process that sliding window is handled is: the Common Pilot Channel descrambling and de-spreading symbol of supposing the input of smoothing processing module is chan (i), i=0, and 1,2...n, the preliminary channel estimation value that sliding window is handled back output is smoother (k), k=0,1,2...n.The smoothing processing module adds up 2n+1 Common Pilot Channel descrambling and de-spreading symbol of input, again divided by 2n+1, obtains the estimated value to middle position Common Pilot Channel descrambling and de-spreading symbol, as shown in Equation (1).

$\mathrm{smoother}\left(k\right)=\frac{1}{2n+1}\underset{i=k-n}{\overset{k+n}{\∑}}\mathrm{chan}\left(i\right),k=\mathrm{0,1,2}...---\left(1\right)$

Wherein, at initial preceding n moment formula (1) chan (i) can appear, the item of i＜0, and not this input at this moment, therefore agreement chan this moment (i) is 0.

Through this step, smoothing processing module 12,22, ..., L2 has obtained preliminary channel estimation value, but can't directly be used for the demodulation Traffic Channel, this is that to different spreading factors, the channel estimation value number that each time slot of Traffic Channel needs is different because the spreading factor of the Traffic Channel of real system has a variety ofly, but smoothing processing module 12,22 ..., each time slot of the channel estimation value that L2 obtains is fixed, as: the spreading factor of WCDMA system protocol regulation Common Pilot Channel is fixed as 256, like this descrambling and de-spreading and after the smoothing processing module each time slot channel estimation value be fixed as 10, therefore need by interpolation processing module 13,23, ..., L3 matches 10 preliminary channel estimation values on the needed different numbers of different spreading factors.

Step 103, interpolation processing module 13,23 ..., L3 is smoothing processing module 12,22 ..., on the needed number of spreading factor that 10 preliminary channel estimation values that L2 sends into match the system business channel.

The interpolation method that this step is concrete has a variety of, generally in order to realize the convenient linear interpolation that adopts.

The demodulating unit that channel estimation value after the interpolation is given receiver carries out the demodulation of Traffic Channel and other channel.

The shortcoming of existing channel method of estimation is the influence that can only weaken white Gauss noise, for non-white Gauss noise, as: the interference of multipath then can not be eliminated, thereby makes the estimation of interchannel noise produce deviation, influences the effect of signal modulation and decoding.

Summary of the invention

In view of this, main purpose of the present invention is to provide a kind of method of channel estimating, both can weaken white Gauss noise, can effectively restrain non-white Gauss noises such as multipath interference again, reduces the deviation that interchannel noise is estimated generation.

Another main purpose of the present invention provides a kind of device of channel estimating, can suppress non-white Gauss noises such as white Gauss noise and multipath interference when channel estimating, makes channel estimating more accurate.

Method according to first goal of the invention a kind of channel estimating provided by the invention comprises:

A) to the Common Pilot Channel descrambling and de-spreading symbol in every footpath in the channel that after descrambling and de-spreading is handled, obtains by air interface data, slide window respectively and handle and Noise Estimation, produce the preliminary channel estimation value and the noise estimation value in the every footpath of channel;

B) according to the footpath deferred message in each footpath of channel and each noise estimation value and preliminary channel estimation value directly of channel of step a) acquisition, produce the preferred channels estimated value in the every footpath of channel by the least mean-square error estimation criterion.

The described air interface data of this method is the air interface data after radio frequency, automatic gain control;

The process that the described descrambling and de-spreading of step a) is handled specifically comprises: the footpath deferred message that obtains the every footpath of channel, behind the position of every footpath air interface data that adjustment is received, relevant with scrambler, the spreading code of local Common Pilot Channel, obtain the Common Pilot Channel descrambling and de-spreading symbol in every footpath.

This method step is slided the process that window handles to the Common Pilot Channel descrambling and de-spreading symbol in every footpath in a) and is specially: in every footpath when the Common Pilot Channel descrambling and de-spreading symbol of pre-treatment, and the Common Pilot Channel descrambling and de-spreading symbol of each similar number adds up before and after this symbol, with the result that adds up number, obtain the preliminary channel estimation value of current Common Pilot Channel descrambling and de-spreading symbol divided by the Common Pilot Channel descrambling and de-spreading symbol that adds up.

The process that this method step carries out Noise Estimation to the Common Pilot Channel descrambling and de-spreading symbol in every footpath in a) specifically comprises: Common Pilot Channel descrambling and de-spreading symbol and self result after sliding window is handled when pre-treatment in every footpath subtract each other, to subtract each other the result who obtains and carry out mould square, result to mould square carries out filtering, obtains the noise estimation value of current Common Pilot Channel descrambling and de-spreading symbol.

The described filtering of this method is a first-order filtering.

This method step b) specifically comprise:

B1) obtain each footpath deferred message directly of channel, generate raised cosine filter multipath correlation matrix according to the raised cosine filter function;

According to each noise estimation value directly of channel that step a) obtains, generted noise covariance matrix;

According to each preliminary channel estimation value directly of channel that step a) obtains, generate the covariance matrix of channel magnitude;

B2) to step b1) the raised cosine filter multipath correlation matrix that obtains carries out matrix transpose, obtains the raised cosine filter multipath correlation matrix of transposition;

To step b1) noise covariance matrix that obtains carries out matrix inversion, obtains contrary noise covariance matrix;

To step b1) covariance matrix of the channel magnitude that obtains carries out matrix inversion, obtains the covariance matrix of inverse channel amplitude;

B3) with step b2) raised cosine filter multipath correlation matrix, contrary noise covariance matrix and the step b1 of the transposition that obtains) the raised cosine filter multipath correlation matrix that obtains carries out matrix multiple successively;

B4) with step b3) the matrix multiple result and the step b2 that obtain) covariance matrix of inverse channel amplitude of acquisition carries out the matrix addition;

B5) to step b4) the matrix addition result that obtains carries out matrix inversion;

B6) with step b5) the matrix inversion result, the step b2 that obtain) the raised cosine filter multipath correlation matrix and the contrary noise covariance matrix of the transposition that obtains carry out matrix multiple successively;

B7) with step b6) after the vector of the preliminary channel estimation value composition in every footpath of obtaining of the matrix multiple result and the step a) that obtain carried out matrix and multiplication of vectors, every result was the preferred channels estimated value in the every footpath of described channel in the gained vector.

This method step b1) described raised cosine filter multipath correlation matrix G is

τ wherein ₀..., τ _L-1Be the footpath deferred message in the every footpath of described channel, R _g(.) is described raised cosine filter function, and L is the footpath number of channel.

This method step b1) described noise covariance matrix I be I=diag (| δ ₀| ²... | δ _L-1| ²), wherein, δ ² ₀..., δ ² _L-1Be the noise estimation value in the every footpath of described channel, L is the footpath number of channel.

This method step b1) the covariance matrix H of described channel magnitude be H=diag (| h ₀| ²... | h _L-1| ²), wherein, diag represents diagonal matrix, h ₀..., h _L-1Be the preliminary channel estimation value of multipath in the every footpath of described channel, L is the footpath number of channel.

Further comprise after the described step b) of this method:

The preferred channels estimated value in the every footpath of channel that c) step b) is obtained matches on the required number of the spreading factor of channel in the system.

The described step c) of this method adopts linear interpolation that the preferred channels estimated value in the every footpath of described channel is matched in the system on the needed number of the spreading factor of channel.

Device according to another goal of the invention a kind of channel estimating provided by the invention, include and be used for the every footpath of channel data are carried out the above descrambling and de-spreading module that descrambling and de-spreading is handled, be used for the every footpath of channel data are slided the above smoothing processing module that window is handled, also comprise:

An above Noise Estimation module is used for the noise of the every footpath of channel data is estimated; And

The interference of estimating based on least mean-square error suppresses module, and the sliding window result, noise estimation value and the footpath deferred message that are used for according to each footpath data of channel produce the preferred channels estimated value in the every footpath of channel;

Described each despreading descrambling module is the air interface data and the footpath deferred message in the every footpath of receive channel respectively, dateout is sent to corresponding Noise Estimation module in every footpath and level and smooth processing module respectively, each Noise Estimation module and level and smooth processing module are sent to described interference with dateout and suppress module, disturb and suppress the dateout that module receives all Noise Estimation modules and level and smooth processing module, and the footpath deferred message in each footpath of receive channel, produce the preferred channels estimated value in the every footpath of channel.

This installs in the described Noise Estimation module and comprises: smoothing processing module, subtracter, mould square submodule and filter;

Described despreading descrambling module is sent to smoothing processing module and subtracter in the corresponding Noise Estimation module respectively with dateout, smoothing processing module in the Noise Estimation module is sent to subtracter with self dateout, subtracter is sent to mould square submodule with self dateout, mould square submodule is sent to filter with self dateout, and filter is sent to described interference with self dateout and suppresses module.

This installs in the described Noise Estimation module and comprises: subtracter, mould square submodule and filter;

Subtracter receive simultaneously the described despreading descrambling module in corresponding footpath and the dateout of level and smooth processing module, subtracter is sent to mould square submodule with dateout, a mould square submodule is sent to filter with dateout, and filter is sent to described interference with dateout and suppresses module.

This installs described filter is firstorder filter.

This installs described interference and suppresses to comprise in the module: raised cosine filter multipath correlation matrix generates covariance matrix that submodule, noise covariance matrix generate submodule, channel magnitude and generates submodule, matrix transpose submodule, three matrix inversion submodules, two matrix multiple submodules, matrix addition submodule and matrixes and multiplication of vectors submodule;

Raised cosine filter multipath correlation matrix generates footpath deferred messages that submodule receives all footpaths of described channel, and self dateout is sent to matrix transpose submodule and first matrix multiple submodule respectively;

Noise covariance matrix generates the dateout that submodule receives described all Noise Estimation modules, and self dateout is sent to first matrix inversion submodule;

The covariance matrix of channel magnitude generates the dateout that submodule receives described all smoothing processing modules, and self dateout is sent to second matrix inversion submodule;

First matrix multiple submodule receives the dateout that raised cosine filter multipath correlation matrix generates submodule, matrix transpose submodule and first matrix inversion submodule, and self dateout is sent to matrix addition submodule;

Matrix addition submodule receives the dateout of first matrix multiple submodule and second matrix inversion submodule, and self dateout is sent to the 3rd matrix inversion submodule;

The dateout of the 3rd matrix inversion submodule receiving matrix addition submodule is sent to second matrix multiple submodule with self dateout;

The data of second matrix multiple submodule receiving matrix transposition submodule, the 3rd matrix inversion submodule and first matrix inversion submodule are sent to matrix and multiplication of vectors submodule with self dateout;

Matrix and multiplication of vectors submodule receive the dateout of second matrix multiple submodule and described smoothing processing module, and self dateout is sent to described interpolation processing module.

This device further comprises: the interpolation processing module on the required number of the spreading factor that preferred channels estimated value that is used for the every footpath of channel matches system's channel;

The preferred channels estimated value in the every footpath of described channel that described interference inhibition module will produce is sent to the interpolation processing module, the data output after the interpolation processing module will be handled.

From above as can be seen, the method and apparatus of channel estimating provided by the invention is when keeping the level and smooth channel estimation methods of prior art, taken all factors into consideration the correlation between the multipath channel characteristic, utilize the least mean-square error method of estimation to the unified channel estimating of carrying out in each footpath of channel, thereby make channel estimating can either weaken white Gauss noise, can effectively restrain non-white Gauss noises such as multipath interference again, make channel estimating more approaching, make later signal modulation and decoding more accurately and reliably with actual truth.

Description of drawings

Fig. 1 is the structural representation of prior art channel estimating apparatus;

Fig. 2 is the structural representation of the channel estimating apparatus of preferred embodiment of the present invention;

Fig. 3 is the structural representation of the Noise Estimation module of channel estimating apparatus in the embodiment of the invention;

Fig. 4 is the structural representation that the interference of channel estimating apparatus in the embodiment of the invention suppresses module.

Embodiment

The present invention is further described in more detail below in conjunction with drawings and the specific embodiments.

Thinking of the present invention is that air interface data is carried out level and smooth channel estimating and Noise Estimation respectively by every footpath of channel, utilization obtains the preliminary channel estimation value and the noise estimation value in every footpath, in conjunction with the footpath deferred message in the every footpath of channel, produce the preferred channels estimated value in the every footpath of channel by the least mean-square error estimation criterion.

The structure of the channel estimating apparatus that the preferable implementation of channel estimation scheme provided by the invention is taked, referring to shown in Figure 2, and in conjunction with Fig. 3, Fig. 4.L bar footpath is arranged in the Traffic Channel of supposing to be received, in the channel estimating apparatus 6 of Fig. 2, comprise: with every footpath L descrambling and de-spreading module 11,21 one to one, ..., L1, L smoothing processing module 12,22 ..., L2 and L Noise Estimation module 14,24 ..., L4, and an interference of estimating based on least mean-square error (MMSE, Minimum Mean Square Error) suppresses module 5 and an interpolation processing module 3.Wherein, descrambling and de-spreading module 11,21 ..., L1 and level and smooth processing module 12,22, ..., identical among L2 and Fig. 1,3 of interpolation processing modules are provided with one, interpolation processing module among its function and structure and Fig. 1 is basic identical, here mainly is to be used for the vector form channel estimation results is carried out interpolation processing.

Carry out signal when receiving, air interface data through receiver radio frequency, AGC to after send into channel estimating apparatus 6.The processing procedure that channel estimating apparatus 6 is carried out is as follows:

Step 201, each descrambling and de-spreading module 11,21 ..., L1 receive respectively its air interface data in corresponding footpath, and according to the footpath deferred message in the every footpath that obtains from multipath, each descrambling and de-spreading module 11,21, ..., L1 adjusts the position of air interface data respectively to pairing footpath separately, again the scrambler of the CPICH that produces with this locality, spreading code is relevant, obtains Common Pilot Channel descrambling and de-spreading symbol, send into the pairing smoothing processing module 12 in every footpath respectively, 22 ..., L2, and also send into simultaneously the pairing Noise Estimation module 14 in every footpath, 24 ..., L4.

Step 202, smoothing processing module 12,22, ..., L2 is slided window to the Common Pilot Channel descrambling and de-spreading symbol of sending into respectively and is handled, and to weaken Gaussian noise, processing procedure is the same with the step 102 of aforementioned prior art, then, the preliminary channel estimation value in the footpath of the L after will handling is respectively sent into to disturb and is suppressed module 5.

Step 203, Noise Estimation module 14,24 ..., L4 estimates the noise of every footpath Common Pilot Channel descrambling and de-spreading symbol respectively, gives the interference of estimating based on least mean-square error with the L noise directly that estimates and suppresses module 5.

The structure of Noise Estimation module that every footpath is used for Noise Estimation is identical, referring to shown in Figure 3.The Noise Estimation module comprises: smoothing processing module 3a, subtracter 3b, mould square submodule 3c and filter 3d.

The process that Noise Estimation is carried out in every footpath is also identical, specifically comprise: after the Common Pilot Channel descrambling and de-spreading symbol of descrambling and de-spreading module output enters the Noise Estimation module, in the Noise Estimation module, the signal subtraction that smoothing processing module 3a directly sends into every footpath descrambling and de-spreading module in subtracter 3b the signal that Common Pilot Channel descrambling and de-spreading symbol carries out exporting after the smoothing processing, then, the result sent into carry out mould among the mould square submodule 3c square with a square energy, enter filter 3d at last and carry out filtering.Filtered output is exactly to force the into noise of theoretical noise variance, and the Noise Estimation result in every footpath is sent into interference supression module 5.

In the Noise Estimation module, filter 3d generally adopts firstorder filter, after the sample noise that firstorder filter is sent into mould square submodule 3c carries out Filtering Processing, makes the sample noise level more force in theoretic noise variance.The function of smoothing processing module 3a and each the smoothing processing module among structure and Fig. 2 are identical, therefore fully can be with the smoothing processing module in every footpath among Fig. 2 as the smoothing processing module 3a in the corresponding Noise Estimation module directly of its institute, like this, the dateout of the smoothing processing module in every footpath disturb to suppress the module 5 except that sending among Fig. 2, also with the subtracter 3b that sends into simultaneously in the corresponding Noise Estimation module directly of institute.

Step 204, the interference of estimating based on Minimum Mean Square Error suppresses module 5 receptions from Noise Estimation module 14,24 ..., the noise estimation value in each footpath of the channel of L4, from each footpath deferred message of channel of multipath with from smoothing processing module 12,22 ..., the preliminary channel estimation value in each footpath of the channel of L2, according to the least mean-square error estimation criterion, calculate the preferred channels estimated value in the every footpath of channel after, send into interpolation processing module 3.

In the computational process of the preferred channels estimated value in the every footpath of this step channel, can weaken the influence that multipath disturbs, thereby improve performance for estimating channel.Realize that the interference based on the Minimum Mean Square Error estimation of this process suppresses the structure of module 5, referring to shown in Figure 4, disturb and suppress to comprise in the module 5: raised cosine filter multipath correlation matrix generates submodule 5a, noise covariance matrix generates submodule 5b, the covariance matrix of channel magnitude generates submodule 5c, matrix transpose submodule 5d, three matrix inversion submodule 5e1,5e2 and 5e3, two matrix multiple submodule 5f1 and 5f2, matrix addition submodule 4g, and matrix and multiplication of vectors submodule 5h.

Disturb the data handling procedure that suppresses module 5 specifically to comprise:

Synchronous execution in step 204a-204c.

204a, raised cosine filter multipath correlation matrix generate submodule 5a and receive each footpath deferred message directly of channel that multipath provides, and by formula (2) generate and output raised cosine filter multipath correlation matrix G.

Wherein, suppose that the footpath deferred message that multipath provides is τ ₀..., τ _L-1, the raised cosine filter function is R _g(.) then raised cosine filter multipath correlation matrix G (2) generation by formula:

204b, noise covariance matrix generates submodule 5b and receives Noise Estimation module 14,24 ..., the noise estimation value in each footpath of channel of L4 output, by formula (3) generate and output multipath noise covariance matrix I.

Wherein, suppose Noise Estimation module 14,24 ..., the noise estimation value in L bar footpath is δ in the Traffic Channel of L4 output ² ₀..., δ ² _L-1, each footpath noise is independent, and then multipath noise covariance matrix I also is a diagonal matrix, promptly

I=diag (| δ ₀| ²... | δ _L-1| ²) diag represents diagonal matrix in (3) formula (3), promptly all the other elements are zero except that diagonal entry.

204c, the covariance matrix of channel magnitude generate submodule 5c reception smoothing processing module 12,22 ..., the preliminary channel estimation value in each footpath of channel of L2 output, by formula (4) generate the also covariance matrix H of delivery channel amplitude.

Wherein, suppose smoothing processing module 12,22 ..., the preliminary channel estimation value in L bar footpath is h in the Traffic Channel of L2 output ₀..., h _L-1, each footpath amplitude is uncorrelated, and then the covariance matrix H of channel magnitude is a diagonal matrix, and diagonal entry is the mould square of preliminary channel estimation value, promptly

H＝diag(|h ₀| ²...|h _L-1| ²) (4)

Then, synchronous again execution in step 204d-204f.

204d, matrix transpose submodule 5d receive the raised cosine filter multipath correlation matrix G that raised cosine filter multipath correlation matrix generates submodule 5a output, and it is carried out matrix transpose function, the transposition G of output matrix G ^T

204e, matrix inversion submodule 5e1 receives the noise covariance matrix I that noise covariance matrix generates submodule 5b output, and it is carried out matrix inversion function, the contrary I of output matrix I ^-1

204f, the covariance matrix of matrix inversion submodule 5e2 receive channel amplitude generate the covariance matrix H of the channel magnitude of submodule 5c output, and it is carried out matrix inversion function, the contrary H of output matrix H ^-1

Step 204g-204k carries out successively.

204g, the transposition G of the matrix G of matrix multiple submodule 5f1 receiving matrix transposition submodule 5d output ^T, the contrary I of the noise covariance matrix of matrix inversion submodule 5e1 output ^-1, and the raised cosine filter multipath correlation matrix G of raised cosine filter multipath correlation matrix generation submodule 5a input, carry out the matrix multiplication function twice, be output as G ^TI ^-1G.

The multiply each other G of submodule 5f1 output of 204h, matrix addition submodule 5g receiving matrix ^TI ^-1The output H of G and matrix inversion submodule 5e2 ^-1, carry out the matrix summation function, be output as G ^TI ^-1G+H ^-1

204i, the output G of matrix inversion submodule 5e3 receiving matrix addition submodule 5g ^TI ^-1G+H ^-1, output (G after the execution matrix inversion function ^TI ^-1G+H ^-1) ^-1

The invert output (G of submodule 5e3 of 204j, matrix multiple submodule 5f2 receiving matrix ^TI ^-1G+H ^-1) ^-1, the output G of matrix transpose submodule 5d ^T, and the output I of matrix inversion submodule 5e1 ^-1, carry out output (G after twice matrix multiplication function ^TI ^-1G+H ^-1) ^-1G ^TI ^-1

The multiply each other output matrix (G of submodule 5f2 of 204k, matrix and multiplication of vectors submodule 5h receiving matrix ^TI ^-1G+H ^-1) ^-1G ^TI ^-1And the preliminary channel estimation value vector of multipath of level and smooth processing module output

$\hat{h}=(h_0...h_{L-1}),$ Be output as vector after carrying out matrix and multiplication of vectors function ${(G^T{I}^{-1}G+H^{-1})}^{-1}{G}^T{I}^{-1}\hat{h}.$ In this vector each is every footpath channel estimation results that the interference of estimating based on least mean-square error suppresses the optimum of module 5 last outputs.

So, the interference inhibition module of estimating based on least mean-square error 5 has obtained optimum channel estimation value, but can't directly be used for the demodulation Traffic Channel, because it is a variety of that the spreading factor of real system Traffic Channel has, to different spreading factors, the channel estimation value number that each time slot of Traffic Channel needs is different, but the every time slot of channel estimation value that obtains based on the interference inhibition module 5 that least mean-square error is estimated is fixing, as: the spreading factor of WCDMA system protocol regulation Common Pilot Channel is fixed as 256, like this descrambling and de-spreading and after the smoothing processing module each time slot channel estimation value be fixed as 10, therefore the channel estimation value of 10 optimums of every time slot that need will obtain by interpolation processing module 3 matches on the needed different numbers of different spreading factors, as described in step 205.

Step 205, interpolation processing module 3 will disturb the channel guess value that suppresses 10 optimums that module 5 sends into to match on the needed different numbers of different spreading factors.

In this step concrete interpolation method have a variety of, generally in order to realize the convenient linear interpolations that adopt more.Channel estimation value after the interpolation can be given the demodulation that follow-up demodulating unit carries out Traffic Channel and other channel.

The channel estimation scheme of prior art is carried out separately each footpath, do not consider the correlation between the multipath channel characteristic, and the present invention has considered this correlation, combine the multipath interference effect between each footpath, the channel estimating of multipath is unified rather than independently carries out, and utilize the least mean-square error method of estimation to realize suppressing this interference, level and smooth channel estimation methods with prior art coexists simultaneously, combine closely, also suppressed the multipath interference thereby both weakened white Gauss noise, so be better than prior art on the performance greatly.

The present invention program is applicable to Wideband Code Division Multiple Access (WCDMA) wireless communication systems such as (WCDMA).

The above only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (17)

1, a kind of method of channel estimating is characterized in that, comprising:

A) to the Common Pilot Channel descrambling and de-spreading symbol in every footpath in the channel that after descrambling and de-spreading is handled, obtains by air interface data, slide window respectively and handle and Noise Estimation, produce the preliminary channel estimation value and the noise estimation value in the every footpath of channel;

B) according to the footpath deferred message in each footpath of channel and each noise estimation value and preliminary channel estimation value directly of channel of step a) acquisition, produce the preferred channels estimated value in the every footpath of channel by the least mean-square error estimation criterion.

2, method according to claim 1 is characterized in that, described air interface data is the air interface data after radio frequency, automatic gain control;

The process that the described descrambling and de-spreading of step a) is handled specifically comprises: the footpath deferred message that obtains the every footpath of channel, behind the position of every footpath air interface data that adjustment is received, relevant with scrambler, the spreading code of local Common Pilot Channel, obtain the Common Pilot Channel descrambling and de-spreading symbol in every footpath.

3, method according to claim 1, it is characterized in that, in the step a) Common Pilot Channel descrambling and de-spreading symbol in every footpath being slided the process that window handles is specially: in every footpath when the Common Pilot Channel descrambling and de-spreading symbol of pre-treatment, and the Common Pilot Channel descrambling and de-spreading symbol of each similar number adds up before and after this symbol, with the result that adds up number, obtain the preliminary channel estimation value of current Common Pilot Channel descrambling and de-spreading symbol divided by the Common Pilot Channel descrambling and de-spreading symbol that adds up.

4, method according to claim 1, it is characterized in that, the process of in the step a) Common Pilot Channel descrambling and de-spreading symbol in every footpath being carried out Noise Estimation specifically comprises: Common Pilot Channel descrambling and de-spreading symbol and self result after sliding window is handled when pre-treatment in every footpath subtract each other, to subtract each other the result who obtains and carry out mould square, result to mould square carries out filtering, obtains the noise estimation value of current Common Pilot Channel descrambling and de-spreading symbol.

5, method according to claim 4 is characterized in that, described filtering is a first-order filtering.

6, method according to claim 1 is characterized in that, step b) specifically comprises:

B1) obtain each footpath deferred message directly of channel, generate raised cosine filter multipath correlation matrix according to the raised cosine filter function;

According to each noise estimation value directly of channel that step a) obtains, generted noise covariance matrix;

According to each preliminary channel estimation value directly of channel that step a) obtains, generate the covariance matrix of channel magnitude;

B2) to step b1) the raised cosine filter multipath correlation matrix that obtains carries out matrix transpose, obtains the raised cosine filter multipath correlation matrix of transposition;

To step b1) noise covariance matrix that obtains carries out matrix inversion, obtains contrary noise covariance matrix;

To step b1) covariance matrix of the channel magnitude that obtains carries out matrix inversion, obtains the covariance matrix of inverse channel amplitude;

B3) with step b2) raised cosine filter multipath correlation matrix, contrary noise covariance matrix and the step b1 of the transposition that obtains) the raised cosine filter multipath correlation matrix that obtains carries out matrix multiple successively;

B4) with step b3) the matrix multiple result and the step b2 that obtain) covariance matrix of inverse channel amplitude of acquisition carries out the matrix addition;

B5) to step b4) the matrix addition result that obtains carries out matrix inversion;

B6) with step b5) the matrix inversion result, the step b2 that obtain) the raised cosine filter multipath correlation matrix and the contrary noise covariance matrix of the transposition that obtains carry out matrix multiple successively;

B7) with step b6) after the vector of the preliminary channel estimation value composition in every footpath of obtaining of the matrix multiple result and the step a) that obtain carried out matrix and multiplication of vectors, every result was the preferred channels estimated value in the every footpath of described channel in the gained vector.

7, method according to claim 6 is characterized in that, step b1) described raised cosine filter multipath correlation matrix G is

τ wherein ₀..., τ _L-1Be the footpath deferred message in the every footpath of described channel, R _g(.) is described raised cosine filter function, and L is the footpath number of channel.

8, method according to claim 6 is characterized in that, step b1) described noise covariance matrix I be I=diag (| δ ₀| ²... | δ _L-1| ²), wherein, δ ² ₀..., δ ² _L-1Be the noise estimation value in the every footpath of described channel, L is the footpath number of channel.

9, method according to claim 6 is characterized in that, step b1) the covariance matrix H of described channel magnitude be H=diag (| h ₀| ²... | h _L-1| ²), wherein, diag represents diagonal matrix, h ₀..., h _L-1Be the preliminary channel estimation value of multipath in the every footpath of described channel, L is the footpath number of channel.

10, method according to claim 1 is characterized in that, further comprises after the described step b):

The preferred channels estimated value in the every footpath of channel that c) step b) is obtained matches on the required number of the spreading factor of channel in the system.

11, method according to claim 10 is characterized in that, described step c) adopts linear interpolation that the preferred channels estimated value in the every footpath of described channel is matched in the system on the needed number of the spreading factor of channel.

12, a kind of device of channel estimating includes an above descrambling and de-spreading module that is used for the every footpath of channel data are carried out the descrambling and de-spreading processing, is used for the every footpath of channel data are slided the above smoothing processing module that window is handled, and it is characterized in that, also comprises:

An above Noise Estimation module is used for the noise of the every footpath of channel data is estimated; And

The interference of estimating based on least mean-square error suppresses module, and the sliding window result, noise estimation value and the footpath deferred message that are used for according to each footpath data of channel produce the preferred channels estimated value in the every footpath of channel;

Described each despreading descrambling module is the air interface data and the footpath deferred message in the every footpath of receive channel respectively, dateout is sent to corresponding Noise Estimation module in every footpath and level and smooth processing module respectively, each Noise Estimation module and level and smooth processing module are sent to described interference with dateout and suppress module, disturb and suppress the dateout that module receives all Noise Estimation modules and level and smooth processing module, and the footpath deferred message in each footpath of receive channel, produce the preferred channels estimated value in the every footpath of channel.

13, device according to claim 12 is characterized in that, comprises in the described Noise Estimation module: smoothing processing module, subtracter, mould square submodule and filter;

Described despreading descrambling module is sent to smoothing processing module and subtracter in the corresponding Noise Estimation module respectively with dateout, smoothing processing module in the Noise Estimation module is sent to subtracter with self dateout, subtracter is sent to mould square submodule with self dateout, mould square submodule is sent to filter with self dateout, and filter is sent to described interference with self dateout and suppresses module.

14, device according to claim 12 is characterized in that, comprises in the described Noise Estimation module: subtracter, mould square submodule and filter;

Subtracter receive simultaneously the described despreading descrambling module in corresponding footpath and the dateout of level and smooth processing module, subtracter is sent to mould square submodule with dateout, a mould square submodule is sent to filter with dateout, and filter is sent to described interference with dateout and suppresses module.

According to claim 14 or 15 described devices, it is characterized in that 15, described filter is a firstorder filter.

16, device according to claim 12, it is characterized in that described interference suppresses to comprise in the module: raised cosine filter multipath correlation matrix generates covariance matrix that submodule, noise covariance matrix generate submodule, channel magnitude and generates submodule, matrix transpose submodule, three matrix inversion submodules, two matrix multiple submodules, matrix addition submodule and matrixes and multiplication of vectors submodule;

Raised cosine filter multipath correlation matrix generates footpath deferred messages that submodule receives all footpaths of described channel, and self dateout is sent to matrix transpose submodule and first matrix multiple submodule respectively;

Noise covariance matrix generates the dateout that submodule receives described all Noise Estimation modules, and self dateout is sent to first matrix inversion submodule;

The covariance matrix of channel magnitude generates the dateout that submodule receives described all smoothing processing modules, and self dateout is sent to second matrix inversion submodule;

First matrix multiple submodule receives the dateout that raised cosine filter multipath correlation matrix generates submodule, matrix transpose submodule and first matrix inversion submodule, and self dateout is sent to matrix addition submodule;

Matrix addition submodule receives the dateout of first matrix multiple submodule and second matrix inversion submodule, and self dateout is sent to the 3rd matrix inversion submodule;

The dateout of the 3rd matrix inversion submodule receiving matrix addition submodule is sent to second matrix multiple submodule with self dateout;

The data of second matrix multiple submodule receiving matrix transposition submodule, the 3rd matrix inversion submodule and first matrix inversion submodule are sent to matrix and multiplication of vectors submodule with self dateout;

Matrix and multiplication of vectors submodule receive the dateout of second matrix multiple submodule and described smoothing processing module, and self dateout is sent to described interpolation processing module.

17, device according to claim 12 is characterized in that, further comprises: the interpolation processing module on the required number of the spreading factor that preferred channels estimated value that is used for the every footpath of channel matches system's channel;

The preferred channels estimated value in the every footpath of described channel that described interference inhibition module will produce is sent to the interpolation processing module, the data output after the interpolation processing module will be handled.

Images