JP5222578B2 - Channel estimation method - Google Patents

Description

  The present invention relates to the field of wireless mobile communication technology, and more particularly to a channel estimation method.

  Orthogonal frequency division multiplexing (OFDM) technology has the ability to resist inter-symbol interference (ISI), and at the same time can provide very high frequency utilization efficiency, so it may be adopted in next generation wireless mobile communication systems Is the highest transmission technology. The technology is often applied in a wide field such as digital user loop, digital audio / video broadcasting, wireless LAN, and wireless MAN.

  In order for an OFDM communication system to have good performance in a wireless mobile channel environment, an estimation that is as accurate as possible for a time-varying multi-fading channel must be made. The quality of channel estimation has a very important impact on the performance of an OFDM system.

  Currently, a practical channel estimation method is performed through known pilot information. Due to differences in processing flows, pilot-based channel estimation can be divided into frequency domain channel estimation and time domain channel estimation. In general, time domain channel estimation is an effective channel estimation method because high channel estimation accuracy can be obtained with appropriate complexity. Since time domain channel estimation is performed through discrete Fourier transform and inverse discrete Fourier transform (DFT / IDFT), it is also called a channel estimation method by DFT.

The specific processing flow of the channel estimation method by DFT is as shown in FIG.
Step 101 for performing least-square (LS) channel estimation for all pilots, and inverse fast Fourier transform (IFFT) for the frequency-domain channel estimation results of the pilot positions to obtain time-domain channel impulse response estimation values Obtaining step 102;
Dividing the time-series data of the channel impulse response estimation value in the time domain into the first half and the second half, and inserting 0 between them;
Corresponding to step 104 of performing fast Fourier transform (FFT) on the channel impulse response estimated value in the time domain after processing to obtain a channel estimated value (CFR) in the entire frequency domain.

  If all of the pilot information can be acquired, the channel estimation method using DFT can acquire a good channel estimation result.

  However, the channel estimation method by DFT needs to use an equi-spaced comb-like pilot across the entire frequency domain range or a block-like pilot that can be regarded as a special case of the comb-like pilot. Also, for an actual OFDM communication system, virtual subcarriers that cannot be used for data and pilot transmission are indispensable, and the low frequency virtual subcarriers of the virtual subcarriers remove the DC portion of the signal and increase the high frequency. Virtual subcarriers provide a guard time that can reduce signal leakage. Compared to the ideal situation without virtual subcarriers, in a real system, some pilots required by the DFT-based channel estimation method always fall within the range of virtual subcarriers. Therefore, if the conventional DFT channel estimation method is used as it is, the channel estimation error is greatly increased. In particular, the channel estimation error becomes more apparent at subcarrier positions around the virtual subcarrier. . In order to simplify the description to be described later, a pilot that falls within the virtual subcarrier range and exists in an ideal situation is referred to as a virtual pilot.

  Two solutions have been proposed as a way to reduce the performance degradation of a DFT-based channel estimation method in an actual OFDM communication system.

  One solution reconstructs the frequency domain channel estimation of the virtual subcarrier position by extrapolation or prediction, and then uses the channel estimation method shown in FIG. However, the performance gained with the solution depends on the accuracy of extrapolation or prediction, and is usually error-prone in actual systems. The solution also increases the computational complexity of the system.

  Another solution is to insert a zero in the frequency domain virtual pilot position and obtain a time domain channel impulse response estimate via IFFT before performing step 102 in FIG. The number of samples is increased by inserting 0 with respect to the channel impulse response estimation value, and then converted into the frequency domain through the FFT conversion in step 104. The solution requires the assumption that the timing synchronization is quite accurate in order to determine the position to insert zeros in the time domain channel impulse response estimate. On the other hand, in an actual OFDM communication system, it cannot be avoided that an error exists in timing synchronization. Further, the channel estimation value obtained by this method still has a large error in subcarriers around the virtual subcarrier.

  In summary, since there must be virtual subcarriers in an actual OFDM communication system, the channel estimation error obtained by the current DFT channel estimation method is large, especially at subcarrier positions around the virtual subcarriers. The error becomes clearer. The two types of reforming methods still cannot solve the above problem effectively.

  In view of the above problems, an object of the present invention is to provide a channel estimation method that effectively reduces the degradation of channel estimation performance due to virtual subcarriers and does not increase the calculation complexity of channel estimation.

  In order to solve the above problems, the present invention provides the following technical solutions.

A channel estimation method comprising:
The signal is composed of subcarriers used for signal transmission and virtual subcarriers not used for signal transmission. For the pilots arranged in the subcarrier range, normal channel estimation is performed, and the signal is within the virtual subcarrier range. In the above, the channel estimation result is used to set the virtual channel estimation value for the virtual pilot so that the continuity of the channel estimation value is guaranteed over all the subcarriers including the virtual subcarrier. Obtaining a frequency domain channel estimate at all pilot positions of the carrier;
Performing an inverse Fourier transform on the frequency domain channel estimate obtained in step a to obtain a time domain channel impulse response estimate;
A step c in which 0 is inserted into the time domain channel impulse response estimate obtained in step b to increase the number of samples, and then Fourier transform is performed to obtain channel estimates in all frequency domains including data symbols. The gist is to include.

  In step a, the normal channel estimation for the pilot is to perform channel estimation using a least squares LS algorithm or a least mean square error MMSE algorithm.

The virtual pilot includes only a virtual pilot in a high frequency region,
In step a, setting the virtual channel estimate for the virtual pilot so that the continuity of the channel estimate across all subcarriers is ensured is for the pilot located at the boundary with the virtual subcarrier. The channel estimation value in the frequency domain is set as a virtual channel estimation value of the virtual pilot in the high frequency domain, or after setting the virtual channel estimation, it is further multiplied by a predetermined windowing function. The frequency pilot channel estimation value of the virtual pilot in the frequency domain is converged to a certain value.

The virtual pilot further includes a virtual pilot in a low frequency region,
In step a, setting the virtual channel estimate for the virtual pilot so that the continuity of the channel estimate is guaranteed across all subcarriers,
The channel estimation value in the frequency domain of the pilot located at the boundary with the virtual subcarrier is set as the virtual channel estimation value of the virtual pilot in the low frequency domain, or further set in advance after setting the virtual channel estimation. The channel estimation value in the frequency region of the virtual pilot in the high frequency region is converged to a certain value by multiplying the calculated windowing function, or the channel estimation value in the low frequency region is calculated using interpolation. .

  In the step c, increasing the number of samples by inserting 0 into the time domain channel impulse response estimation value obtained in the step b means that the time-series data of the time domain channel impulse response estimation value is set to the first half and the second half. That is, inserting 0 in the middle.

  The position where the time-series data of the channel impulse response estimation value in the time domain is divided into the first half and the second half is set by the channel delay profile and the timing offset.

The method
Prior to step c, time domain smoothing is performed on the acquired time domain channel impulse response estimate and / or frequency domain smoothing is performed on the frequency domain channel estimate obtained in step c. It further includes performing processing.

  When the pilot subcarrier start position is 0, the method uses the frequency domain channel estimation value obtained in step c as a channel estimation result. When the pilot subcarrier start position is a value other than 0, step c The frequency domain channel estimation value acquired in step (1) is cyclically shifted in the higher frequency direction, and the frequency domain channel estimation value after the shift is further used as a channel estimation result.

  The present invention guarantees continuity by inserting zeros into the virtual pilot, and reserves the channel energy in the most time domain as much as possible, so that the virtuality for the channel estimation performance is almost never increased. The influence of subcarriers can be effectively reduced and can be eliminated, and in particular, channel estimation errors at subcarrier positions around the virtual subcarrier can be reduced.

  In the best mode of the present invention, channel estimation is performed using a linear interpolation method for a virtual pilot in a low frequency region, and simple boundary value repetition is performed for a virtual pilot in a high frequency region. Very simple.

  Hereinafter, the beneficial effects of the present invention will be analyzed with reference to FIGS.

  First, for the simulation parameters of the present invention, the number of subcarriers of an OFDM symbol is 1024, the bandwidth is 20 MHz, the OFDM symbol length is 51.2 μs, the cyclic prefix length is 5 μs, and the virtual subcarriers Are 0 and 413 to 611, the number of equidistant comb-like pilots (including effective pilots and virtual pilots) is 128, and the channel model adopts the COST207 indoor channel model. λ takes 3/4.

  The MSE performance of the channel estimation of the present invention is as shown in FIGS. The present invention can effectively reduce or eliminate the influence of virtual subcarriers on channel estimation performance on the premise that the complexity is not increased. In particular, the channel estimation error of subcarrier positions around the virtual subcarriers can be reduced. Can be reduced.

  Hereinafter, the technical solution of the present invention will be described in more detail with reference to the drawings and embodiments.

となる。そのうち、Ｋ、Ｐ、Ｉはいずれも２のパワーである。バーチャルサブキャリアが存在すると、バーチャルサブキャリアとサブキャリア範囲内に配置されたパイロット数はそれぞれ

となり、サブキャリアのサフィックスはｋであり、且つ、

である。パイロットサブキャリアのスタート位置が

であると、取得した周波数領域のチャネル推定値が必要とする結果になる。その他の場合は、取得した周波数領域のチャネル推定値に対して該当するサイクリックシフトを行う必要がある。例えば、サブキャリアのスタート位置が

の場合、取得した周波数領域のチャネル推定値を周波数の高い方向へ２ビットサイクリックシフトさせることが必要となる。他の場合も類似であるので、ここでは重複に説明しない。ここのパイロットサブキャリアは、サブキャリアも含むし、バーチャルサブキャリアも含む。 First, the number of subcarriers in the OFDM system is K, the number of virtual subcarriers is V, the number of subcarriers is D, and D is usually an even number. Then, obviously, the following equation holds. That is, K = V + D. Without virtual subcarriers, the number of equally spaced comb pilots is P, and the pilot interval I is

It becomes. Among them, K, P, and I are all 2 powers. If there are virtual subcarriers, the number of pilots allocated within the virtual subcarrier and subcarrier range is

And the subcarrier suffix is k, and

It is. Pilot subcarrier start position

, The obtained frequency domain channel estimation value is required. In other cases, it is necessary to perform a corresponding cyclic shift on the acquired channel estimation value in the frequency domain. For example, if the subcarrier start position is

In this case, it is necessary to shift the obtained channel estimation value in the frequency domain by a 2-bit cyclic shift in a higher frequency direction. Since the other cases are similar, they will not be described redundantly here. The pilot subcarriers here include subcarriers as well as virtual subcarriers.

と示すことができる。 Usually, the cyclic prefix (CP) length of the OFDM symbol is larger than the maximum multipath delay L of the radio channel, and the channel is constant within the time length of one OFDM symbol. Then, the OFDM system is a parallel transmission model

Can be shown.

は、それぞれ、k個目のサブキャリア上の送信信号、受信信号、周波数領域チャネルゲイン及び加算性ホワイトガウスノイズを示す。パイロットサブキャリアについて、

そのうち、

Ｉは、パイロット間隔を示す。

の場合を考慮すると、

と

がバーチャルサブキャリアになる。Ｐ個のパイロットのうち、

がバーチャルサブキャリアの範囲に入る。そのうち、

は、Ｉに対して除算した商を示す。

のバーチャルパイロットは、高周波数領域におけるバーチャルサブキャリアの位置にあるパイロットを含み、もちろん、低周波数領域におけるバーチャルサブキャリアが設置されている場合、低周波数領域におけるバーチャルサブキャリアの位置にあるパイロットも含む。 Of which

Respectively denote a transmission signal, a reception signal, a frequency domain channel gain, and additive white Gaussian noise on the kth subcarrier. About pilot subcarriers

Of which

I indicates the pilot interval.

Considering the case of

When

Becomes a virtual subcarrier. Of the P pilots

Falls within the range of virtual subcarriers. Of which

Indicates a quotient obtained by dividing I.

The virtual pilots include the pilots located at the virtual subcarriers in the high frequency region and, of course, the pilots located at the virtual subcarriers in the low frequency region when the virtual subcarriers are installed in the low frequency region. .

  Hereinafter, the channel estimation flow will be described in detail. The channel estimation flow is as shown in FIG. 2 and corresponds to the following steps.

  In step 201, normal channel estimation is performed for pilots arranged in the subcarrier range, and virtual channel estimation values are set for virtual pilots arranged in the virtual subcarrier range, ie, virtual Fill the pilot to ensure continuity and obtain the frequency domain channel estimate CFR1 at the pilot position.

  In this step, the processing method adopted for the pilot and the virtual pilot is not limited.

のパイロットの場合、ＬＳアルゴリズムを採用してもよいし、ＭＭＳＥアルゴリズムを採用してもよい。ここでは、ＬＳアルゴリズムのみを例として説明する。具体的には、下記公式（１）でこれらパイロットの周波数領域のチャネル推定値を取得できる。

For example,

, The LS algorithm may be employed or the MMSE algorithm may be employed. Here, only the LS algorithm will be described as an example. Specifically, the channel estimation values in the frequency domain of these pilots can be acquired by the following formula (1).

のバーチャルパイロットに対する処理方法を説明する。 Hereinafter, in this embodiment, it was mentioned above.

A processing method for virtual pilots will be described.

  When the virtual pilot includes only the virtual pilot in the high frequency region, the process is performed only on the virtual pilot in the high frequency region.

For virtual pilots in the high frequency region, in order to avoid large errors in the channel estimation results in subcarriers around the virtual subcarrier, ie, Gibbs effect, a virtual channel estimation value is set in the virtual pilot, that is, The channel estimates at the virtual subcarrier boundaries may be made continuous by performing appropriate filling on the virtual pilots. Specifically, the channel estimation values in the frequency domain of the pilot located at the virtual subcarrier boundary may be used as the virtual channel estimation values of the virtual pilot in the high frequency domain, with simple boundary value overlap. Further, the channel estimation value in the frequency domain of the virtual pilot in the two-stage high frequency domain may be converged to a certain value by further multiplying a predetermined windowing function after the boundary value overlap. The value may be 0 or another value. Multiplying a predetermined windowing function may use any conventional windowing algorithm. When the simplest boundary value overlap is used, the filling value of the virtual pilot in the high frequency region can be obtained by the following formula (2) and formula (3).

を得ることができる。 By the above processing, the frequency domain channel estimate of the pilot position of length P, ie CFR1:

Can be obtained.

  Of course, if the virtual pilot further includes a virtual pilot in the low frequency region, it is necessary to process the virtual pilot in the low frequency region. Normally, there is one virtual pilot in the low frequency region, but there may be a plurality of virtual pilots.

  Specifically, when processing a virtual pilot in the low frequency region, the processing method employed for the virtual pilot in the high frequency region, that is, a simple boundary value overlap or a predetermined boundary value overlap is further determined in advance. You may also use a method that multiplies the windowing functions.

For the virtual pilot in the low frequency region, the channel estimation value of the corresponding frequency region may be acquired using the interpolation method without using the above processing. In this embodiment, only linear interpolation will be described as an example. Specifically, the channel estimation value in the frequency domain of the virtual pilot in the low frequency domain can be obtained by the following formula (4).

  When the boundary value duplication processing is performed on the virtual pilot in the high frequency region and the linear interpolation processing is performed on the virtual pilot in the low frequency region, the frequency of the virtual subcarrier boundary is obtained by these processing as shown in FIG. Since the channel estimation of the region is prevented from drastically changing, the degradation problem of the channel estimation error due to the Gibbs effect can be effectively solved.

  In step 202, P-point IFFT conversion is performed on CFR1 acquired in step 201, and a channel impulse response estimation value in the time domain of length P, that is, CIR1 can be acquired.

The CIR is shown as follows.

  In step 203, CIR1 obtained in step 202 is divided into the first half and the second half, 0 is inserted to obtain CIR2, K-point FFT conversion is performed on CIR2, and all subcarriers including the last data symbol are obtained. Obtain the frequency domain channel estimate, ie, CFR2.

である。 CIR2 in this step is

It is.

であり、同時に、無線チャネルの指数フェージング特性及びシステム中の回避できないタイミングオフセットを考慮すると、ＣＰの長さに基づいて、上記公式（５）で取得したＣＩＲ１に対して、公式（６）に示す前半と後半に分けることと、０の挿入処理を行うことで、長さがＫのＣＩＲ２を得ることができる。

In this step, the position where CIR1 is divided into the first half and the second half needs to be set in consideration of the delay profile of the channel and the timing offset. Ideally, the time domain channel energy is concentrated in the CIR1 head, but in the actual system, the time domain channel energy is leaked due to the presence of virtual subcarriers, that is, the CIR1 tail is partially included. Including channel energy. Obviously, most of the time domain channel energy of CIR1 should be reserved. The cyclic prefix length of the OFDM symbol is greater than or equal to the maximum multipath delay of the channel, i.e.

At the same time, taking into account the exponential fading characteristics of the radio channel and the unavoidable timing offset in the system, the formula (6) shows the CIR1 obtained by the formula (5) based on the CP length. CIR2 with a length of K can be obtained by dividing into the first half and the second half and performing the insertion process of 0.

は比例因子である。具体的な値は、チャネルの遅延プロファイルとタイミング誤差の統計特性に依頼する。通常、λは、３/４又は７/８を取ればいい。 Of which

Is a proportional factor. Specific values are requested from the statistical characteristics of the channel delay profile and timing error. Usually, λ should be 3/4 or 7/8.

となる。 The CFR2 acquired in this step is

It becomes.

  Of course, in order to further improve the channel estimation performance, a time domain smoothing process is performed on CIR1 acquired in step 202 and / or a frequency domain smoothing process is performed on CFR2 acquired in step 203. Thus, noise can be further suppressed.

  The above is not limited to the preferred embodiment of the present invention. It is obvious to those skilled in the art that various improvements can be made on the premise that the principles of the present invention are not deviated, and these improvements also belong to the protection scope of the present invention.

FIG. 1 is a flowchart for realizing a channel estimation method based on the current DFT. FIG. 2 is an implementation flowchart of the channel estimation method according to the present invention. FIG. 3 is a diagram illustrating a simulation effect of the relationship between the mean square error (MSE) and the signal-to-noise ratio (SNR) according to the present invention. FIG. 4 is a diagram illustrating a simulation effect of the relationship between the MSE and the subcarrier index (Index) according to the present invention. FIG. 5 is a diagram illustrating processing performed on the high-frequency virtual pilot and the low-frequency virtual pilot according to the present invention.

Claims (6)

The signal consists of subcarriers used for signal transmission and virtual subcarriers not used for signal transmission.
For the pilots arranged in the subcarrier range, normal channel estimation is performed, and in the virtual subcarrier range, the channel estimation result is used, and the channel estimation value over all subcarriers including virtual subcarriers. A virtual channel estimate is set for the virtual pilot so that continuity of the sub-carrier is guaranteed, and frequency-domain channel estimates at all pilot positions of all subcarriers are obtained,
Performing an inverse Fourier transform on the frequency domain channel estimate obtained in step a to obtain a time domain channel impulse response estimate;
A step c in which 0 is inserted into the time domain channel impulse response estimate obtained in step b to increase the number of samples, and then Fourier transform is performed to obtain channel estimates in all frequency domains including data symbols. Including
The virtual pilot includes a virtual pilot in a high frequency region and a virtual pilot in a low frequency region ,
In step a, setting the virtual channel estimate for the virtual pilot so that the continuity of the channel estimate across all subcarriers is ensured is for the pilot located at the boundary with the virtual subcarrier. The channel estimation value in the frequency domain is set as a virtual channel estimation value of the virtual pilot in the high frequency domain, or after setting the virtual channel estimation, it is further multiplied by a predetermined windowing function. A channel estimation method comprising: converging a channel estimation value in a frequency domain of a virtual pilot in a frequency domain to a certain value; and calculating a channel estimation value in a low frequency domain using a linear interpolation method.   The method of claim 1, wherein in step a, the normal channel estimation for the pilot is to perform channel estimation using a least squares LS algorithm or a least mean square error MMSE algorithm.   In the step c, increasing the number of samples by inserting 0 into the time domain channel impulse response estimation value obtained in the step b means that the time-series data of the time domain channel impulse response estimation value is set to the first half and the second half. 2. The method according to claim 1, characterized in that a zero is inserted in between. 4. The method according to claim 3 , wherein the position of dividing the time-series data of the channel impulse response estimation value in the time domain into the first half and the second half is set by a channel delay profile and a timing offset. Before step c, perform time domain smoothing on the obtained time domain channel impulse response estimate, and / or
2. The method of claim 1, further comprising performing frequency domain smoothing on the frequency domain channel estimate obtained in step c. When the start position of the pilot subcarrier is 0, the channel domain estimation value obtained in step c is the channel estimation result,
When the pilot subcarrier start position is a value other than 0, the frequency domain channel estimation value obtained in step c is cyclically shifted in the higher frequency direction, and the frequency domain channel estimation value after the shift is channeled. 2. The method of claim 1 further comprising providing an estimation result.

Images