WO2010118588A1

WO2010118588A1 - Channel estimation method and device of orthogonal frequency division multiplexing system

Info

Publication number: WO2010118588A1
Application number: PCT/CN2009/072451
Authority: WO
Inventors: 许百成; 牟秀红; 冯心睿
Original assignee: 北京天碁科技有限公司
Priority date: 2009-04-14
Filing date: 2009-06-25
Publication date: 2010-10-21
Also published as: CN101534266A; CN101534266B

Abstract

A channel estimation method and device of orthogonal frequency division multiplexing system are provided. The method includes the following steps: A. performing a calculation according to the frequency-domain channel estimations of the reference signals, and obtaining the delay spread of the channel(201); B. generating frequency-domain interpolation coefficients according to the delay spread(202); C. performing the interpolation disposal according to the frequency-domain interpolation coefficients and the frequency-domain channel estimations of the reference signals, and obtaining the whole frequency-domain channel estimations on the time-domain position of the reference signals(203); D. performing a calculation according to the frequency-domain channel estimations at different time-domain positions, and obtaining the coherent time of the channel(204); E. generating the time-domain interpolation coefficients according to the coherent time(205); F. performing the interpolation disposal according to the time-domain interpolation coefficients and the whole frequency-domain channel estimations in the range of coherent time, and obtaining the whole time-domain and frequency-domain channel estimations(206). According to the implementation of the present invention, the result of channel estimation can adapt itself to the change of the channel.

Description

Channel estimation method and device for orthogonal frequency division multiplexing system

Technical field

The invention belongs to the field of wireless communication technologies, and particularly relates to an orthogonal frequency division multiplexing

Channel estimation method and apparatus for (OFDM) systems. Background of the invention

In Long Term Evolution (LTE) wireless communication systems, OFDM is its core physical layer technology. It modulates the data stream on multiple orthogonal subcarriers, and the spectrum between orthogonal subcarriers can be overlapped, which greatly improves spectrum utilization.

Referring to FIG. 1, in an orthogonal frequency division multiplexing system, information transmitted may be described by a resource grid, and the resource grid represents resources in the entire time domain and frequency domain in the form of resource elements (RE, Resource Element). That is, a minimum square in Figure 1, which corresponds to the length of time of one subcarrier on the frequency domain and one symbol on the time domain. All information that needs to be transmitted is carried by the resource element. Multiple resource elements form a resource block. Specifically, in the case of a normal cyclic prefix, 12 (number of subcarriers) x 7 (number of symbols) constitutes one resource block; in the case of extended cyclic prefix, 12 (number of subcarriers) χ 6 (number of symbols) constitutes one Resource block.

In order to correctly demodulate the data, it is necessary to know the channel condition at each RE position. In the LTE system, the channel estimation at each RE location is obtained by inserting a known reference signal RS ( Reference Signal ) at a specific resource element position of each resource block to calculate a channel at the RE location. Estimate, then use the interpolation method to get the channel estimates at all other resource element locations.

In actual application scenarios, the channel conditions vary widely, but the prior art uses a fixed interpolation method to cope with all channel conditions, which results in lower accuracy of channel estimation. To demodulate the effect of the data. Summary of the invention

The technical problem to be solved by the present invention is to provide a channel estimation method and apparatus for an orthogonal frequency division multiplexing system, so that the channel estimation result can be adaptively changed as the channel changes.

In order to solve the above technical problem, the present invention provides the following technical solutions:

A channel estimation method for an orthogonal frequency division multiplexing system includes the following steps:

A, calculating according to the frequency domain channel estimation of the reference signal, and obtaining a delay spread of the channel;

B. Generate a frequency domain interpolation coefficient according to the delay extension;

C. Perform interpolation processing according to the frequency domain interpolation coefficient and the frequency domain channel estimation of the reference signal, to obtain a channel estimation of the entire frequency domain at the time domain position where the reference signal is located.

The above method, wherein, after the step C, further comprises:

D. Perform calculation according to frequency domain channel estimation at different time domain locations to obtain a coherence time of the channel;

E. generating a time domain interpolation coefficient according to the coherence time;

F. Perform interpolation processing according to the time domain interpolation coefficient and the channel estimation of the entire frequency domain in the coherent time range, to obtain channel estimation of the entire time domain and the frequency domain.

The above method, wherein the step A includes:

A1: Performing an inverse Fourier transform on the frequency domain channel estimation of the reference signal to obtain a time domain channel estimation of the reference signal;

A2. Obtain an effective path of the time domain channel estimation.

A3. Perform calculation according to the effective path to obtain a delay spread of the channel.

The above method, wherein the step A2 comprises:

Denoising the time domain channel estimation; Finding a channel window in a time domain channel estimation after denoising processing;

Obtaining an effective path according to the channel window.

The above method, wherein the step A2 comprises:

Finding a channel window in the time domain channel estimation;

Performing a denoising process on the channel window;

The effective path is obtained according to the channel window after the denoising process.

The above method, wherein the delay spread is:

Maximum delay spread, average delay spread, or rms delay spread.

The above method, wherein after the step A3, the method further includes:

A4. Smoothing the delay spread of the channel.

The above method, wherein the smoothing process performed in the step A4 is: wherein, „the current value is extended for the smoothed delay, f„ is the undelayed delay extended current value, ^ is smoothed The processed delay extended history value is the forgetting factor, 0 < β ≤ 1.

The above method, wherein the step B includes:

Using the delay spread as a parameter, generating the frequency domain interpolation coefficient based on a raised cosine, a root raised cosine or a sampling function, or first generating a channel correlation matrix, and then generating the frequency domain by using a minimum mean square error criterion Interpolation factor.

The above method, wherein the step C includes:

Constructing a frequency domain channel estimate S before interpolation in the time domain position of the reference signal: If there is a reference signal at the subcarrier, then if there is no reference signal at the subcarrier, then let =0, where /^ is the frequency domain of the reference signal Channel estimation, k = 0, l- N-1, N is the number of subcarriers; Frequency domain interpolation by convolution operation: H' = S*F , where the symbol "*" is a convolution operation and F is a frequency domain interpolation coefficient;

The frequency domain interpolation result H' is processed to obtain a channel estimation H: k = H _K ' _{I2 + k of the} entire frequency domain, where the length of the frequency domain interpolation coefficient is reduced by 1.

The above method, wherein the step D includes:

D1. Calculating a difference factor between a frequency domain channel estimate of the reference signal and an available frequency domain channel estimate in the time domain;

D2. Determine a coherence time of the channel according to the difference factor.

The above method, wherein the available frequency domain channel is estimated as:

The frequency domain channel estimate of the reference signal, or the channel estimate of the entire frequency domain at the time domain location of the reference signal.

The above method, after the step D2, further includes:

D3. Smoothing the coherence time of the channel.

The above method, wherein the smoothing process performed in the step D3 is

, _n = round[ _cn + (1— γ) _{£ η} _ _χ ]

Where is the current value of the smoothed coherence time, the current value of the coherent time without smoothing, ^^ is the uncoordinated coherence time history value, ^ is the forgetting factor, 0 < 1 , ^round is the nearest Whole operation.

The above method, wherein the step E includes:

The coherence time is taken as a parameter, and the time domain interpolation coefficient is generated based on a linear method or a nonlinear method.

The above method, wherein the step E includes:

The time domain interpolation coefficients are generated using a difference factor that is within the coherence time range. In the above method, in step E, the weight of the first reference signal in the coherence time range is: , where is the difference factor corresponding to the first reference signal.

A channel estimation apparatus for an Orthogonal Frequency Division Multiplexing system includes:

a delay estimator for calculating a frequency domain channel estimate of the reference signal to obtain a delay spread of the channel;

a frequency domain interpolation coefficient generator, configured to generate a frequency domain interpolation coefficient according to the time delay extension; a frequency domain interpolator, configured to perform interpolation processing according to the frequency domain interpolation coefficient and the frequency domain channel estimation of the reference signal, to obtain The channel estimate of the entire frequency domain at the time domain location where the reference signal is located.

The above device, wherein:

a coherence time estimator, configured to calculate according to frequency domain channel estimation at different time domain locations, to obtain a coherence time of the channel;

a time domain interpolation coefficient generator, configured to generate a time domain interpolation coefficient according to the coherence time; a time domain interpolator, configured to perform interpolation processing according to the time domain interpolation coefficient and channel estimation of the entire frequency domain in a coherent time range , obtain channel estimates for the entire time domain and frequency domain.

Compared with the prior art, the embodiment of the present invention generates frequency domain interpolation coefficients by estimating channel delay spread and using it as an important parameter, such that the frequency domain interpolation coefficients are adaptively changed as the channel changes. Further, the coherence time of the channel is estimated, and the estimation result is used to generate the time domain interpolation coefficient, so that the time domain interpolation coefficient also adaptively changes as the channel changes. BRIEF DESCRIPTION OF THE DRAWINGS

1 is a normal cyclic prefix and a distribution of reference signals in a resource block in the case of single antenna transmission; 2 is a general flowchart of a channel estimation method according to an embodiment of the present invention; FIG. 3 is a schematic structural diagram of a channel estimation apparatus according to an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of a channel estimation apparatus according to another embodiment of the present invention. Mode for carrying out the invention

The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

In the embodiment of the present invention, the frequency domain interpolation coefficient used for channel estimation is adaptively calculated according to the correspondence between the channel delay spread and the coherence bandwidth; and the time domain interpolation coefficient in the channel estimation is adaptively calculated according to the coherence time.

The so-called channel delay spread means that the path components of the channel experience different propagation paths, and therefore each path component has a different time delay, which causes the energy of the signal to be extended in time. The so-called coherent bandwidth refers to a frequency domain bandwidth in which the amplitude of the received signal has a strong correlation, that is, the amplitude-frequency response of the channel remains basically unchanged in the coherent bandwidth. The channel delay spread has a strong correspondence with the coherent bandwidth. The larger the channel delay spread, the narrower the coherence bandwidth. Conversely, the smaller the channel delay spread, the wider the coherence bandwidth. The so-called coherence time refers to a time interval in which the amplitude of the received signal has a strong correlation, that is, the channel impulse response remains substantially unchanged during the coherence time.

As described above, the amplitude of the received signal within the coherent bandwidth has a strong correlation, and therefore, the frequency domain channel estimation of the reference signal can be utilized to estimate the channel at other frequency domain locations within the relevant bandwidth and can be obtained. More accurate estimation results. At the same time, the delay spread of the channel has a strong correspondence with the coherent bandwidth. Therefore, the channel estimation in the relevant bandwidth can be converted into delay extension through the channel. Specifically, in the embodiment of the present invention, when the frequency domain interpolation coefficient is determined, the delay spread of the channel is used to obtain the channel estimation in other frequency domain locations within the relevant bandwidth by using the frequency domain channel estimation of the reference signal. Including the following steps:

Step 201

Extended extension

Step 202: Generate a frequency domain interpolation coefficient according to the delay extension.

Step 203: Perform interpolation processing according to the frequency domain interpolation coefficient and the frequency domain channel estimation of the reference signal to obtain a channel estimation of the entire frequency domain in the time domain position where the reference signal is located;

Step 204: Calculate according to the frequency domain channel estimation at different time domain locations, and obtain the coherence time of the channel.

Step 205: Generate a time domain interpolation coefficient according to the coherence time;

Step 206 performs interpolation processing according to the time domain interpolation coefficient and the channel estimation of the entire frequency domain in the coherent time range, and obtains the channel estimation of the entire time domain and the frequency domain.

It should be noted that the foregoing steps 204-206 are optional steps. After obtaining the channel estimation of the entire frequency domain at the time domain position of the reference signal, other existing methods may be used for time domain interpolation. Channel estimation for the entire time domain and frequency domain.

For a specific implementation of the above method, reference may be made to the following description of the channel estimation apparatus of the embodiment of the present invention.

Referring to FIG. 3 and FIG. 4, the channel estimation apparatus of the embodiment of the present invention mainly includes: a delay estimator, a frequency domain interpolation coefficient generator, a coherent time estimator, a time domain interpolation coefficient generator, a frequency domain interpolator, and a time domain interpolation. The following describes the working principle of each module described above in detail.

Time delay estimator

The delay estimator calculates the delay spread in the wireless channel using the frequency domain channel estimate of the reference signal. There are multiple representations of delay spread, and the most commonly used delay spread, average delay Extensions and root mean square delay extensions, etc. Regardless of which representation is used, the effective paths of the time domain channel impulse response must be determined first. The delay extension can obtain the time domain channel estimation by performing inverse domain Fourier transform on the frequency domain channel estimation of the position of the reference signal, and then obtain the effective path of the time domain channel estimation, and calculate according to the effective path to obtain the channel delay. Expansion.

Specifically, the time delay estimator includes an inverse Fourier transform unit, an effective path acquisition unit, and a delay calculation unit. The inverse Fourier transform unit is configured to perform inverse domain Fourier transform on the frequency domain channel estimation of the reference signal to obtain a time domain channel estimation of the reference signal, and an effective path obtaining unit, configured to obtain an effective path of the time domain channel estimation; And a calculating unit, configured to perform calculation according to the effective path to obtain a delay spread of the channel.

Generally, the effective path is not distributed over the entire time domain channel estimation range, but is concentrated in a certain width of the channel window. Therefore, as an optimization scheme, the power of each point in the time domain channel estimation can be calculated first, and then the width is The rectangular window of W is slidably summed over the entire power sequence to find the power and the largest window is the channel window. By finding the effective paths in the channel window, the delay spread of the channel is obtained. In the implementation, one of the following two options can be used:

(1) Denoising the time domain channel estimation first, then finding the channel window on the denoised channel estimation foundation, and finally finding the effective path in the channel window to find the delay spread of the channel.

(2) Firstly, the channel window is searched directly on the time domain channel estimation before denoising, then the channel window is denoised, and finally the effective path is found in the channel window to obtain the delay spread of the channel.

For the search of the channel window, the amount of calculation can also be reduced by reducing the channel window finding range method as follows:

(1) Find the power maximum path of the time domain channel estimation, and determine a window with width w as the center, which is larger than w but smaller than the length of the entire time domain channel estimation, and find the channel within the range of the wide window. window. (2) In wireless systems, the synchronization timing is performed in real time. This is to synchronize the channel window at a desired position. Due to the inevitable error, the channel window usually wobbles around the desired position. It is therefore also possible to center the position, determine a window of width W', W' is greater than W but less than the length of the entire time domain channel estimate, looking for a channel window over a wide window.

If the form of maximum delay spread is used, then it is only necessary to find the first effective path and the last effective path in the channel window and calculate the time difference.

If the form of average delay spread is used, then the delay spread should be calculated as follows:

In the formula, and represent the delay and power of the ith path, respectively, and M is the number of effective paths. If the form of rms delay extension is used, the delay spread should be calculated as follows

The calculation of the channel delay spread can be calculated once per time slot or once per subframe. If each time slot is calculated once, one of the two sets of reference signals can be selected for calculation, or two sets of reference signals can be separately calculated, and then the average value is taken as the final result. The group referred to herein refers to a reference signal having the same position on the time domain axis. If a plurality of time slots are calculated once, a set of reference signals in a plurality of time slots may be taken to calculate or take several sets of signals to be separately calculated, and then the average value is taken as a final result. In order to improve the stability of the delay estimation, the current delay spread estimation value and the historical value may be smoothed as an optimization scheme, that is, the delay estimator may further include: a delay smoothing unit, configured to The delay spread of the channel is smoothed, and the specific processing is: where „ is the current value extended by the smoothed delay, f „ is the unsmoothed delay extended current value, ^ is the smoothed time The extended history value is the forgetting factor, 0 < ^ ≤ 1.

Frequency domain interpolation coefficient generator

The frequency domain interpolation coefficient generator uses the delay spread of the channel as an important parameter to generate frequency domain interpolation coefficients.

There are several methods for generating frequency domain interpolation coefficients by using channel delay extension. The frequency domain interpolation coefficients can be generated directly based on the raised cosine function, the root raised cosine function or the sampling function. Alternatively, the frequency domain channel correlation matrix can be formed into a frequency domain channel correlation matrix. The frequency domain interpolation coefficients are generated according to the correlation matrix, for example, the frequency domain interpolation coefficients are generated by using the MMSE (Minimum Mean Square Error) criterion.

Here, taking the sampling function as an example to illustrate how to generate the frequency domain interpolation coefficient F according to the delay spread, as follows:

F

Here Δ/ is the subcarrier spacing, and there are two types of LTE in 15 kHz and 7.5 kHz. The length of the interpolation coefficient is + l. The Κ value is determined as follows, = 2round( ^ - ^ ) , round means the nearest rounding operation. The mapping factor for delay spread and coherence bandwidth, ie the coherence bandwidth _{β =} ^_. In the above manner, the length of the interpolation coefficient is made to satisfy the requirement of the relevant bandwidth, and the interpolation coefficients in the relevant bandwidth range are determined for subsequent frequency domain interpolation.

Coherent time estimator The coherence time estimator performs calculations based on frequency domain channel estimates at different time domain locations to obtain the coherence time of the channel. Specifically, the coherence time estimator may include: a difference factor calculation unit, configured to calculate a frequency domain channel estimation of the reference signal and a difference factor of the available frequency domain channel estimation in the time domain; a coherence time calculation unit, configured to The difference factor determines the coherence time of the channel. The so-called available frequency domain channel estimation here has different meanings for the two structures shown in Figures 3 and 4. For Figure 3, the available frequency domain channel estimate refers to the channel estimate at the location of the reference signal. For Figure 4, the coherent estimator uses the output of the frequency domain interpolator as input, such that the available frequency domain channel estimate refers to the entire frequency domain channel estimate of the symbol location at which the reference signal is located.

Coherence time estimation is calculated using frequency domain channel estimates that are temporally located at different symbol locations and can be calculated as follows:

H — H

H

Positive = max^ I Pj < a)

t ^positive ^minus

To ensure symmetry, the coherence time can also be selected as follows: I = 2 min{"bU _Uve } or

' positive *

Among them, is the difference factor with 1^, the smaller the value, the expression. And, the stronger the coherence of the channel impulse response at the two locations in the time domain; the smaller the converse. N is the number of symbols used to calculate the coherence time. Is a preset parameter for coherent decision, that is, when ?. <", it is considered that the channel impulse response at both positions ^ and ^ is coherent. It means that all or part of the available frequency domain is located at the reference position. a sequence of channel estimates, H _; represents a sequence of all or part of the available frequency domain channel estimates at the jth position relative to the reference position, j is positive The number represents the position in front of the relative reference position, and when j is negative, it indicates the position behind the relative reference position. f _£ is the coherence time, the unit is the time taken by one symbol. The frequency domain channel estimates taken by H. and H. should be in the same position in the frequency domain, ie

H _i0 = , ,

― \.

It is true that if all the available frequency domain channel estimates in the frequency domain are used to calculate the coherence time, the best results will be obtained, but the amount of computation will be large. Therefore, as a compromise, a part of the frequency domain channel estimation can also be used to calculate the coherence time. .

The calculation of the difference factor can also take other forms such as:

In order to ensure the stability of the system, you can also protect t _£ , ie,

Where is the upper threshold of the preset coherence time, t _mn is the lower threshold of the preset coherence time.

The calculation of the coherence time can be calculated once per subframe or once in multiple subframes.

In order to improve the stability of the coherent time estimation, the current coherence time estimation value and the historical value may be smoothed as an optimization scheme, that is, the coherent time estimator further includes: a coherent time smoothing unit, configured to The coherence time is smoothed, and the specific processing is: , _n = round[ _cn + (1 - γ) _{£ η} _ _χ ]

Wherein, the current value of the coherent time after smoothing is the current value of the coherent time without smoothing, ^ is the coherent time history value without smoothing, and is the forgetting factor, 0 < γ ≤ 1, round is the nearest rounding operation.

Time domain interpolation coefficient generator

The time domain interpolation coefficient generator is a coefficient that uses coherence time as an important parameter to generate time domain interpolation. In the embodiment of the present invention, the frequency domain channel estimation in the time position of each reference signal in the coherent time range is used to perform time domain interpolation on the position of the data symbol, because the channel impulse response in the coherent time range has strong correlation. Therefore, the linear method of the cartridge can be used for interpolation, or the nonlinear method can be used for interpolation. Accordingly, the time domain interpolation coefficient generator may generate time domain interpolation coefficients based on a linear method or a nonlinear method.

As one of the most compact embodiments, the time domain interpolation coefficients can be generated by an averaging method, that is, if there are M sets of reference signals in the coherence time range, then ^ can be used as the interpolation coefficients to the data symbols in the coherent time range. Interpolate.

As an optimization scheme, the difference factor ^ can also be used to generate the interpolation coefficient, that is, the smaller the ^ is, the larger the weight of the corresponding reference signal in the interpolation coefficient is, and vice versa. The weight coefficient sum of each group of reference signals in the coherence time is 1. As an embodiment, the weight on each reference signal can be assigned as follows:

Frequency domain interpolator

The frequency domain interpolator uses the frequency domain channel coefficients of the frequency domain interpolation coefficients and the reference signal to interpolate over the entire frequency domain at the time position of the reference signal.

Specifically, first, construct a frequency domain channel estimation before interpolation in the time domain position of the reference signal.

5:

- f ^ss ) , there is RS at the subcarrier position where fc is located, _{Λ 1 ΛΓ 1}

H _t = ^k k = 0,1· · · Ν - 1

k 1θ, where there is no RS in the subcarrier position, where / ^s ) is the frequency domain channel estimate of the reference signal, and N is the number of subcarriers. Then performing frequency domain interpolation by convolution, ie,

H = H *F , where the symbol " * " is a convolution operation and F is the frequency domain interpolation coefficient output by the aforementioned frequency domain interpolation coefficient generator; by using the frequency domain interpolation operation here, using any reference according to the current correlation bandwidth Frequency domain channel estimation of the signal, estimating a frequency domain channel of other frequency domain locations within a relevant bandwidth of the reference signal, thereby obtaining a full frequency domain channel estimate that matches the current channel environment;

Finally, the frequency domain interpolation result H' is processed to obtain a channel estimation H of the entire frequency domain:

^H k = H' , where K is the length of the frequency domain interpolation coefficient minus one.

Due to the initial transient effect of the convolution calculation and the tailing effect at the end, the interpolation result of some points before and after is incomplete interpolation, so special processing is needed in the actual implementation, and certain compensation or other processing is performed. method. For example, the channel estimate for this portion of the point may take the nearest value to replace the interpolation result.

Time domain interpolator

The time domain interpolator performs interpolation processing based on the time domain interpolation coefficient and the channel estimation of the entire frequency domain in the coherent time range, and obtains the channel estimation of the entire time domain and the frequency domain.

In summary, embodiments of the present invention generate frequency domain interpolation coefficients by estimating channel delay spread and using it as an important parameter such that the frequency domain interpolation coefficients are adaptively varied as the channel changes. Further, the coherence time of the channel is estimated and the estimation result is used to generate the time domain interpolation coefficient, such that the time domain interpolation coefficient also adaptively changes as the channel changes. Compared with the fixed interpolation coefficients used in the prior art, the frequency domain and time domain interpolation coefficients obtained by the channel adaptive change obtained in the embodiments of the present invention can obtain more accurate channel estimation results after being used for channel estimation. Applying the channel estimation result to demodulate the data can improve the performance of data demodulation accordingly.

Although the invention has been illustrated and described by reference to a preferred embodiment of the invention It will be apparent to those skilled in the art that various modifications may be made in the form and the details thereof without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

Claim

A channel estimation method for an Orthogonal Frequency Division Multiplexing system, comprising the steps of:

2. The method of claim 1 further comprising, after said step C, further comprising:

3. The method according to claim 1, wherein the step A comprises:

A2. Obtain an effective path of the time domain channel estimation.

The method according to claim 3, wherein the step A2 comprises: performing denoising processing on the time domain channel estimation;

Finding a channel window in a time domain channel estimation after denoising processing;

Obtaining an effective path according to the channel window.

5. The method of claim 3, the step A2 comprising: Finding a channel window in the time domain channel estimation;

Performing a denoising process on the channel window;

The method according to claim 3, wherein the delay spread is: a maximum delay spread, an average delay spread, or a root mean square delay spread.

7. The method according to claim 3, wherein after the step A3, the method further comprises:

A4. Smoothing the delay spread of the channel.

8. The method according to claim 7, characterized in that, the smoothing process performed in Step A4: wherein, ₇ "for the delay after the extension of the smoothed current value, F" without the smoothing process is The delay extends the current value, ^^ is the smoothed extended delay history value, which is the forgetting factor, 0<^≤1.

9. The method according to claim 1, wherein the step B comprises: using the delay spread as a parameter, generating the frequency domain interpolation coefficient based on a raised cosine, a root raised cosine or a sampling function;

Alternatively, a channel correlation matrix is first generated, and the minimum mean square error criterion is used to generate the frequency domain interpolation coefficients.

10. The method according to claim 1, wherein the step C comprises: constructing a frequency domain channel estimation S before interpolation in a time domain position of the reference signal: if there is a reference signal at the subcarrier, then let = / ^s ), if there is no reference signal at the subcarrier, then

H _k =0, where ^ss ) is the frequency domain channel estimate of the reference signal, k = Q, '-'N~ , W is the number of subcarriers; Frequency domain interpolation by convolution operation: H' = S*F , where the symbol "*" is a convolution operation and F is the frequency domain interpolation coefficient;

The frequency domain interpolation result H' is processed to obtain a channel estimation H: H _k = H _K ' _{ll + k of the} entire frequency domain, where the length of the frequency domain interpolation coefficient is reduced by 1.

11. The method of claim 2, wherein the step D comprises:

12. The method of claim 11, wherein the available frequency domain channel is estimated to be:

The method according to claim 11, wherein after the step D2 and before the step E, the method further comprises:

D3. Smoothing the coherence time of the channel.

The method according to claim 13, wherein the smoothing process performed in the step D3 is

t _c „ = round[ yt _cn + (1— γ)ΐ _εη _ _χ ]

Where t _£ „S is the current value of the coherent time after smoothing, the current value of the coherence time without smoothing, ^ is the coherence time history value without smoothing, y is the forgetting factor, 0< 1, roiind Round up the operation for the nearest.

The method according to claim 2, wherein the step E comprises: generating the time domain interpolation coefficient based on a linear method or a nonlinear method by using the coherence time as a parameter.

16. The method of claim 11, wherein the step E comprises: generating the time domain interpolation coefficient using a difference factor that is within the coherence time range.

17. The method according to claim 16, wherein: in step E, the weight of the first reference signal in the coherence time range is:

Where ^ is the difference factor corresponding to the first reference signal.

A channel estimation apparatus for an OFDM system, comprising: a delay estimator for performing calculation based on a frequency domain channel estimation of a reference signal to obtain a delay spread of the channel;

19. The device of claim 18, further comprising:

a time domain interpolation coefficient generator for generating a time domain interpolation coefficient according to the coherence time; a time domain interpolator for performing interpolation processing according to the time domain interpolation coefficient and a channel estimation of the entire frequency domain in a coherent time range , obtain channel estimates for the entire time domain and frequency domain.

The apparatus according to claim 18, wherein the delay estimator comprises: an inverse Fourier transform unit, configured to perform inverse frequency Fourier transform of a frequency domain channel estimation of the reference signal to obtain a time domain channel estimation of the reference signal ;

An effective path obtaining unit, configured to obtain an effective path of the time domain channel estimation; The delay calculation unit is configured to perform calculation according to the effective path to obtain a delay spread of the channel.

The device of claim 20, wherein the delay estimator further comprises:

A delay smoothing unit is configured to perform smoothing processing on the delay spread of the channel.

22. Apparatus according to claim 18 wherein:

The frequency domain interpolation coefficient generator is further configured to: use the delay spread as a parameter, generate the frequency domain interpolation coefficient based on a raised cosine, a root raised cosine or a sampling function, or first generate a channel correlation matrix, and then The frequency domain interpolation coefficients are generated using a minimum mean square error criterion.

23. The apparatus of claim 18, wherein the frequency domain interpolator is further used to:

Constructing the frequency domain channel estimate δ before interpolation in the time domain position of the reference signal: If there is a reference signal at the subcarrier, let = / ^s ), if there is no reference signal at the subcarrier, let H _k =0, where / ^s ) is the frequency domain channel estimate of the reference signal, k = Q, '-'N~ , W is the number of subcarriers;

Frequency domain interpolation by convolution operation: H' = S*F , where the symbol "*" is a convolution operation and F is a frequency domain interpolation coefficient;

The frequency domain interpolation result H' is processed to obtain a channel estimation of the entire frequency domain.

H _k =H _K ' _ll+k , where is the length of the frequency domain interpolation coefficient minus one.

24. The apparatus of claim 19, wherein the coherence time estimator comprises:

a difference factor calculation unit, configured to calculate a difference factor between a frequency domain channel estimate of the reference signal and an available frequency domain channel estimate in the time domain; A coherence time calculation unit is configured to determine a coherence time of the channel according to the difference factor.

25. The apparatus of claim 24, wherein the available frequency domain channel is estimated to be:

The device according to claim 24, wherein the coherence time estimator further comprises:

A coherent time smoothing unit is configured to smooth the coherence time of the channel.

27. Apparatus according to claim 19 wherein:

The time domain interpolation coefficient generator is further configured to generate the time domain interpolation coefficient based on a linear method or a nonlinear method by using the coherence time as a parameter.

28. Apparatus according to claim 24 wherein:

The time domain interpolation coefficient generator is further configured to generate the time domain interpolation coefficient using a difference factor within the coherence time range.