WO2010135946A1

WO2010135946A1 - Method and apparatus for realizing decimal frequency offset estimation

Info

Publication number: WO2010135946A1
Application number: PCT/CN2010/072436
Authority: WO
Inventors: 邱宁; 李强; 曹南山; 陈力; 张涛; 游月意; 姚扬中
Original assignee: 中兴通讯股份有限公司
Priority date: 2009-05-27
Filing date: 2010-05-04
Publication date: 2010-12-02
Also published as: CN101902426B; CN101902426A

Abstract

A method and apparatus for realizing decimal frequency offset estimation are provided by the present invention. The relationship of decimal frequency offset values between in the frequency domain and in the time domain is utilized, and the frequency domain correlation is adopted to replace the time domain correlation, and correlation sequence components corresponding to the highest power value and its neighbourhood are deleted from a frequency domain correlation sequence, only the remained correlation sequence is accumulated to obtain a frequency offset. Because the frequency of the single tone interference is generally not equal to the subcarrier frequencies, the single tone interference, after a frequency domain transform, will be decomposed onto the nearest frequency point and the neighboring frequency points. Hence, by deleting the components near the highest value, the influence of the single tone interference is better eliminated, thus the robustness of the decimal frequency offset estimation is improved greatly, and the influence of the single tone interference and direct-current component on the performance of the decimal frequency offset estimation is eliminated.

Description

Method and device for realizing fractional frequency offset estimation

The present invention relates to a carrier frequency offset estimation and compensation technique, and more particularly to a method and apparatus for implementing fractional frequency offset estimation based on a terminal device of China Mobile Multimedia Broadcasting (CMMB) technology. Background technique

The CMMB standard uses the mobile TV receiving standard STiMi independently developed by China. The CMMB standard is the first domestically developed system for mobile phones, PDAs, MP3s, MP4s, digital cameras, laptops, and a variety of mobile terminals, using S-band satellite signals. Achieve integrated coverage of the world, national roaming, support 25 sets of TV programs and 30 sets of radio programs. The CMMB standard specifies the frame structure, channel coding and modulation of the broadcast channel transmission signal of the mobile multimedia broadcasting system in the frequency range of the broadcasting service. The CMMB standard is applicable to the broadcasting service frequency in the frequency range of 30 MHz to 3000 MHz, via satellite and/or terrestrial wireless. A national broadcasting system that transmits multimedia signals such as television, radio, and data information can achieve national roaming.

The CMMB standard uses Orthogonal Frequency Division Multiplexing (OFDM) technology. OFDM is a well-known multi-carrier modulation technology. Its main principle is: divide the channel into several orthogonal subchannels, and convert high-speed data into Parallel low-speed sub-data streams are modulated onto each sub-channel for transmission. The orthogonal signals can be separated by using correlation techniques at the receiving end, which can reduce mutual interference between subchannels. The signal bandwidth on each subchannel is smaller than the associated bandwidth of the channel, so each subchannel can be considered as flatness fading, thereby eliminating intersymbol interference. And since each subchannel bandwidth is only a small fraction of the original channel bandwidth, channel equalization becomes relatively easy. OFDM is currently used in several wireless system standards, such as European digital audio and digital video broadcasting systems such as DAB, DVB-T, DVB-H, 5GHz high data rate wireless LAN (IEEE802.11a, HiperLan2, MMAC) systems. Wait.

Since the sub-channels of OFDM have a small bandwidth and are highly sensitive to carrier frequency deviation, very accurate frequency synchronization is required. According to the relationship between the frequency deviation and the subcarrier spacing, it can be divided into a fractional part and an integer part. The integer part only shifts the information signal on the subchannel without destroying the orthogonality between the subcarriers; Subchannel interference, which destroys the orthogonality between subcarriers, causes the BER performance of the system to be seriously degraded. Therefore, the time domain first-order fractional frequency offset estimation and compensation is performed at the receiving end to eliminate the ICI caused by the fractional part, and then the translation in the frequency domain can eliminate the influence of the carrier frequency offset.

At present, the fractional frequency offset estimation method is implemented by using the training symbol information based on the time domain correlation method. It generally includes the following steps:

The implementation steps of the existing estimation method are as follows:

First, the received signals are stored in the preset arrays in the corresponding order / («) and / " ₂ («);

L-1

Then, use R _t {n)r* (n) to multiply and accumulate r _y (n) and r ₂ * (n) to obtain the correlation value R, and finally, use Δ / _ε = ' ^{WAF to} obtain the fractional frequency Partial estimation results. Suppose two time-domain sync symbols are sent; there are delays of N _d samples between (n) and 3⁄4 (n), and the repeat symbol length is L, as shown in Figure 1, Figure 1 is separated by N samples. Schematic diagram of repeated training symbols for two time domains. The above steps are specifically implemented as follows:

When there is carrier frequency deviation Δ/ _ε and carrier phase deviation Δζ3, the two received training symbols) and r ₂ («) can be expressed as equation (1) and formula ( ² ), respectively:

r (n) = r{n) = x _l (n) exp (- j{27iAf _c nT _s + Αφ)) + η _ι (nT _s )

(1)

r ₂ (n) = r(n + N _d ) = x ₂ (n) exp(-j(2 _c (n + N _d )T _s + Αφ)) + η, {{n + N _d )T _s ) (2)

Where /i = 0,l,-",L- 1

At the receiving end, by correlating the time domain of the two training symbols, the intermediate variable, the correlation value R, is obtained, as shown in equation (3):

_ (n) exp(- (2 / r _s +Αφ)) + η _γ (nT _s )] x = exp(j2^f _c N _d T _s x, (n)x ₂ (n) + η

.

Without considering noise, and; = « = 0,1,2,~, L-1, then, the correlation value R, as shown in equation (4):

It can be seen from equation (4) that the phase of the correlation value R _f is as shown in equation (5):

Arg(R _t ) = 2 Af _c N _d T _s

(5)

It can be seen that in this case, the phase of the obtained correlation value R _t has no relationship with the carrier phase deviation. Therefore, the carrier frequency deviation is as shown in formula (6):

Arg(R _f ) _ arg(R _t )NAF

c =f _c ~f _c

2πΝ , 1 gas

In equation (6), the carrier spacing A = 1/Nr _s For the phase of the correlation value R _t , the range of variation of arg(R _f ).

Jan et al. have shown that in the AWGN channel, the existing fractional frequency offset estimation is the maximum likelihood estimation; in addition, in the AWGN channel, Schimdl et al. analyzed the performance of the existing fractional frequency offset estimation and gave The variance of the carrier deviation Δ / _ε estimate i7 _A ² _f oc ^ _.

f ^c LSNR

When there is monophonic interference, the price of the existing fractional frequency offset estimation method will have a great influence. The introduction of monophonic interference will not only degrade the existing fractional frequency offset method from unbiased estimation to biased estimation, but also estimate. The variance will also increase significantly. The specific instructions are as follows:

Assume that the single tone interference signal in the passband = A. _ex p( 2 t) , where / is the frequency of the single tone interference signal. Then, the two training symbols received can be represented as formula (1') and formula (2'), respectively:

r in) = r{n) = x _l (n) exp (- j{l /f _c nT _s + Αφ)) + A · exp(j'2« ) + η _ι (nT _s )

r ₂ (n) = r(n + N _d ) = x ₂ (n) exp(-;(2 _c (n + N _d )T _S + Αφ)) + A · exp(;2^ _f (n + N _d )T _S ) + _l {{n + N _d )T _s )

Where w = 0,l,-", L-1. At the receiving end, by correlating the time domain of the two training symbols, the intermediate variable, ie the correlation value, is given by the formula (3'):

L-1

^ i [j (n) exp (- j, Hf _c nT _s + Αφ)) + α + η _ν (nT _s )] x

~ [x ₂ (n) exp( 2nAf _c (n + N _d )T _s +Αφ)) + *+ η ((n + N _d )T _s )]

L-1

= exp(;2 - _c N _d r _s )g _Xl (n)x ₂ (n) + LA ² - exp(-j2 _d T _s ) + η

among them, (n) exp (- j, Hf _c nT _s + Αφ)) - * + x ₂ * (n) exp(j(27iAf _c (n + N _d )T _s + Αφ)) · + =∑3⁄4 (n ) exp(— (2 / r _s + Αφ)) - η{ ((n + N _d )T _S ) + x ₂ (n) exp(;(2 _c (n + N _d )T _S + Αφ)) · η _γ (nT _s ) + " ⁼⁰ a · η, {{n + N _d )T _S ) + - η, (nT _s ) + η _γ (nT _s ) · η, {{n + N _d ) T _S )]

The expectation is 0.

The phase estimate of the correlation value R _f is expected to be the formula (5'):

(5' )

It can be seen that the introduction of monophonic interference not only degrades the existing fractional frequency offset method from unbiased estimation to biased estimation, but also increases the variance of the estimated value.

Figure 2 is a schematic diagram showing the simulation results of the influence of single-tone interference on the existing fractional frequency offset estimation. As shown in Fig. 2, the abscissa represents the signal-to-noise ratio (SNR), and the ordinate represents the estimated standard frequency offset between the estimation result and the true value. For example, if the H is not tested in the AWGN channel, the fractional frequency offset is 1/4 subcarriers and there is 1.25MHz single tone interference, the simulation result of the fractional frequency offset estimation accuracy of the CMMB mobile TV simulation platform. Curve 21 is the simulation result without single tone interference, curve 22 is the simulation result obtained by -20dB single tone interference, curve 23 is the simulation result obtained by the presence of -10dB tone interference, and curve 24 is obtained by the presence of OdB tone interference. Simulation results. It is obvious from the simulation curve that when there is no single-tone interference, the estimated standard frequency offset decreases monotonously with the increase of the signal-to-noise ratio.

SNR=-4dB reaches 0.01, and SNR=10dB is close to one thousandth; however, after the receiver introduces the single tone interference with the equal signal power (ie, OdB), the existing frequency offset estimation method collapses directly, and the estimation result Error, and is not sensitive to signal-to-noise ratio changes. Even if the tone interference is reduced to -10dB relative to the wanted signal, the estimated standard frequency offset is very large, and the performance pelvic floor is inferior to one percent, which does not meet the requirements for high quality demodulation of OFDM. Estimated performance can only be met in most SNR intervals when the tone interference drops to -20dB.

It can be seen from the simulation results shown in Fig. 2 that the estimation accuracy of the existing methods for implementing fractional frequency offset estimation is seriously affected by external interference, which greatly reduces the robustness of fractional frequency offset estimation. Summary of the invention

In view of this, the main object of the present invention is to provide a method for implementing fractional frequency offset estimation, which can greatly reduce the influence of external interference, thereby greatly improving the robustness of fractional frequency offset estimation.

Another object of the present invention is to provide an apparatus for implementing fractional frequency offset estimation, which can greatly reduce the influence of external interference, thereby greatly improving the robustness of fractional frequency offset estimation.

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

A method for implementing fractional frequency offset estimation, the method comprising the following steps:

Converting the received time domain signal into a frequency domain signal and performing frequency domain correlation;

Calculating and searching for the position of the corresponding power in the ranked related sequence, and deleting the relevant sequence corresponding to the position and the related sequence of the position related to the position;

The remaining related sequences after the arrangement are accumulated, and the frequency deviation is obtained by using the accumulated values.

The transform method is a Fourier FFT transform.

After performing the frequency domain correlation, before calculating and searching for the position of the corresponding power in the aligned correlation sequence, the method further includes: arranging the correlation sequences obtained by the frequency domain correlation according to the frequency from small to large; The subsequent related sequence R" (k) is: R" (k)

a correlation sequence obtained after correlation in the frequency domain; L is a repetition symbol length of the received time domain signal.

When the position corresponding to the maximum power is '<10, the position-related position is: φ = {0Χ···, ^,···, ^ + 9, k' + lO; L + k' -lO, L + k' -9,---,Ll};

When the position corresponding to the maximum power satisfies ZJ2-ll<^≤L/2-l, the position-related position is: = μ-10, ^-9, ···, ^,···, /2 - 1};

The position-dependent when the position corresponding to the maximum power satisfies L/2 ≤ < L/2 + 10 The position is: i3 = {L/2, L/2 + l, ---, t', ---, t' +9, ^ +10};

When the position corresponding to the maximum power satisfies f > L-ll, the position-related position is: ^ = {0,1,···,^' -L + 9,fc' -L + 10;^' -I0,k' -9,---,k',---,Ll};

Otherwise, the position associated with the position ^ is: = {fc — 10,^—9,···, ^,···, ^ + 9,^ + 10}; where L is the received The repeat symbol length of the time domain signal.

The correlation sequence of the position is: each of the ten related sequences on both sides of the position of the highest power in the correlation sequence after the corresponding arrangement.

And accumulating the remaining related correlation sequences, and obtaining the frequency deviation Δ/ _ε by using the accumulated value is specifically:

△ / = / - / = ^ ^ = ^{arg (} learn ^F , where is the subcarrier spacing, A = 1/Nr,

N is the number of subcarriers, N _d is the number of samples, and R is the accumulated correlation sequence.

An apparatus for implementing fractional frequency offset estimation, the apparatus comprising a preprocessing module, a frequency domain correlation module, a sequencing module, an interference deletion module, and a frequency deviation acquisition module, wherein

a pre-processing module, configured to transform the received time domain signal into a frequency domain signal;

a frequency domain correlation module, configured to receive a frequency signal transformed from the preprocessing module, and perform frequency domain correlation on the frequency domain signal;

The interference deletion module is configured to calculate and search for a position corresponding to the highest power in the correlation sequence output by the frequency domain correlation module, delete the correlation sequence corresponding to the position and the correlation sequence of the preset position related to the position, and output the remaining correlation sequence Giving a frequency deviation acquisition module;

The frequency deviation obtaining module is configured to accumulate the remaining related sequences after the arrangement, and obtain the frequency deviation by using the accumulated value.

The apparatus further includes a sorting module disposed between the frequency domain correlation module and the interference deletion module, configured to arrange related sequences output from the frequency domain correlation module according to a frequency from small to large, and sort the related sequences. Output to the interference removal module.

It can be seen from the technical solution provided by the present invention that the present invention utilizes fractional frequency offset values in frequency. The relationship between the domain and the time domain, the frequency domain correlation is used to replace the time domain correlation, and the power maximum and its related sequence components are deleted from the frequency domain correlation sequence, and only the remaining residual correlation sequences are used to obtain the frequency deviation. . Since the tone interference frequency is usually not equal to the subcarrier frequency, the monophonic interference after the frequency domain transformation is decomposed into the nearest frequency point and the adjacent frequency point. Therefore, by removing the component adjacent to the maximum value, the influence of the single-tone interference is better eliminated, thereby greatly improving the robustness of the fractional frequency offset estimation, and eliminating the influence of the single-tone interference and the DC component on the fractional frequency offset estimation performance. . DRAWINGS

1 is a schematic diagram of two time domain repetition training symbols separated by N _d samples;

2 is a schematic diagram of simulation results of influence of single tone interference on existing fractional frequency offset estimation; FIG. 3 is a flowchart of a method for implementing fractional frequency offset estimation according to the present invention;

4 is a schematic structural diagram of a device for implementing fractional frequency offset estimation according to the present invention;

FIG. 5 is a schematic diagram showing the simulation results of the influence of single tone interference on the fractional frequency offset estimation of the present invention. Detailed ways

FIG. 3 is a flowchart of a method for implementing fractional frequency offset estimation according to the present invention. As shown in FIG. 3, the method of the present invention includes the following steps:

Step 300: Convert the received time domain signal into a frequency domain signal.

Assume that two time-domain sync symbols are transmitted; there are delays of N _d samples between (n) and 3⁄4 (n), and the repeat symbol length is L, as shown in Figure 1, when there is carrier frequency deviation Δ/ _ε When the carrier phase deviation occurs, the two training symbols r» and r ₂ («) are shown in equation (1) and formula (2), respectively. In this step, by Fourier transform (FFT), "» and /" ₂ («) are converted and stored in the preset array to obtain the frequency domain sync symbol sequence r; k) and r ₂ (fc) as shown in equations (7) and (8), respectively: r _Y ( k) = ^ r(/i) exp( - j27ikn l N)

(7) ^r 2 =∑ r(i + N _d ) exp( - jlTikn IN)

(8)

Where = ο, υ_ι.

Step 301: Perform frequency domain correlation on the frequency domain signal.

In this step, first, at the receiving end, by correlating the frequency domain of the two frequency domain synchronization symbol sequences, the frequency domain correlation sequence is obtained as shown in the formula (9) (the * symbol indicates the conjugate operation):

R _f {k) = _ri {k)r;{k)

=

IN)

·∑{[3⁄4 (n) exp(-;(2 _c (n + N _d )T _S +Αφ)) + η[ ((n + N _d )T _S )] exp( - jlTtim IN) }* = Exp(2 " _c N _d r _s )3⁄4 (k)x* (k) + η

(9)

Through the frequency domain correlation of this step, the frequency domain transform of the two segments of the same transmission sequence is also the same feature, the phase information of the transmission sequence itself is eliminated, and further, the same transmission characteristics of the same frequency channel are used to eliminate the channel. Phase information.

Further, the step 301 may further include arranging the obtained correlation sequence R _f (fc) according to the frequency from small to large, and obtaining the aligned correlation sequence (W is represented by the formula (10):

(10)

The sorting method uses half of the total number of the serial numbers as the demarcation point, divides the sequence into upper and lower halves, and pairs the upper and lower halves to obtain a new sequence, which is reordered by the order of frequency from small to large. The post-row sequence is in descending order of the frequency from small to large.

Step 302: Calculate and search for the location of the corresponding power in the correlation sequence after the frequency domain correlation, The relevant sequence corresponding to the location and the related sequence of the location associated with the location are deleted.

Search for the value in the range of = 0,1,.,L-1, so that =, the power value |R; ( )| is the maximum value of all power values |R; (w|, that is, the single-tone interference is found Frequency position: According to the specific value of the position corresponding to the maximum power, the preset position related to the position is determined as the deletion set:

1) When ^ <10, φ = {0Χ····^',····^' + 9,k' +lO;L + k' -lO,L + k' -9,---,Ll };

2) When L/2_ll< '≤L/2_1, φ = {Κ -\Q,k -9,---,k ,---,LI2-\};

3) When ZJ2≤ ' <L/2 + 10, f = {LI2,LI2 + \,---,k ,---,k +9,k +

4) When ^〉 L—11, f = {Q,\,---,k -L + 9,k -L + \Q;k -\Q,k -9,---,k ,- --,L-\}\

5) When ^ does not belong to the range listed in 1) ~ 4), φ = {^ -10,k' -9,---,k',---,k' + 9,k' + 10}. If in step 301, the obtained correlation sequence has been obtained in order of frequency from small to large.

R^fc) is arranged, then, in this step, the position associated with the position 'set in advance is the deletion set, and the deletion set is = {fc'- 1( - 9,... , . '+9 +10} , instead of Then use the above 1) ~ 5) to determine the location associated with the location 'as a deletion set. It should be noted that the number of deletions depends on the actual scheme design. Generally, it is possible to achieve better results by deleting 10 related sequences on both sides of the position corresponding to the highest power.

Since the single-tone interference frequency is usually not equal to the sub-carrier frequency, the single-tone interference after the frequency domain transformation is decomposed into the nearest frequency point and the adjacent frequency point. Therefore, by this step, the component adjacent to the maximum value is deleted, preferably Eliminates the effects of monophonic interference.

Step 303: Accumulate the remaining correlation sequences, and use the accumulated values to obtain the frequency deviation.

After obtaining the set of deleted locations, use Equation (11) to accumulate the correlation values of all remaining positions except the deleted location set. Equation (11) is as follows:

(11)

L-l L-l

Where = ^^» _ j27ikn/ Ν,, χ ₂ {Κ) = ^χ ₂ (n) exp( - ]2ται / N) , η The expectation is 0.

Without considering the noise, and («) = («)," = 0,1,2,... - 1, the correlation value obtained by the method of the invention ?' is as shown in equation (12):

R=exp(j2 Af _c N _d T _s ) ( (12)

The phase of the correlation value R is arg(R) = Hf _c N _d T _s . With the method of the present invention, the phase of the correlation value R obtained has no relationship with the carrier phase deviation, so the carrier frequency deviation is as shown in the formula (13): _ = arg(R ) = _arg (R)NAF

(13)

Where Δ F is the subcarrier spacing, A = 1/Nr _s , N is the number of subcarriers, and N _d is the number of samples.

R is the accumulated correlation sequence. The phase angle of the accumulation result obtained fractional frequency offset estimation result 2 _d T _s times, given in the standard case, I IN S is a constant, for example, in the CMMB standard _{^{N d = 2048,7; = 1x10- 7}} , the phase The angle is divided by 2πΝ _ά Ί to obtain the actual fractional frequency offset estimate. It can be seen that the improved estimation method of the present invention is an unbiased estimation method of fractional frequency offset, and substantially eliminates the influence of single tone interference and DC component (which can be understood as special tone interference with frequency 0) on frequency offset estimation. .

The invention utilizes the relationship between the fractional frequency offset value in the frequency domain and the time domain, and uses the frequency domain delay correlation to replace the time domain delay correlation, and deletes the power maximum value and its nearby components in the frequency domain delay correlation value, and greatly improves The robustness of the fractional frequency offset estimation eliminates the influence of single-tone interference and DC components on the performance of fractional frequency offset estimation.

For the method of the present invention, an apparatus for implementing fractional frequency offset estimation is provided. FIG. 4 is a schematic structural diagram of a device for implementing fractional frequency offset estimation according to the present invention. As shown in FIG. 4, the method includes a preprocessing module and a frequency domain correlation module. a sorting module, an interference removing module, and a frequency deviation obtaining module, wherein a pre-processing module, configured to transform the received time domain signal into a frequency domain signal.

The frequency domain correlation module is configured to receive the transformed frequency signal from the preprocessing module, and perform frequency domain correlation on the frequency domain signal.

The interference deletion module is configured to calculate and search for a position corresponding to the highest power in the correlation sequence output by the frequency domain correlation module, delete the correlation sequence corresponding to the position and the correlation sequence of the preset position related to the position, and output the remaining correlation sequence Give the frequency deviation acquisition module.

A frequency deviation acquisition module is configured to accumulate the remaining correlation sequences and obtain the frequency deviation using the accumulated values.

5 is a schematic diagram showing the simulation results of the influence of the single tone interference on the fractional frequency offset estimation of the present invention. As shown in FIG. 5, the abscissa represents the SNR, and the ordinate represents the estimated standard frequency offset between the estimation result and the real value, for example, 5 is 4 The simulation environment is the AWGN channel, the decimal frequency offset is 1/4 subcarriers and there is 1.25MHz single tone interference, the simulation result of the fractional frequency offset estimation accuracy of the CMMB mobile TV simulation platform. Curve 51 is the simulation result obtained by the method of the present invention in the presence of -10 dB tone interference, curve 52 is the simulation result obtained by the method of the present invention in the presence of OdB tone interference, and curve 53 is the presence of -10 dB tone interference, using existing The simulation result obtained by the method, curve 54 is the simulation result obtained by using the existing method for the presence of OdB monophonic interference. It is obvious from the simulation curve that the method of the invention is insensitive to monophonic interference, and more accurate estimation results can be obtained under different degrees of monophonic interference.

The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included. Within the scope of protection of the present invention.

Claims

Claim

A method for implementing fractional frequency offset estimation, the method comprising the steps of: converting a received time domain signal into a frequency domain signal and performing frequency domain correlation;

2. The method according to claim 1, wherein the transform method is a Fourier FFT transform.

The method according to claim 1, wherein after the frequency domain correlation is performed, before calculating and searching for the location of the aligned correlation sequence having the highest power, the method further comprises: following the frequency from small to large. The sequence is related to the correlation sequence obtained by the frequency domain correlation; the aligned correlation sequence R" (k) is: R" (k)

4. The method of claim 1 wherein:

When the position corresponding to the maximum power is '<10, the position related to the position' is: φ = {0Χ···, ^,···, ^ + 9, k' + lO; L + k' -lO , L + k' -9, ---, Ll};

When the position corresponding to the maximum power satisfies ZJ2-ll< ≤ L/2-l, the position associated with the position ' is: μ μ — 10, fc — 9,···, L/2—1};

When the position corresponding to the maximum power satisfies L / 2 ≤ ^ < L / 2 + 10, the position associated with the position 'is: i3 = {L/2, L/2 + l, ---, fc', ---,fc' +9,^ +10};

When the position corresponding to the maximum power satisfies >L-ll, the position related to the position ' is: φ = {0Χ···, ^ -L + 9, k' -L + I0; k' -\ k - 9,--- ,k ,L-\} Otherwise, the position-dependent position is: = μ - 10, fc - 9, '", fc', '", fc' + 9, fc' + 10}; where L is the received time The repeat symbol length of the domain signal.

The method according to claim 3, wherein the correlation sequence of the position is: each of ten related sequences on both sides of the position with the highest power among the corresponding ranked correlation sequences.

6. The method according to claim 1, wherein the accumulating the remaining related sequences and obtaining the frequency deviation Δ/ _ε by using the accumulated value is:

/ =^^= ^{arg (} learn ^F , where AF is the subcarrier spacing, A = 1/Nr ,

7. A device for implementing fractional frequency offset estimation, the device comprising a preprocessing module, a frequency domain correlation module, a sequencing module, an interference deletion module, and a frequency deviation acquisition module, wherein the preprocessing module is configured to receive The time domain signal to be converted into a frequency domain signal;

The device according to claim 7, further comprising a sorting module disposed between the frequency domain correlation module and the interference deletion module, configured to correlate from the frequency domain according to a frequency from small to large The relevant sequences output by the module are arranged, and the sorted related sequences are output to the interference deletion module.