JP2007089167A - Method of channel estimation in orthogonal frequency division multiplexing system and channel estimator - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of channel estimation in an orthogonal frequency division multiplexing (OFDM) system. <P>SOLUTION: This method includes: step A which inserts a pilot signal into an OFDM symbol, estimates a channel at the position of the pilot signal in a frequency domain, and obtains a channel estimation result in the frequency domain; step B which changes the channel estimation result in the frequency domain to that in a time domain, obtains channel impulse response estimation in the time domain, and performs subspace tracking concerning an autocorrelation matrix of the channel impulse response estimation in the time domain; and step C which calculates the channel impulse response estimation based on the subspace tracking, changes the channel impulse response estimation based on the subspace tracking to that in the frequency domain, and obtains channel response estimation in the frequency domain. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a channel estimation technical field in a mobile communication system, and more particularly to a channel estimation method and a channel estimator in an orthogonal frequency division multiplexing (OFDM) system.

  Currently, along with the development of mobile communication systems, personal mobile communication terminals that can support various new services have already been provided to subscribers. Since these services require transmission of a large amount of data, a higher bit transmission rate is required for the mobile communication system. In an ordinary single carrier system, using a higher bit transmission rate makes it difficult to receive signals effectively due to intersymbol interference (ISI) and severe frequency selective fading of the radio channel. The OFDM technology has the ability to suppress ISI and can provide high frequency spectrum efficiency. Therefore, it is regarded as one of the transmission technologies most likely to be adopted in the next generation wireless mobile communication system. OFDM technology has already been widely applied in various fields such as digital subscriber lines, digital audio / video broadcasting, wireless LAN and wireless MAN.

  In order to guarantee good performance in a radio mobile channel environment of a communication system, a constantly changing multipath radio fading channel should be estimated as accurately as possible, which is particularly difficult in the case of high-speed movement. Although the quality of channel estimation has a great influence on the performance of an OFDM system, currently practical channel estimation methods generally use known pilot information.

  On the other hand, pilot-based channel estimation is divided into least square (LS) estimation and minimum mean square error (MMSE) estimation, depending on the estimation algorithm used. Usually, MMSE estimation has higher channel estimation accuracy than LS estimation. Have. On the other hand, depending on the processing process, pilot-based channel estimation is divided into frequency domain estimation and time domain estimation, and in general, time domain estimation is higher than frequency domain estimation. It has channel estimation accuracy and lower computational complexity. At the same time, the signal and noise can be effectively divided by the subspace tracking technique, thereby greatly improving the accuracy of channel estimation.

FIG. 1 is a flowchart of an MMSE channel estimation method in the time domain in a conventional OFDM system. As shown in Figure 1,
An OFDM pilot signal used for channel estimation generated by the system transmitter is inserted into an OFDM symbol, and the inserted pilot signal is temporally continuous and arranged for every certain number of subcarriers (hereinafter, comb-shaped arrangement). Step 101,
Performing step 102 LS channel estimation in the frequency domain for the pilot signal position channel;
Step 103 that preferentially adopts inverse fast Fourier transform (IFFT) and transforms the LS channel estimation result in the frequency domain in step 102 from the frequency domain to the time domain;
Performing step 104 of time domain MMSE channel estimation;
Transforming the time domain MMSE channel estimate back to the frequency domain to obtain a complete channel estimate result 105.

  However, this method does not separate the signal and noise in the channel. That is, since the channel impulse response and noise are not separated, the noise interferes with the signal, thereby significantly reducing the channel estimation accuracy. At the same time, in this method, when performing MMSE channel estimation in the time domain, it is necessary to perform an operation for obtaining an inverse matrix, so that the complexity becomes very high.

  The present invention has been made in view of the above, and a main object of the present invention is to provide a channel estimation method in an OFDM system capable of improving the accuracy of channel estimation.

  Another object of the present invention is to provide a channel estimator in an OFDM system capable of improving the accuracy of channel estimation.

  In order to achieve the above object, the technical solution of the present invention is realized as follows.

A channel estimation method in an OFDM system,
Inserting a pilot signal into an OFDM symbol, performing channel estimation in the frequency domain for a channel at the pilot signal position, and obtaining a channel estimation result in the frequency domain; and
Transform the channel estimation result in the frequency domain from the frequency domain to the time domain to obtain a time domain channel impulse response estimate, and perform subspace tracking in the time domain for the autocorrelation matrix of the time domain channel impulse response estimation. Step B to do,
Calculating a time domain channel impulse response estimate based on the subspace tracking in the time domain, and converting the time domain channel impulse response estimate into the frequency domain to obtain a frequency domain channel response estimate. .

  In step A, the pilot signal to be inserted into the OFDM symbol is to insert a comb-arranged pilot signal or a block-type pilot signal.

  In step A, frequency domain LS channel estimation is performed on the channel at the pilot signal position, or frequency domain MMSE channel estimation is performed.

  In step B, the frequency domain channel estimation result is transformed from the frequency domain to the time domain by performing an inverse fast Fourier transform on the frequency domain channel estimation result.

In step A, perform a least-squares LS channel estimation in the frequency domain for the channel at the position of the pilot signal,
In step B, subspace tracking is performed on the autocorrelation matrix of the channel impulse response estimation in the time domain, where the subspace tracking performs repeated calculation of intermediate variables using a predetermined forgetting factor, In the iterative calculation, the QR decomposition value of the intermediate variable, the phase rotation matrix, and the autocorrelation matrix of the channel impulse response estimation are used, and at the same time, the QR decomposition value and the phase rotation matrix used for the next iterative calculation are updated. .

  In step B, the noise power is further estimated using the number of pilot subcarriers in each OFDM symbol, the maximum value of the subspace dimension, and the data power.

The forgetting factor

The range of

It is.

The forgetting factor

Range is

Is preferred.

Pre-threshold

And in step B, further noise power

of

Eigenvalues of at least double

Select and the selected eigenvalue

Subspace dimension based on

Where

It is.

Said

Range is

Is preferred.

In step C, calculate a time domain LS channel impulse response estimate based on time domain subspace tracking, said step C comprising:
After calculating the LS channel impulse response estimation in the time domain based on subspace tracking in the time domain, zero padding the LS channel impulse response estimation, and performing a fast Fourier transform to estimate the channel response in the frequency domain Including getting.

In step C, calculate an MMSE channel impulse response estimate in the time domain based on subspace tracking in the time domain, said step C,
Calculating an MMSE channel impulse response estimate based on time-space subspace tracking, zero padding the MMSE channel impulse response estimate, and performing a fast Fourier transform to obtain a channel response estimate in the frequency domain.

A channel estimator in an OFDM system,
A pilot signal position frequency domain channel estimation unit that performs channel estimation in the frequency domain for a channel at the position of the pilot signal and obtains a channel estimation result in the frequency domain;
A frequency domain / time domain transform unit that transforms the channel estimation result in the frequency domain from the frequency domain to the time domain so as to obtain a channel impulse response estimate in the time domain;
A subspace tracking unit that performs subspace tracking in the time domain for an autocorrelation matrix of channel impulse response estimation in the time domain;
An estimation unit based on subspace tracking in the time domain that calculates a channel impulse response estimate in the time domain based on the subspace tracking in the time domain;
A time domain / frequency domain transform unit that transforms the channel impulse response estimate in the time domain into the frequency domain to obtain a channel response estimate in the frequency domain.

  The pilot signal position frequency domain channel estimation unit is a pilot signal position frequency domain channel LS estimation unit or a pilot signal position frequency domain channel MMSE estimation unit.

The channel estimator further includes a noise power estimation unit that estimates noise power in a time domain, and a subspace dimension determination unit that determines a dimension of a subspace used for channel estimation in the subspace tracking,
An estimation unit based on subspace tracking in the time domain calculates a channel impulse response estimate based on the subspace tracking, the noise power and the dimension of the subspace.

  The estimation unit based on subspace tracking in the time domain is an LS estimation unit based on subspace tracking in the time domain or an MMSE estimation unit based on subspace tracking in the time domain.

  The OFDM system is a single antenna OFDM system or a multi-antenna OFDM system.

  As can be seen from the above technical scheme, in the channel estimation method in the OFDM system of the present invention, the channel estimation result in the frequency domain is converted from the frequency domain to the time domain to obtain the channel impulse response estimation in the time domain, Subspace tracking is performed on the autocorrelation matrix for channel impulse response estimation in the time domain. Therefore, the present invention greatly increases the channel estimation accuracy by calculating the channel impulse response based on subspace tracking, thereby separating the signal and noise, and significantly reducing interference with the signal due to noise. Further, in the present invention, the application of MMSE channel estimation can be specifically linked, thereby further improving channel estimation accuracy, and when performing MMSE channel estimation in the time domain, an operation for obtaining an inverse matrix is performed. There was no need to do so, thereby reducing complexity. Also, the application of the present invention does not require channel statistics, and always tracks channels that are changing. Note that the application range of the present invention is wide, and can be applied not only to a single antenna OFDM system but also to a multi-antenna OFDM system.

  At the same time, in the channel estimator in the OFDM system according to the present invention, the subspace tracking unit performs subspace tracking on the autocorrelation matrix of channel estimation in the time domain, and the estimation unit based on subspace tracking in the time domain is subspace. By calculating the channel impulse response based on the tracking, the separation of the signal and the noise is realized as well. Therefore, the channel estimation accuracy is greatly enhanced by significantly reducing interference with signals due to noise. Also, the estimation unit based on subspace tracking in the time domain preferably calculates the channel impulse response using MMSE channel estimation based on subspace tracking and noise power and subspace dimensions, whereby the channel The estimation accuracy is further increased. Note that when performing MMSE channel estimation in the time domain using the channel estimator of the present invention, it is not necessary to perform an operation for obtaining an inverse matrix, thereby greatly reducing the complexity.

  In order to further clarify the objects, technical solutions and merits of the present invention, the present invention will be described in more detail below with reference to the drawings and specific embodiments.

  The channel estimation method of the present invention is based on subspace tracking in the time domain. FIG. 2 is a flowchart of a channel estimation method in the time domain in an OFDM system according to the present invention.

As shown in Figure 2,
In step 201, a pilot signal is inserted into an OFDM symbol, and channel estimation is performed in the frequency domain for the channel at the position of the pilot signal.

  Here, generally, an OFDM pilot signal used for channel estimation is generated by a system transmitter, and the generated OFDM pilot signal is inserted into an OFDM symbol. Among these, the inserted pilot signal may be a comb-type pilot signal or a block type pilot signal, but in order to effectively track a radio fading channel that changes at high speed, the pilot pattern is equally spaced. It is preferable to insert a pilot-shaped pilot signal. If block type pilot insertion is employed, it is necessary to add a decision feedback module on the receiving side in order to realize tracking to a constantly changing channel. Here, the channel estimation in the frequency domain performed for the channel at the pilot signal position may be LS channel estimation or MMSE channel estimation, but it is preferable to adopt LS channel estimation here for simplicity.

  Step 202 transforms the channel estimation result in the frequency domain from the frequency domain to the time domain to obtain a channel impulse response estimate in the time domain, and subspace tracking the autocorrelation matrix of the channel impulse response estimation in the time domain I do.

  Here, it is preferable to convert the channel estimation result in the frequency domain from the frequency domain to the time domain by performing IFFT transform for frequency domain channel estimation. However, the method of converting the channel estimation result in the frequency domain from the frequency domain to the time domain is not limited to IFFT conversion. Also, performing subspace tracking on the autocorrelation matrix for impulse response estimation in the time domain includes updating of intermediate variables, QR decomposition of intermediate variables, and updating of phase rotation matrices. It is preferable to further estimate the noise power simultaneously with performing the subspace tracking.

  In step 203, a channel impulse response estimate based on subspace tracking is calculated, and the channel impulse response estimate based on subspace tracking is converted to the frequency domain to obtain a channel response estimate in the frequency domain.

  Here, eigenvectors and eigenvalues used for channel estimation are selected based on the results of subspace tracking and noise power estimation in the time domain, and then LS estimation based on subspace tracking in the time domain or time domain. MMSE estimation based on subspace tracking is calculated respectively. In general, MMSE estimation is preferably used here because the accuracy of MMSE estimation is better. Then, in order to acquire the channel response estimation in the complete frequency domain, each of the acquired LS estimation and MMSE estimation is zero-padded and then subjected to FFT conversion. In order to effectively remove noise, it is preferable to determine the dimension of the subspace more dynamically. Only eigenvalues greater than a certain multiple of noise power and the corresponding eigenvectors are used for channel estimation.

  Further, the channel estimation method according to the present invention is not only adapted to a single antenna OFDM system, but also to a multi-input multi-output OFDM (MIMO-OFDM) system. For MIMO-OFDM systems, orthogonal pilot signals can be assigned to different transmit antennas.

  Hereinafter, the algorithm principle of the method of the present invention will be described in detail.

First, a code is defined.

Represents the number of subcarriers in the OFDM symbol,

Represents the number of pilot subcarriers in each OFDM symbol,

Represents the number of samples corresponding to the maximum multipath delay (usually less than or equal to the cyclic prefix length)

Is a time variable.

When designing a system, from the sampling theorem in the frequency domain

It is understood that it is necessary to guarantee.

, Time

Represents the channel response vector in the frequency domain at.

, Time

Represents the channel impulse response vector in the time domain at.

Is the DFT matrix, where

.

Represents a mapping matrix.

, Time

Represents the received signal at.

, Time

Represents additive white Gaussian noise at.

, Time

Represents the transmitted signal at.

The corresponding variable of pilot signal position

When expressed as:

First, channel estimation is performed in the frequency domain for the position of the pilot signal. Here, LS estimation is preferred. Then

.

At this time,

about

The LS estimation of the channel impulse response in the time domain obtained by performing the IFFT transform of the points is as follows.

As the input signal,

Autocorrelation matrix

Perform subspace tracking for.

Where initialization

so,

Is the maximum subspace dimension,

Is an orthonormal matrix,

Is the phase rotation matrix,

Is the forgetting factor,

Is an intermediate variable.

Subspace tracking

Includes the following:

Update

Where

Is

Is a conjugate transpose of

QR decomposition for

And

Update

Where

Is

Is a conjugate transpose of
As the algorithm converges,

Is

Approaches the eigenvector matrix of

Is

Diagonal matrix of eigenvalues of

Get closer to.

Noise power at the same time as subspace tracking

It is preferable to estimate for.

here,

Is data power.

On the other hand, when performing channel estimation based on subspace tracking in the time domain, subspace dimensions are used to effectively remove noise.

Need to be determined dynamically, and only eigenvalues and corresponding eigenvectors that satisfy the following relationship are used for channel estimation. That is,

,here,

.

In addition, parameters that need to be set artificially in a specific implementation process are mainly forgetting factors.

And threshold

It is. As can be seen from the simulation, the selection of these two parameters is not very sensitive to the channel environment, so a certain range should be satisfied. usually,

It is. To reach a more ideal channel estimation performance,

Is preferred.

Eigenvector matrix used for channel estimation based on the results of subspace tracking and noise power estimation in the time domain

And eigenvalue diagonal matrix

Select. here,

Is

Before

Column,

Is

In the upper left corner

Contains only dimension elements.

And LS estimation based on subspace tracking in time domain

And MMSE estimation

Can be calculated respectively.

here,
When LS estimation is performed, LS estimation

Is

It is.

When MMSE estimation is performed, MMSE estimation

Is

It is.

After zero-padding for K-point FFT transform, channel response estimation in full frequency domain

Can be obtained respectively.

When LS estimation is performed,

.

When MMSE estimation is performed,

.

What should be explained is that when implementing the algorithm specifically, the power delay spectrum of the radio channel satisfies the characteristic of exponential decay,

In the time domain, taking into account that the latter half of the sample in time means that it mainly consists of noise.

When doing subspace tracking, complete

Instead of using

You may use only the sample in the front part. The specific proportion of disposal is determined by the channel environment and is usually

Only the first half of is used. In this way, not only the channel estimation accuracy can be increased, but also the computational complexity can be reduced.

  So far, the algorithm principle of the present invention has been described in detail.

  FIG. 3 is a diagram illustrating a channel estimator in an OFDM system according to an embodiment of the present invention. As shown in FIG. 3, the channel estimator includes a pilot signal position frequency domain channel LS estimation unit 301 that performs LS channel estimation in the frequency domain for the pilot signal position channel, and the LS channel estimation result in the frequency domain in time. Frequency domain / time domain transform unit 302 that transforms into a domain, subspace tracking unit 303 that performs subspace tracking on the autocorrelation matrix of channel LS estimation in the time domain, and noise power estimation that estimates noise power in the time domain A unit 304, a dimension determination unit 306 for determining the dimension of the subspace used for channel estimation in the subspace tracking, and the MMSE estimation based on the subspace tracking, the dimension of the subspace and the noise power Subspace tracking in time domain The MMSE estimation unit 305 based converts the MMSE estimate to the frequency domain, and a time-domain / frequency-domain conversion unit 307 to obtain the channel response estimates in the frequency domain.

  Here, when the demand for channel estimation accuracy is not so high, the noise power estimation unit 304 and the dimension determination unit 306 may be omitted in order to simplify the structure of the channel estimator. At this time, the MMSE estimation unit 305 based on subspace tracking in the time domain calculates MMSE estimation based only on subspace tracking. The frequency domain / time domain conversion unit 302 is preferably an IFFT unit. The time domain / frequency domain conversion unit 307 is preferably an FFT unit.

  FIG. 4 is a diagram illustrating a channel estimator in an OFDM system according to another embodiment of the present invention. As shown in FIG. 4, the channel estimator includes a pilot signal position frequency domain channel LS estimation unit 401 that performs LS channel estimation in the frequency domain for the channel of the pilot signal position, and time domain LS channel estimation in the frequency domain. A frequency domain / time domain transform unit 402 that transforms into a subspace tracking unit 403 that performs subspace tracking on the autocorrelation matrix of channel LS estimation in the time domain, and a noise power estimation unit that estimates noise power in the time domain 404, a dimension determination unit 406 for determining the dimension of the subspace used for channel estimation during the subspace tracking, the LS estimation is calculated based on the subspace tracking, the dimension of the subspace and the noise power Based on subspace tracking in the time domain Ku includes a LS estimation unit 405 converts the LS estimation into a frequency domain and a time obtaining a channel response estimate in the frequency domain domain / frequency domain conversion unit 407.

  Here, when the demand for channel estimation accuracy is not so high, the noise power estimation unit 404 and the dimension determination unit 406 may be omitted in order to simplify the structure of the channel estimator. At this time, the LS estimation unit 405 based on the subspace tracking in the time domain calculates the LS estimation based only on the subspace tracking. At the same time, the frequency domain / time domain transform unit 402 is preferably an IFFT unit. The time domain / frequency domain conversion unit 407 is preferably an FFT unit.

  In short, the present invention performs channel estimation based on subspace tracking in the time domain, and separates signal and noise. Therefore, the interference with the signal due to noise is remarkably reduced, thereby greatly increasing the channel estimation accuracy. In the present invention, in order to further improve the channel estimation accuracy, it is preferable to perform estimation using an MMSE algorithm based on subspace tracking in the time domain.

  FIG. 5 is a comparison diagram of simulations using the present invention and the prior art. Among them, curves 1, 2, and 3 are channel estimators obtained by using the conventional LS channel estimation method, the MMSE channel estimation method, and the method that combines LS and frequency subspace tracking, respectively. And curve 4 is a performance curve diagram of the channel estimator based on the method of the present invention. As can be seen in FIG. 5, the channel estimation accuracy is significantly improved and the computational complexity is reduced compared to various prior art techniques. Therefore, it is very advantageous to use the present invention in a high mobility and low SNR environment.

  The above are only preferred embodiments of the present invention and do not limit the protection scope of the present invention. Various modifications, equivalent switching, improvements and the like made within the spirit and principle of the present invention should all be included in the protection scope of the present invention.

3 is a flowchart of an MMSE channel estimation method in a time domain in an OFDM system according to the prior art. 3 is a flowchart of a channel estimation method in a time domain in an OFDM system according to the present invention. FIG. 3 shows a channel estimator in an OFDM system according to an embodiment of the present invention. FIG. 6 is a diagram illustrating a channel estimator in an OFDM system according to another embodiment of the present invention. It is a simulation comparison figure which each uses this invention and a prior art.

Explanation of symbols

301 ... Pilot signal position frequency domain channel estimation unit
302 ... Frequency domain / time domain conversion unit
303… Subspace tracking unit
304 ... Noise power estimation unit
305 ... MMSE estimation unit
306 ... Dimension determination unit
307… Time domain / frequency domain conversion unit
401 ... Pilot signal position frequency domain channel estimation unit
402: Frequency domain / time domain conversion unit
403 ... Subspace tracking unit
404 ... Noise power estimation unit
405 ... LS estimation unit
406 ... Dimension determination unit
407… Time domain / frequency domain conversion unit

Claims (17)

A channel estimation method in an orthogonal frequency division multiplexing OFDM system, comprising:
Inserting a pilot signal into an OFDM symbol, performing channel estimation in the frequency domain for the channel at the pilot signal position to obtain a channel estimation result in the frequency domain; and
The channel estimation result in the frequency domain is converted from the frequency domain to the time domain to obtain a channel impulse response estimation in the time domain, and the autocorrelation matrix of the channel impulse response estimation in the time domain is subspaced in the time domain. Step B for tracking,
Calculating a channel impulse response estimate in the time domain based on subspace tracking in the time domain, and converting the channel impulse response estimate in the time domain to the frequency domain to obtain a channel response estimate in the frequency domain When,
A channel estimation method comprising:   2. The channel estimation according to claim 1, wherein in step A, the pilot signal is temporally continuous in an OFDM symbol, and a pilot signal is inserted for every certain subcarrier, or a block type pilot signal is inserted. Method.   2. The channel estimation method according to claim 1, wherein in step A, least square LS channel estimation in the frequency domain or minimum mean square error MMSE channel estimation in the frequency domain is performed on the channel at the pilot signal position.   2. In step B, the channel estimation result in the frequency domain is transformed from the frequency domain to the time domain by performing an inverse fast Fourier transform on the channel estimation result in the frequency domain. Channel estimation method.   In step A, the least square LS channel estimation in the frequency domain is performed on the channel at the position of the pilot signal, and in step B, subspace tracking is performed on the autocorrelation matrix of the channel impulse response estimation in the time domain, where In the subspace tracking, an intermediate variable is repeatedly calculated using a predetermined forgetting factor. In the iterative calculation, the QR decomposition value of the intermediate variable, the phase rotation matrix, and the autocorrelation matrix of the channel impulse response estimation are included. 2. The channel estimation method according to claim 1, wherein a QR decomposition value and a phase rotation matrix that are used simultaneously and used for the next iterative calculation are updated.   6. The channel estimation method according to claim 5, wherein the noise power is further estimated in step B using the number of pilot subcarriers in each OFDM symbol, the maximum value of the subspace dimension, and the data power. The forgetting factor

Range is

7. The channel estimation method according to claim 6, wherein The forgetting factor

Range is

8. The channel estimation method according to claim 7, wherein Pre-threshold

In step B, the noise power

Multiples

Select and the selected eigenvalue

To determine the dimension r of the subspace based on

6. The channel estimation method according to claim 5, wherein: Said

Range is

10. The channel estimation method according to claim 9, wherein In step C, calculate an LS channel impulse response estimate in the time domain based on subspace tracking in the time domain, said step C,
LS channel impulse response estimate in time domain based on subspace tracking in time domain is calculated, zero padded to the LS channel impulse response estimate, fast Fourier transform is performed, and channel response estimate in the frequency domain is obtained 10. The channel estimation method according to claim 9, further comprising: In step C, calculate an MMSE channel impulse response estimate in the time domain based on subspace tracking in the time domain;
Calculating an MMSE channel impulse response estimate based on subspace tracking in the time domain, zero padding the MMSE channel impulse response estimate, and performing a fast Fourier transform to obtain the channel response estimate in the frequency domain 10. The channel estimation method according to claim 9, wherein A channel estimator in an OFDM system,
A pilot signal position frequency domain channel estimation unit that performs channel estimation in the frequency domain for a channel in the pilot signal position and obtains a channel estimation result in the frequency domain;
A frequency domain / time domain transform unit that transforms the channel estimation result in the frequency domain from the frequency domain to the time domain to obtain a channel impulse response estimate in the time domain;
A subspace tracking unit that performs subspace tracking in the time domain for an autocorrelation matrix of channel impulse response estimation in the time domain;
An estimation unit based on subspace tracking in the time domain to calculate a channel impulse response estimate in the time domain based on the subspace tracking in the time domain;
A channel estimator comprising: a time domain / frequency domain transform unit that converts the channel impulse response estimate in the time domain into the frequency domain to obtain a channel response estimate in the frequency domain.   14. The channel estimator according to claim 13, wherein the pilot signal position frequency domain channel estimation unit is a pilot signal position frequency domain channel LS estimation unit or a pilot signal position frequency domain channel MMSE estimation unit. The channel estimator further includes a noise power estimation unit that estimates noise power in a time domain, and a subspace dimension determination unit that determines a dimension of a subspace used for channel estimation in the subspace tracking,
14. The channel of claim 13, wherein the estimation unit based on subspace tracking in the time domain calculates a channel impulse response estimate based on the subspace tracking, the noise power and the dimension of the subspace. Estimator.   The estimation unit based on subspace tracking in the time domain is an LS estimation unit based on subspace tracking in the time domain or an MMSE estimation unit based on subspace tracking in the time domain. The described channel estimator.   14. The channel estimator according to claim 13, wherein the OFDM system is a single antenna OFDM system or a multi-antenna OFDM system.

Images