CN101702696B - Implement method and device of channel estimation - Google Patents

Abstract

The invention discloses an implement method and a device of channel estimation. The method comprises the following steps: converting the extracted channel estimation result of the reference signal to time domain to obtain the time domain initial channel estimation sequence of the reference signal; weighing the time domain initial channel estimation sequence to obtain the time domain channel shock response sequence of the reference signal; obtaining the power delay spectrum of the reference signal according to the time domain channel shock response sequence, and obtaining the relevant function of the frequency domain of the reference signal according to the power delay spectrum; and carrying out normalization treatment on the relevant function of the frequency domain, and carrying out Wiener filtering calculation according to the normalized relevant function of frequency domain to obtain the final channel estimation results. In the invention, corresponding Wiener filtering coefficient can be obtained according to the real situation of RS, thus ensuring that the channel estimation process can be better carried out based on RS, avoiding the problem that the channel estimation process does not fit for the change of the actual environment due to the adoption of fixed Wiener filtering coefficient, and being capable of effectively improving the performance and accuracy of channel estimation.

Description

The implementation method of channel estimating and device

Technical field

The present invention relates to the communications field, relate in particular to a kind of implementation method and device of channel estimating.

Background technology

At third generation partner program (3rd Generation Partnership Project; Abbreviate 3GPP as) wireless standard of future generation; Promptly; Long Term Evolution (Long Term Evolution; Abbreviate LTE as) first-selected quadrature article minute multiplexing (Orthogonal Frequency Division Multiplexing abbreviates OFDM as) technology of broadband wireless communications that the descending technology of the basic transmission technology of standard has adopted, the OFDM technology has the higher availability of frequency spectrum and good anti-multipath interference performance.In the downlink receiver design, channel estimation technique is a key technology that improves systematic function.

Usually; The LTE down channel is estimated based on frequency domain reference signal (Reference Signal; Abbreviate RS as) carry out; RS is distributed on the specific subcarrier on time domain and the frequency domain regularly, and Fig. 1 is the RS distribution schematic diagram that adopts in the correlation technique under single transmit antenna and regular circulation prefix (Cyclic Prefix abbreviates CP as) the length situation.In Fig. 1, R ₀Expression RS position.Usually, channel estimating can briefly be divided into for three steps: insert in the RS channel estimating, frequency domain and time domain in insert.Wherein, the RS channel estimating is meant to be estimated the channel of RS position, as inserting processing basis in carrying out.Then, carry out frequency domain and time domain interpolation respectively, thereby obtain the channel estimating that all use sub-carrier positions.The RS channel estimating adopts the lower least square of complexity (LS) method usually, specifically can be expressed as following formula:

${\hat{H}}_{\mathrm{RS}}\left(k\right)=r_{\mathrm{RS}}\left(k\right)/t_{\mathrm{RS}}\left(k\right)$ Formula (1)

In formula (1), r _RS(k) and t _RS(k) represent on the k number of sub-carrier position RS signal that receives and send respectively.

For fear of the too much data of buffer memory, insert the general simple linear interpolation of using in the time domain.Compare above two parts, insert in the frequency domain and use more complicated interpolation algorithm usually so that premium properties to be provided.Wherein, receive Wiener based on the dimension of linear least mean-square poor (LMMSE) criterion) interpolating method has good performance, is widely used.When being used for inserting in the frequency domain, inserting in the Wiener and accomplish channel estimating at frequency domain, can be expressed as with the form of filter

${\hat{H}}_{\mathrm{LMMSE},k}=w_{\mathrm{LMMSE}}{H}_{\mathrm{Raw}}$ Formula (2)

H in the formula (2) _RawBe initial RS position channel estimation results sequence, can pass through following formulate:

$H_{\mathrm{Raw}}\left(k\right)=\left\{\begin{array}{cc}{\hat{H}}_{\mathrm{RS}}\left(k\right),& \mathrm{ForRS}\\ 0,& \mathrm{Others}\end{array}\right.$ Formula (3)

And the Wiener filter factor can be represented through following formula:

w _LMMSE=(R _Kk+ σ ²I) ^-1r _KkFormula (4)

σ in the formula (4) ²Be the noise variance on the subcarrier, I is a unit matrix, R _KkBe the frequency domain autocorrelation matrix of channel, r _KkFrequency domain auto-correlation vector for channel.

Because the receiver that adopts usually in the reality is difficult to accurately obtain channel statistic property; And; Owing to need consider the complexity issue of receiver simultaneously; Therefore, in present system, can suppose channel model usually, and come shortcut calculation based on the channel model calculating fixed filters coefficient of this supposition.

Usually, can establish channel power delay the spectrum (Power Delay Profile abbreviates PDP as) be that a length is N _CPRectangular function, N wherein _CPRepresent CP length, it is a sinc function frequency domain, and at this moment, the correlation function on the frequency domain can pass through following formulate:

$\mathrm{Corr}\left(k\right)=\mathrm{Sin}c\left(\frac{{\mathrm{KN}}_{\mathrm{CP}}}{N_{\mathrm{FFT}}}\right)$ Formula (5)

Sinc () in the formula (5) is a normalization sinc function, N _FFTBe FFT length, and the calculating of element can be carried out in correlation matrix and the vector through following formula:

R _Kk(k1, k2)=corr (k1-k2), 0≤k1, k2≤N _FFT-1 formula (6)

r _Kk(k)=and corr (k), 0≤k≤N _FFT-1 formula (7)

Wherein, R _Kk(k1, k2) representing matrix R _KkThe capable k2 column element of k1, r _Kk(k) represent k vectorial element, finally obtain the result of channel estimating.

In aforesaid way; Though adopting fixing wiener filter coefficients estimates effectively to simplify the process of channel estimating and the complexity of receiver; But owing to can not consider actual channel situation based on the channel estimating of fixed filters coefficient; Therefore when using, the higher signal to noise ratio and the error rate can be occurred, thereby accuracy of channel estimation can be reduced.

Problem to causing channel estimating performance to reduce owing to the variation of adopting fixing filter coefficient to make channel estimating can not adapt to actual environment in the correlation technique does not propose effective solution at present as yet.

Summary of the invention

To the problem that causes channel estimating performance to reduce owing to the variation of adopting fixing filter coefficient to make channel estimating can not adapt to actual environment in the correlation technique; The present invention proposes a kind of implementation of channel estimating; Can combine actual channel conditions to carry out channel estimating, improve accuracy of channel estimation.

According to an aspect of the present invention, a kind of implementation method of channel estimating is provided, has been used for realizing channel estimating based on the Wiener filtering technology.

Implementation method according to channel estimating of the present invention comprises: the channel estimation results of the reference signal of extracting is transformed into time domain, obtains the time domain initial channel estimation sequence of reference signal; Time domain initial channel estimation sequence is carried out weighting, obtain the time domain channel shock response sequence of reference signal; The power that obtains reference signal according to time domain channel shock response sequence is delayed spectrum, and delays composing the frequency domain correlation function that obtains reference signal according to power; The frequency domain correlation function is carried out normalization handle, and carry out Wiener filtering according to normalized frequency domain correlation function and calculate, obtain final channel estimation results.

Wherein, can the channel estimation results of the reference signal of extracting be transformed into time domain according to following formula:

$h_{\mathrm{raw}}\left(n\right)=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{k=0}^{N_{\mathrm{FFT}}-1}{H}_{\mathrm{raw}}\left(k\right)\&CenterDot;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,$ 0≤n≤N _FFT-1，

Wherein, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length.

In addition, can carry out weighting to time domain initial channel estimation sequence according to following formula:

h _CIR(n)＝h _raw(n)·c(n)，0≤n≤N _FFT-1，

Wherein, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length, C (n) equals N for length _FFTWindow function.

Preferably, the kind of window function can comprise one of following: rectangular window function, hamming window function, Haining window function, triangle window function.

In addition, can delay spectrum according to the power that following formula obtains reference signal:

P _PDP(n)＝h _CIR(n)·conj(h _CIR(n))，0≤n≤N _FFT-1，

Wherein, P _PDP(n) be the energy value at n time-domain sampling point place, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, conj is a conjugate operation, N _FFTBe Fourier transform length.

In addition, the frequency domain correlation function of reference signal can for:

${\mathrm{corr}}^{\&prime;}\left(k\right)=\mathrm{real}[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1，

Wherein, corr ' is the frequency domain correlation function of reference signal (k), P _PDP(n) be the energy value at n time-domain sampling point place, N _FFTBe Fourier transform length, real calculates for getting real part.

In addition, normalized frequency domain correlation function can for:

corr _new(k)＝corr′(k)/corr′(0)，0≤k≤N _FFT-1，

Wherein, corr _New(k) be normalized frequency domain correlation function.

According to a further aspect in the invention, a kind of implement device of channel estimating is provided, has been used for realizing channel estimating based on the Wiener filtering technology.

Implement device according to channel estimating of the present invention comprises: modular converter, be used for the channel estimation results of the reference signal of extracting is transformed into time domain, and obtain the time domain initial channel estimation sequence of reference signal; Weighting block is used for time domain initial channel estimation sequence is carried out weighting, obtains the time domain channel shock response sequence of reference signal; Determination module is used for confirming that according to time domain channel shock response sequence the power of reference signal delays spectrum, and is used for delaying composing the frequency domain correlation function of confirming reference signal according to power; Channel estimation module is used for that the frequency domain correlation function is carried out normalization and handles, and carries out Wiener filtering according to normalized frequency domain correlation function and calculate, and obtains final channel estimation results.

Wherein, modular converter can be used for according to following formula the channel estimation results of the reference signal of extracting being transformed into time domain:

$h_{\mathrm{raw}}\left(n\right)=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{k=0}^{N_{\mathrm{FFT}}-1}{H}_{\mathrm{raw}}\left(k\right)\&CenterDot;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,$ 0≤n≤N _FFT-1，

Wherein, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length.

In addition, weighting block can be used for according to following formula time domain initial channel estimation sequence being carried out weighting:

h _CIR(n)＝h _raw(n)·c(n)，0≤n≤N _FFT-1，

Wherein, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length, C (n) equals N for length _FFTWindow function.

In addition, the determination module power that can be used for confirming to obtain according to following formula reference signal is delayed spectrum:

P _PDP(n)＝h _CIR(n)·conj(h _CIR(n))，0≤n≤N _FFT-1，

Wherein, P _PDP(n) be the energy value at n time-domain sampling point place, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, conj is a conjugate operation, N _FFTBe Fourier transform length.

And, the frequency domain correlation function of the reference signal that determination module obtains can for:

${\mathrm{corr}}^{\&prime;}\left(k\right)=\mathrm{real}[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1，

Wherein, corr ' is the frequency domain correlation function of reference signal (k), P _PDP(n) be the energy value at n time-domain sampling point place, N _FFTBe Fourier transform length, real calculates for getting real part.

In addition, the normalized frequency domain correlation function that obtains of channel estimation module can for:

corr _new(k)＝corr′(k)/corr′(0)，0≤k≤N _FFT-1，

Wherein, corr _New(k) be normalized frequency domain correlation function.

The present invention is through obtaining corresponding Wiener filter factor according to the actual state of RS; Make channel estimation process to carry out based on RS better; Avoided that the Wiener filter factor causes channel estimation process can not adapt to the problem of the variation of actual environment owing to adopt fixedly, can effectively improve performance for estimating channel and accuracy.

Description of drawings

Fig. 1 is according to the mapping sketch map of RS in the regular circulation prefix under the single transmit antenna situation of correlation technique;

Fig. 2 is the flow chart according to the implementation method of the channel estimating of the inventive method embodiment;

Fig. 3 is the principle schematic when carrying out channel estimating according to the implementation method of the channel estimating of the inventive method embodiment;

Fig. 4 is the mean square error emulation sketch map that adopts the channel estimation methods of fixed filters coefficient in the correlation technique;

Fig. 5 is the mean square error emulation sketch map according to the channel estimating implementation method of the inventive method embodiment;

Fig. 6 is the error rate emulation sketch map that adopts the channel estimation methods of fixed filters coefficient in the correlation technique;

Fig. 7 is the error rate emulation sketch map according to the channel estimating implementation method of the inventive method embodiment;

Fig. 8 is the block diagram according to the implement device of the channel estimating of apparatus of the present invention embodiment.

Embodiment

To the problem that causes channel estimating performance to reduce owing to the variation of adopting fixing filter coefficient to make channel estimating can not adapt to actual environment in the correlation technique; The present invention proposes the channel estimation results of RS is transformed into time domain; Afterwards the sequence after the conversion is carried out weighting and obtain time domain channel shock response sequence; The power that obtains RS is afterwards delayed spectrum, and delays composing the frequency domain correlation function that obtains RS according to power, based on this frequency domain correlation function carry out inserting in the time domain with frequency domain in insert; And then obtain final channel estimation results; Make frequency domain correlation function and filter coefficient be associated, thereby make channel estimating can adapt to different situation, effectively improve performance for estimating channel and accuracy with actual RS.

To combine accompanying drawing below, describe embodiments of the invention in detail.

Method embodiment

In the present embodiment, a kind of implementation method of channel estimating is provided, has been used for realizing channel estimating, can be applied to adopt the plurality of communication systems of OFDM technology based on the Wiener filtering technology, for example, the LTE system.

To combine the processing procedure of Fig. 2 and Fig. 3 detailed description below according to the implementation method of the channel estimating of present embodiment.

Fig. 2 is the flow chart according to the implementation method of the channel estimating of the inventive method embodiment.Fig. 3 is the principle schematic when carrying out channel estimating according to the implementation method of the channel estimating of the inventive method embodiment.

When carrying out channel estimating; At first need extract frequency domain RS (RS as shown in Figure 3 extracts); RS to extracting carries out channel estimating (that is, the channel estimating among Fig. 3), obtains the frequency domain channel estimated result; This frequency domain channel estimated result only is a result who tentatively obtains, and need use this preliminary result follow-up when obtaining final channel estimation results;

Afterwards, extract the channel estimation results of RS, particularly, the channel frequency domain estimated result of RS can be used H _Raw(k) expression, wherein, k is a subcarrier sequence number span, and 0≤k≤N _FFT-1, N _FFTBe FFT length.Use

Expression RS position channel estimation results, then H _Raw(k) can pass through following formulate:

As shown in Figure 2, comprise according to the implementation method of the channel estimating of present embodiment:

Step S202 is transformed into time domain with the channel estimation results of the RS that extracts, obtains the time domain initial channel estimation sequence of RS;

Step S204 carries out weighting to time domain initial channel estimation sequence, obtains the time domain channel shock response sequence of RS;

Step S206, the power that obtains RS according to time domain channel shock response sequence is delayed spectrum, and delays composing the frequency domain correlation function (step S202 is equivalent to the processing of the calculating frequency domain correlation function among Fig. 3 to step S206) that obtains RS according to power;

Step S208 carries out normalization to the frequency domain correlation function and handles, and carries out Wiener filtering according to normalized frequency domain correlation function and calculate, and obtains final channel estimation results.

Particularly, in step S208, need calculate the Wiener filter factor according to normalized frequency domain correlation function, in conjunction with before the H that obtains _RAWCarry out inserting in the Wiener shown in Fig. 3, carry out afterwards inserting in the time domain shown in Fig. 3, obtain the channel estimation results of RS code element on the sub-carrier positions of all uses, that is, obtain the channel estimating that all use sub-carrier positions through inserting in the time domain.

Through above-mentioned processing; Can obtain corresponding Wiener filter factor according to the actual state of RS; Make channel estimation process to carry out based on RS better; Avoided that the Wiener filter factor causes channel estimation process can not adapt to the problem of the variation of actual environment owing to adopt fixedly, can effectively improve performance for estimating channel and accuracy.

To describe the processing in above each step below in detail; It should be noted that; Hereinafter cited formula only is concrete instance; And be not used in qualification the present invention, and in the middle of practical application, can as required following formula being out of shape and revising, these all should be included in the scope of the present invention.

Wherein, in step S202, can the channel estimation results of the RS that extract be transformed into time domain according to following formula:

$h_{\mathrm{Raw}}\left(n\right)=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{k=0}^{N_{\mathrm{FFT}}-1}{H}_{\mathrm{Raw}}\left(k\right)\&CenterDot;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,$ 0≤n≤N _FFT-1, formula (8)

Wherein, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length (for example, N _FFT=1024).

And, can carry out weighting to time domain initial channel estimation sequence according to following formula:

h _CIR(n)=h _Raw(n) c (n), 0≤n≤N _FFT-1, formula (9)

Wherein, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length, C (n) equals N for length _FFTWindow function.

Alternatively, the window function that is adopted here can comprise a variety of, for example, can comprise rectangular window function commonly used, hamming window function, Haining window function, triangle window function etc., and this paper does not enumerate one by one.

In step S206, can delay spectrum (that is the energy of each sample point size) according to the power that following formula obtains RS:

P _PDP(n)=h _CIR(n) conj (h _CIR(n)), 0≤n≤N _FFT-1, formula (10)

Wherein, P _PDP(n) be the energy value at n time-domain sampling point place, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, conj is a conjugate operation, N _FFTBe Fourier transform length.

And in step S206, the frequency domain correlation function that obtains RS can be represented through following formula (11):

${\mathrm{corr}}^{\&prime;}\left(k\right)=\mathrm{real}[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1，

Formula (11),

Wherein, corr ' is the frequency domain correlation function of RS (k), P _PDP(n) be the energy value at n time-domain sampling point place, N _FFTBe Fourier transform length, real calculates for getting real part.

Particularly, in step S208, normalized frequency domain correlation function specifically can for:

Corr _New(k)=corr ' (k)/corr ' (0), 0≤k≤N _FFT-1, formula (12)

Wherein, corr _New(k) be normalized frequency domain correlation function.

Particularly, obtaining normalized frequency domain correlation function corr _NewAfter; Can utilize this normalized function to replace the corr function calculation Wiener filter factor in formula formula (6) and the formula (7); And, inserts frequency domain in carrying out Wiener; All use the channel estimation results on the sub-carrier positions to obtain the RS code element, through inserting the use sub-carrier channels estimated result that (for example, adopting linear interpolation method) do not contained the RS code element in the time domain.

It should be noted that; Although in above description, enumerated a plurality of formula; But, with channel estimation results from frequency domain be transformed into time domain, to time domain initial channel estimation sequence carry out weighting, obtain that power is delayed composing, the expression of the expression of frequency domain auto-correlation function, normalized frequency domain auto-correlation function and not only be confined to the represented content of above formula.In practical application, can change above-mentioned formula as required, for example, when the expression formula distortion that makes the frequency domain correlation function, can the real computing be omitted, the expression formula of the frequency domain correlation function after the distortion that obtains is:

${\mathrm{corr}}^{\&prime;}\left(k\right)=[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1

Can be out of shape equally for other formula, concrete mapping mode will become apparent to those skilled in the art that this paper enumerates no longer one by one.

To combine instantiation to describe detailed process below according to channel estimating implementation method of the present invention.

In the instance that will describe below; Suppose that emission mode is the single antenna 10MHz bandwidth mode of 3GPP LTE physical layer 3GPP TS36.211 V8.5.0 standard [1]; Circulating prefix-length is conventional length; The energy of the reference pilot on each subcarrier and general data signal energy are all carried out normalization and are handled modulation system employing 16QAM.

Can equal N for length time domain initial channel estimation sequence being added temporary employed weighting function c (n) _CPThe rectangular window function, expression formula is following:

In this example, suppose sub-district ID number be 485, then receiving terminal uses the LS algorithm at 100 pilot frequency locations k=5,11 ... 299,729,735 ..., 1023 carry out the LS channel estimating, to other position data zero setting, suc as formula formula (1).

Just can try to achieve normalized frequency domain correlation function corr through formula (8) to formula (12) _New

Preferably, in order to reduce the complexity of estimation procedure, the channel estimating of a data subcarrier is only used 8 nearest RS, that is, and and the R in formula (6) and the formula (7) _KkAnd r _KkDimension be respectively 8 * 8 and 8 * 1, and for the RS subcarrier, then use self and 6 nearest RS channel estimation results, the R of this moment _KkAnd r _KkDimension be respectively 7 * 7 and 7 * 1.Utilizing the formula formula (4) can be in the hope of the Wiener filter factor, just can be to utilize this coefficient to carry out inserting in the Wiener filtering then, and all that obtain the RS code element are used sub-carrier positions channel estimation results.At last, use linear interpolation to obtain the channel estimating of all code elements in time domain.

Fig. 4 is carrier A (the Extended VehicularA of interpolating method channel, expansion in single footpath of fixedly Wiener filter factor; EVA) and the expansion typical urban (Extended Typical Urban; Abbreviate ETU as) channel estimating mean square error (mean square error abbreviates MSE as) performance simulation result under the channel [2]; Fig. 5 is carrier A (the Extended Vehicular A that the present invention proposes channel estimating implementation method channel, expansion in single footpath; EVA) and the expansion typical urban (Extended Typical Urban; Abbreviate ETU as) channel estimating mean square error (mean square error under the channel [2]; Abbreviate MSE as) the performance simulation result, wherein, the maximum doppler frequency of EVA and ETU channel is respectively 70Hz and 300Hz.In simulation process, the weighting function c (n) that adopts according to channel estimating implementation method of the present invention gets the rectangular window function that length is 72 samplings.

Through contrasting visible to Fig. 4 and simulation result shown in Figure 5; Under the EVA channel; Though this moment, fixedly a channel mistake problem of Wiener filtering interpolating method was not very serious relatively, channel estimating implementation method according to the present invention still is superior to adopting in the correlation technique the fixedly interpolating method of Wiener coefficient; And under single footpath channel and ETU channel, channel estimating implementation method according to the present invention obviously is superior to adopting in the correlation technique the fixedly interpolating method of Wiener filtering.

Fig. 6 is carrier A (the Extended Vehicular A of interpolating method channel, expansion in single footpath of fixedly Wiener filter factor; EVA) and the expansion typical urban (Extended Typical Urban; Abbreviate ETU as) the performance simulation result of the reception error rate (Bit Error Rate abbreviates BER as) under the channel [2]; Fig. 7 is carrier A (the Extended Vehicular A that the present invention proposes channel estimating implementation method channel, expansion in single footpath; EVA) and the expansion typical urban (Extended Typical Urban; Abbreviate ETU as) the performance simulation result of the reception error rate (Bit Error Rate abbreviates BER as) under the channel [2].

Like Fig. 6 and shown in Figure 7; Wherein curve trend and the simulation result shown in Fig. 4 and Fig. 5 are similar; Under the EVA channel; But slightly be superior to adopting the fixedly interpolating method of Wiener coefficient in the correlation technique according to the BER performance of channel estimating implementation method of the present invention, and under single footpath channel and ETU channel, obviously be superior in the correlation technique the fixedly interpolating method of Wiener filtering of employing according to the BER performance of channel estimating implementation method of the present invention.

Simulation curve under the extended cyclic prefix situation and the class of a curve shown in Fig. 4 to Fig. 7 seemingly, this paper no longer enumerates.

Through above-mentioned processing, can confirm the Wiener filter factor again according to the situation of RS itself, thereby make channel estimating can adapt to various environment, effectively improve accuracy of channel estimation and performance.

Device embodiment

In the present embodiment, a kind of implement device of channel estimating is provided, has been used for realizing channel estimating, can be applied to adopt the plurality of communication systems of OFDM technology based on the Wiener filtering technology, for example, the LTE system.

As shown in Figure 8, comprise according to the implement device of the channel estimating of present embodiment:

Modular converter 1 is used for the channel estimation results of the RS that extracts is transformed into time domain, obtains the time domain initial channel estimation sequence of RS;

Weighting block 2 is connected to modular converter 1, is used for time domain initial channel estimation sequence is carried out weighting, obtains the time domain channel shock response sequence of RS;

Determination module 3 is connected to weighting block 2, is used for confirming that according to time domain channel shock response sequence the power of RS delays spectrum, and is used for delaying composing the frequency domain correlation function of confirming RS according to power;

Channel estimation module 4 is connected to determination module 3, is used for that the frequency domain correlation function is carried out normalization and handles, and carry out Wiener filtering according to normalized frequency domain correlation function and calculate, and obtains final channel estimation results.

Wherein, modular converter 1 can be used for according to following formula the channel estimation results of the RS that extracts being transformed into time domain:

$h_{\mathrm{raw}}\left(n\right)=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{k=0}^{N_{\mathrm{FFT}}-1}{H}_{\mathrm{raw}}\left(k\right)\&CenterDot;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,$ 0≤n≤N _FFT-1，

Wherein, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length.

And weighting block 2 can be used for according to following formula time domain initial channel estimation sequence being carried out weighting: h _CIR(n)=h _Raw(n) c (n), 0≤n≤N _FFT-1,

Wherein, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, h _Raw(n) be time domain initial channel estimation sequence, N _FFTBe Fourier transform length, C (n) equals N for length _FFTWindow function.

In addition, determination module 3 power that can be used for confirming to obtain according to following formula RS is delayed spectrum:

P _PDP(n)＝h _CIR(n)·conj(h _CIR(n))，0≤n≤N _FFT-1，

Wherein, P _PDP(n) be the energy value at n time-domain sampling point place, h _CIR(n) time domain channel shock response sequence for obtaining after the weighting, conj is a conjugate operation, N _FFTBe Fourier transform length.

And determination module 3 also can be used for obtaining according to following formula the frequency domain correlation function of RS:

${\mathrm{corr}}^{\&prime;}\left(k\right)=\mathrm{real}[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2009102412202E00031.png,H2009102412202E00051.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1，

Wherein, corr ' is the frequency domain correlation function of RS (k), P _PDP(n) be the energy value at n time-domain sampling point place, N _FFTBe Fourier transform length, real calculates for getting real part.

The normalized frequency domain correlation function that is obtained by channel estimation module can be expressed as:

corr _new(k)＝corr′(k)/corr′(0)，0≤k≤N _FFT-1，

Wherein, corr _New(k) be normalized frequency domain correlation function.

Through said apparatus, can confirm the Wiener filter factor again according to the situation of RS itself, thereby make channel estimating can adapt to various environment, effectively improve accuracy of channel estimation and performance.

Describe before the device shown in Figure 8 processing shown in equally can execution graph 2 and 3, and obtain similar Fig. 4, its principle and concrete processing procedure, no longer repeat here to channel estimating simulation result shown in Figure 7.

In sum; By means of technical scheme of the present invention; Obtain corresponding Wiener filter factor through actual state according to RS; Make channel estimation process to carry out based on RS better, avoided that the Wiener filter factor causes channel estimation process can not adapt to the problem of the variation of actual environment owing to adopt fixedly, can effectively improve performance for estimating channel and accuracy.

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

Claims (13)

1. the implementation method of a channel estimating is used for realizing channel estimating based on the Wiener filtering technology, it is characterized in that said method comprises:

The channel estimation results of the reference signal of extracting is transformed into time domain, obtains the time domain initial channel estimation sequence of said reference signal;

Said time domain initial channel estimation sequence is carried out weighting, obtain the time domain channel shock response sequence of said reference signal;

The power that obtains said reference signal according to said time domain channel shock response sequence is delayed spectrum, and delays composing the frequency domain correlation function that obtains said reference signal according to said power;

Said frequency domain correlation function is carried out normalization handle, and carry out Wiener filtering according to normalized said frequency domain correlation function and calculate, obtain final channel estimation results.

2. method according to claim 1 is characterized in that, the channel estimation results of the said reference signal that will extract according to following formula is transformed into time domain:

$h_{\mathrm{raw}}\left(n\right)=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{k=0}^{N_{\mathrm{FFT}}-1}{H}_{\mathrm{raw}}\left(k\right)\&CenterDot;e^{j2\mathrm{\&pi;nk}/N_{\mathrm{FFT}},}$ 0≤n≤N _FFT-1，

Wherein, h _Raw(n) be said time domain initial channel estimation sequence, H _Raw(k) be the channel frequency domain estimated result of reference signal, k is the subcarrier sequence number, N _FFTBe Fourier transform length.

3. method according to claim 1 is characterized in that, according to following formula said time domain initial channel estimation sequence is carried out weighting:

h _CIR(n)＝h _raw(n)·c(n)，0≤n≤N _FFT-1，

Wherein, h _CIR(n) the said time domain channel shock response sequence for obtaining after the weighting, h _Raw(n) be said time domain initial channel estimation sequence, N _FFTBe Fourier transform length, c (n) equals N for length _FFTWindow function.

4. method according to claim 3 is characterized in that, it is one of following that the kind of said window function comprises: rectangular window function, hamming window function, Haining window function, triangle window function.

5. method according to claim 1 is characterized in that, the power that obtains said reference signal according to following formula is delayed spectrum:

P _PDP(n)＝h _CIR(n)·conj(h _CIR(n))，0≤n≤N _FFT-1，

Wherein, P _PDP(n) be the energy value at n time-domain sampling point place, h _CIR(n) the said time domain channel shock response sequence for obtaining after the weighting, conj is a conjugate operation, N _FFTBe Fourier transform length.

6. method according to claim 1 is characterized in that, the frequency domain correlation function of said reference signal is:

${\mathrm{corr}}^{\&prime;}\left(k\right)=\mathrm{real}[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-j2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1，

Wherein, corr ' is the frequency domain correlation function of said reference signal (k), and k is the subcarrier sequence number, P _PDP(n) be the energy value at n time-domain sampling point place, N _FFTBe Fourier transform length, real calculates for getting real part.

7. method according to claim 6 is characterized in that, normalized said frequency domain correlation function is:

corr _new(k)＝corr′(k)/corr′(0)，0≤k≤N _FFT-1，

Wherein, k is the subcarrier sequence number, corr _New(k) be normalized said frequency domain correlation function.

8. the implement device of a channel estimating is used for realizing channel estimating based on the Wiener filtering technology, it is characterized in that said device comprises:

Modular converter is used for the channel estimation results of the reference signal of extracting is transformed into time domain, obtains the time domain initial channel estimation sequence of said reference signal;

Weighting block is used for said time domain initial channel estimation sequence is carried out weighting, obtains the time domain channel shock response sequence of said reference signal;

Determination module is used for confirming that according to said time domain channel shock response sequence the power of said reference signal delays spectrum, and is used for delaying composing the frequency domain correlation function of confirming said reference signal according to said power;

Channel estimation module is used for that said frequency domain correlation function is carried out normalization and handles, and carries out Wiener filtering according to normalized said frequency domain correlation function and calculate, and obtains final channel estimation results.

9. device according to claim 8 is characterized in that, the channel estimation results of the said reference signal that said modular converter is used for will extracting according to following formula is transformed into time domain:

$h_{\mathrm{raw}}\left(n\right)=\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{k=0}^{N_{\mathrm{FFT}}-1}{H}_{\mathrm{raw}}\left(k\right)\&CenterDot;e^{j2\mathrm{\&pi;nk}/N_{\mathrm{FFT}},}$ 0≤n≤N _FFT-1，

Wherein, h _Raw(n) be said time domain initial channel estimation sequence, H _Raw(k) be the channel frequency domain estimated result of reference signal, k is the subcarrier sequence number, N _FFTBe Fourier transform length.

10. device according to claim 8 is characterized in that, said weighting block is used for according to following formula said time domain initial channel estimation sequence being carried out weighting:

h _CIR(n)＝h _raw(n)·c(n)，0≤n≤N _FFT-1，

Wherein, h _CIR(n) the said time domain channel shock response sequence for obtaining after the weighting, h _Raw(n) be said time domain initial channel estimation sequence, N _FFTBe Fourier transform length, c (n) equals N for length _FFTWindow function.

11. device according to claim 8 is characterized in that, said determination module is used for confirming to obtain according to following formula the power of said reference signal and delays spectrum:

P _PDP(n)＝h _CIR(n)·conj(h _CIR(n))，0≤n≤N _FFT-1，

Wherein, P _PDP(n) be the energy value at n time-domain sampling point place, h _CIR(n) the said time domain channel shock response sequence for obtaining after the weighting, conj is a conjugate operation, N _FFTBe Fourier transform length.

12. device according to claim 8 is characterized in that, the frequency domain correlation function of the said reference signal that said determination module obtains is:

${\mathrm{corr}}^{\&prime;}\left(k\right)=\mathrm{real}[\frac{1}{\sqrt{N_{\mathrm{FFT}}}}{\mathrm{\&Sigma;}}_{n=0}^{N_{\mathrm{FFT}}-1}{P}_{\mathrm{PDP}}\left(n\right)\&CenterDot;e^{-j2\mathrm{\&pi;nk}/N_{\mathrm{FFT}}}],$ 0≤k≤N _FFT-1，

Wherein, corr ' is the frequency domain correlation function of said reference signal (k), and k is the subcarrier sequence number, P _PDP(n) be the energy value at n time-domain sampling point place, N _FFTBe Fourier transform length, real calculates for getting real part.

13. device according to claim 12 is characterized in that, the normalized said frequency domain correlation function that said channel estimation module obtains is:

corr _new(k)＝corr′(k)/corr′(0)，0≤k≤N _FFT-1，

Wherein, corr _New(k) be normalized said frequency domain correlation function.

Images