WO2012155459A1

WO2012155459A1 - Method and device for correcting frequency offset

Info

Publication number: WO2012155459A1
Application number: PCT/CN2011/081276
Authority: WO
Inventors: 刘宜佳; 郭军平
Original assignee: 中兴通讯股份有限公司
Priority date: 2011-05-18
Filing date: 2011-10-25
Publication date: 2012-11-22
Also published as: CN102790738B; CN102790738A

Abstract

The invention provides a method and device for correcting frequency offset. Through the response of all the pilot carrier for every tilestrip in every sub-channel, the corresponding matrix is Constructed, and the constructed matrix is normalized, and the average of the normalized matrix of every antenna is determined, thus the average of the normalized matrix of every received antenna for the base station is determined, and the Fast Fourier Transform (FFT) is proceeded according to it, thereby the estimation of the frequency offset is proceeded, then the frequency offset is corrected according to the result of the estimation of the frequency offset. Consequently the estimation and correction of the frequency offset is realized in the Orthogonal Frequency Division Multiplexing (OFDM) system.

Description

Method and device for frequency offset correction

The present invention relates to the field of wireless communication technologies, and in particular, to a frequency offset correction method and apparatus. Background technique

In the Orthogonal Frequency Division Multiplexing (OFDM) technology, the subcarrier spectrums are required to overlap each other, so strict requirements are imposed on the orthogonality between subcarriers. Once the orthogonality between the subcarriers is not strictly satisfied, interference occurs between the subcarriers, thereby deteriorating the performance of the system. So for OFDM systems and multiple input multiple outputs

(Multi-Input Multiple-Out-put, MIMO) - In OFDM systems, the synchronization of the carrier frequency at the transceiver end and the stability of the carrier phase pose high requirements.

The frequency deviation from the transmitter to the local oscillator (LO) of the receiver is called the carrier frequency offset.

(CFO), Doppler shift is also one of the CFO sources in mobile channel environments. The CFO not only causes the phase rotation of the demodulation constellation point to affect the symbol decision, but also causes the orthogonality between the subcarriers of the OFDM system to be destroyed, resulting in mutual interference (ICI) between the subcarriers, resulting in serious system performance. decline. In addition, early rise has shown that a smaller CFO will result in a large system performance loss of OFDM, and also shows that under the Additive White Gaussian Noise (AWGN) channel, it is greater than 4% of the subcarrier spacing. The loss caused by the CFO of the multipath fading channel greater than 2% of the subcarrier spacing can not be neglected, and CFO estimation and compensation measures must be taken. MIMO-OFDM systems are also very sensitive to CFOs.

Due to the sensitivity of the OFDM system and the MIMO-OFDM system to the CFO, the actual system design must adopt CFO estimation and correction measures to ensure system performance, but there is no relevant technical support at present. Summary of the invention

In view of this, an embodiment of the present invention provides a method and apparatus for frequency offset correction, which are used to

The carrier frequency offset in the OFDM system and the MIMO-OFDM system is estimated and corrected.

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

A method for frequency offset correction, comprising:

For the data received on one antenna of the base station, the average of the responses of the different pilot subcarriers is obtained according to the response of all the pilot subcarriers of each Tilestrip in each subchannel, and the column vector is constructed according to this, and according to the structure The column vector constructs the matrix, scales the constructed matrix, and normalizes the scaled matrix;

Determining, based on each Tilestri normalized matrix, an average of the normalized matrix of each antenna, and determining an average of the normalized matrix of the base station;

The fast Fourier transform is performed according to the elements in the average value of the matrix normalized by the base station, and the frequency offset correction is performed according to the result of the fast Fourier transform.

The process of constructing a column vector according to the average of the responses of the different pilot subcarriers obtained includes:

Determining a sequence number corresponding to the average value of the different pilot subcarriers according to the number of orthogonal frequency division multiplexing OFDM symbols of the pilot subcarrier in the time domain;

After obtaining the average of the responses of the different pilot subcarriers, the average of the obtained responses is divided into two groups, wherein the average of the pilot subcarrier responses with the odd sequence number is a group, and the pilot number with the serial number is even. The average of the subcarrier responses is a group, and column vectors are constructed according to each group.

The process of constructing a matrix according to the constructed column vector includes:

Construct two matrices and ^ according to the constructed column vector ^ζ ι , where

Where '— 2 , , '′ and, '′ are responses of different pilot subcarriers; a matrix constructed by combining the sum of the two matrices ^R i as a column vector.

Wherein, before constructing the two matrices and ^, the method further includes:

Move each of the two matrices and right one bit to the right.

The process of performing scale processing on the constructed matrix includes:

Determining the largest absolute element of the real and imaginary parts of all elements in the constructed matrix

M;

Determine the number of bits S1 of the element M relative to the 32-bit signed number of left shifts;

According to the determined number of bits S1, the real part and the virtual step of all elements in the matrix are shifted to the left by S1 bit;

The height 16 bits of the real part and the imaginary part of all elements in the matrix after the left shift are intercepted, and this is used as a scale processed matrix.

The process of normalizing the scaled matrix includes: determining a sum of real parts of the diagonal elements of the matrix after the scale processing, using the sum as a trace of the matrix; determining the matrix and the The quotient of the trace of the matrix, which is the result of normalization of the matrix.

The process of performing frequency offset correction according to the result of the fast Fourier transform includes:

_ -;'-Δ? ! -1

Frequency offset correction according to ^{_ 3} , where i represents the ith OFDM symbol,

D is the normalized frequency offset, which is the value of the carrier before the frequency offset correction, and Si is the corrected value.

The method further includes: before performing the frequency offset correction, the method further includes:

Determining the maximum value of the column vector of the real part structure according to the column vector constructed by the element in the average value of the normalized matrix and the column vector constructed by the real part of the column vector fast Fourier transform Position, based on which frequency offset estimation is performed to determine the normalized frequency offset.

A frequency offset correction device includes:

a first normalization processing module, configured to receive data on an antenna of a base station, Obtaining the average of the responses of different pilot subcarriers according to the response of all the pilot subcarriers of each Tilestrip in each subchannel, constructing a column vector according to the structure, and constructing a matrix according to the constructed column vector to scale the constructed matrix Processing, normalizing the scaled matrix; a second normalization processing module for determining an average of the normalized matrix of each antenna according to the normalized matrix of each Tilestrip And determining an average value of the normalized matrix of the base station according to the method;

And a frequency offset correction module, configured to perform fast Fourier transform according to an element in an average value of the matrix normalized by the base station, and perform frequency offset correction according to the result of the fast Fourier transform.

The first normalization processing module is configured to: when constructing a column vector according to the average of the responses of the different pilot subcarriers obtained:

Determining a sequence number corresponding to an average value of different pilot subcarriers according to a number of OFDM symbols of the pilot subcarrier in the time domain;

The first normalization processing module constructs a matrix according to the constructed column vector, and is configured to: construct two matrices and ^ according to the constructed column vector ^ζ ι, ^z 2 , where

z ₂ z ₄ ''Z. 2N

among them

, , '' and the response for different pilot subcarriers;

The sum of the two matrices ^ and ^R 2 is constructed as a matrix of column vectors.

The first normalization processing module performs scale processing on the constructed matrix, and is used to: determine an element with the largest absolute value of the real part and the imaginary part of all elements in the constructed matrix. M;

Determining the number of bits of the element M relative to the 32-bit signed number of left shifts SI;

The first normalization processing module is used to: when normalizing the scale processed matrix:

The sum of the real parts of the diagonal elements of the matrix after scale processing is determined, and the sum is taken as a trace of the matrix; the quotient of the matrix and the trace of the matrix is determined, and the quotient is used as a result of normalization to the matrix.

Wherein the offset correction module corrects the frequency offset based on a result of the fast Fourier transform, for: ^_ offset correction is performed according to ^3, where i represents the i-th OFDM symbol,

The device further includes:

a frequency offset estimation module, configured to determine a real part structure according to a column vector constructed by an element in an average value of the normalized matrix and a column vector constructed by the real part of the column vector fast Fourier transform The position of the maximum value in the column vector, from which the frequency offset estimation is performed to determine the normalized frequency offset.

Embodiments of the present invention provide a method and apparatus for frequency offset correction, constructing corresponding matrices by using responses of all pilot subcarriers of each Tilestrip in each subchannel, and normalizing and processing the constructed matrix to determine The average of the normalized matrix of each antenna, thereby determining the average of the normalized matrix of each receiving antenna in the base station, and performing fast Fourier transform according to determining the average value of the normalized matrix of the base station Therefore, the frequency offset estimation is performed, and the frequency offset correction is performed according to the result of the frequency offset estimation. The invention achieves estimation and correction of frequency offset in an OFDM system. DRAWINGS

1 is a flow chart of frequency offset correction according to an embodiment of the present invention;

2 is a schematic diagram of a basic resource structure of an uplink data channel in an IEEE 802.16e system; FIG. 3 is a diagram of a device for frequency offset correction according to an embodiment of the present invention. detailed description

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments in order to make the present invention. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

FIG. 1 is a process of frequency offset correction according to an embodiment of the present invention, where the process includes the following steps:

S101: Obtain an average value of responses of different pilot subcarriers according to responses of all pilot subcarriers of each Tilestrip in each subchannel for data received on one antenna of the base station.

Among them, Tilestrip is a collection of Tiles of a user on the same frequency resource, which is a 2-dimensional concept. For subcarrier part occupancy (PUSC) carrier mapping, the Tilestrip contains (4*3*SlotDurNum) subcarriers.

S102: Construct a column vector according to the average of the responses of the obtained different pilot subcarriers, and construct a matrix according to the constructed column vector, and perform scale processing on the constructed matrix.

The scaled scale processing, that is, a set of data is enlarged or reduced by 2a times to adapt to its storage range; wherein a is a scale value, the 2a times is equivalent to shifting a bit; a is positive means left shift, a is Negative means right shift.

Configuring the column vector according to the average of the responses of the different pilot subcarriers obtained includes: determining the sequence number corresponding to the average value of the different pilot subcarriers according to the number of the OFDM symbols of the pilot subcarriers in the time domain;

After obtaining the average of the responses of the different pilot subcarriers, the average of the obtained responses is divided into two groups, wherein the average of the pilot subcarrier responses with the odd sequence number is a group, and the pilot number with the serial number is even. The average of the subcarrier responses is a group, and column vectors are constructed according to each group. Constructing a matrix based on the constructed column vector includes:

Construct two matrices according to the constructed column vector ^Z 1, ^Z 2 ,

^R l = 3⁄4 ^Z 4 ''Z. 2N

among them

, A'' and '· are responses of different pilot subcarriers;

Determine the sum of the two matrices of the construct and ^, and construct the sum as a matrix of the column vectors.

S103: The matrix processed by the scale and the trace of the processed matrix are normalized to the processed matrix.

Specifically, in the embodiment of the present invention, the matrix is normalized according to the scale processed matrix and the trace of the matrix, including:

Determining the sum of the real parts of the diagonal elements of the matrix after the scale processing, and using the sum as a trace of the matrix;

The quotient of the matrix and the trace of the matrix is determined, and the quotient is used as a result of normalization to the matrix. S104: Determine, according to the normalized matrix of each Tilestrip, an average value of the normalized matrix of each antenna, and determine the base station normalization according to the determined average value of the normalized matrix of each antenna. The average of the matrix after the transformation.

S105: Perform fast Fourier transform according to the elements in the average value of the matrix normalized by the base station, and perform frequency offset correction according to the result of the fast Fourier transform.

Before the frequency offset correction is performed according to the result of the fast Fourier transform, the method further includes: a column vector constructed according to an element in an average value of the normalized matrix, and a fast Fourier transform on the column vector a column vector constructed by the real part of the column vector, determining the position of the maximum value in the column vector of the real part structure;

Based on the determined position of the maximum value, a frequency offset estimation is performed to determine a normalized frequency offset. Since the embodiment of the present invention constructs a corresponding matrix by responding to all pilot subcarriers of each Tilestrip in each subchannel, and normalizing the constructed matrix, determining the average of the normalized matrix of each antenna. a value, thereby determining an average value of a matrix normalized by each receiving antenna of the base station, performing fast Fourier transform according to determining an average value of the normalized matrix of the base station, thereby performing frequency offset estimation, and estimating by frequency offset The result is frequency offset correction. Thereby, the estimation and correction of the frequency offset in the OFDM system is realized.

2 is a schematic diagram of a basic resource structure of an uplink data channel in an IEEE 802.16e system. In FIG. 2, the horizontal axis is a time domain OFDM symbol, and the vertical axis is a frequency domain subcarrier, including N physical resource units, where the background is a shaded square. The grid represents the pilot carrier, and the other squares represent the data carrier. The scheme of frequency offset estimation and correction proposed by the embodiment of the present invention is implemented based on this structure.

Since six Tilestrips are included in one subchannel, a base station includes a plurality of subchannels on one antenna. In the embodiment of the present invention, first, for a data received on a receiving antenna of a base station, a Tilestrip is selected, and responses of all pilot subcarriers in the Tilestrip are taken out, for example, ^^, ^ ... ^ and ^,)^...)^ , where ₁₁₁ is the mth subcarrier in the frequency domain, and n is the nth OFDM symbol in the time domain.

The responses to different pilot subcarriers are averaged for the same time domain OFDM symbols. According to the above FIG. 2, the process of finding the average value of the response is to divide the upper and lower pilots.

. Specifically, when obtaining the average value of the response of the different pilot subcarriers, in the process of the fixed point implementation, the response of the acquired different pilot subcarriers needs to be shifted right by 1 bit, and the response of the pilot subcarrier after the right shift is performed. Adding, then dividing by 2, is the average of the responses.

When determining the average of the responses of the different pilot subcarriers, the sequence number corresponding to the average of the responses of the different pilot subcarriers is determined according to the number of the OFDM symbols of the pilot subcarriers in the time domain. After obtaining the average of the responses of different pilot subcarriers, the average value of the obtained responses is divided into In the two groups, the average of the responses of the pilot subcarriers with the odd sequence number is a group, and the average of the responses of the pilot subcarriers with the even sequence number is a group. That is due to

Therefore, when ' , ' is divided into two groups, the first group includes , 3⁄4ν- ι , and the second group includes

..., 3⁄4. A column vector is constructed based on the average of the responses of the different pilot subcarriers contained in each group after grouping. The column vectors constructed here are ^Ζ ι = ......, 3⁄4v— and

^Z 2 = [3⁄4, ,..., 3⁄4ν , and construct two matrices ^ and ^R 2 according to the constructed two column vectors ^Ζ 1, ^Ζ 2 , where

^R l = ^Z l■ ^Z lz ₂ z ₄ ''B 2N

Determine the sum of the two matrices of the construct, that is, find the sum of ^ and ^, and sum it as R.

^{R = R} l ^{+ R} 2. Before determining the sum of the two matrices constructed, it is necessary to shift each element of the matrix and ^ to the right by 1 bit, and the sum of the two matrices shifted by 1 bit to the right as the matrix R.

After the matrix R is determined, the matrix R needs to be scaled, wherein the process of scaling the matrix R includes: determining an element M having the largest absolute value of the real part and the imaginary part of all elements in the matrix, where M is a value greater than 0; determining the number of bits S1 of the element M relative to the left bit shift of the 32-bit signed number; and according to the determined number of bits S1, the real part and the virtual step of all elements in the matrix R are shifted to the left by S1; The high 16 bits of the real part and the imaginary part of all the elements in the matrix after the left shift are intercepted as the matrix R after the scale processing. After the matrix R is scaled, the resulting matrix can be represented by R.

After the matrix R is obtained, the matrix processed by the scale needs to be normalized. The process of the normalization process in the embodiment of the present invention includes: the matrix processed according to the scale R, and the trace of the R, matrix, normalize the matrix R after the scale processing. And normalizing the matrix R after the scale processing according to the matrix R after the scale processing, and the trace of the R, the matrix, comprising: determining a diagonal element of the matrix R after the scale processing The sum of the real part, the sum of the sum as the matrix R, the quotient of the trace R, and the trace of the matrix R, which is the normalized result R of the matrix R, ie irace( R ')

The trace of the matrix.

Specifically, when the normalization process of the matrix R is implemented by a computer, the trace ^ira (R ') of the matrix R ' may be first determined according to the sum of the real parts of the diagonal elements of the matrix R ', and the calculation ira^ With respect to the 32-bit signed number, the maximum number of bits S2 can be shifted to the left, ^) is shifted left (S2-15), and scale(TileIdx) = -S2 is recorded. Call the 32-bit unsigned number division function, divide 0x3fffffff by ^{trace R} , and record the result W. At this time, the range of W is 0x3fff~0x7fff. All elements of the matrix R ' are multiplied by W to obtain a normalized matrix.

In the above process, it is processed for one Tilestrip in the subchannel. Since each subchannel includes 6 Tilestrips, each Tilestrip is processed according to the above process, and the normalized matrix R corresponding to each Tilestrip is obtained. And the value of the scale(Tileldx) corresponding to the matrix, and the normalized matrix corresponding to all the Tilestrips in the subchannel is averaged for each subchannel.

The averaged matrix corresponding to all the tiles in the subchannel is averaged for each subchannel, including:

Finding a minimum value of a scale (Tileldx) corresponding to the six Tilestrips according to a value of a scale (Tileldx) saved in the normalization process corresponding to each Tilestrip in the subchannel, and the minimum value i is a scaleMin; The matrix R after the normalization of the Tilestrip corresponding to the minimum value of scale(Tileldx) is not performed. Processing, but for the real and imaginary parts of all elements in the normalized matrix R corresponding to other Tilestrip, right shift the corresponding number of bits, where the right shift is the scale of the other Tilestrip (Tileldx) and The difference of scaleMin, the matrix after the right shift is used as the unified matrix corresponding to the other Tilestri;

Determining, according to the unified matrix corresponding to each Tilestrip, an average value R of the unified matrix of the subchannel, and using the scaleMin as the scale (channelldx) of the subchannel, that is,

⁶ = ^! , for each unified matrix corresponding to the Tilestrip, the normalized matrix of the Tilestrip corresponding to the scaleMin is the unified matrix corresponding to the Tilestrip.

In the above process, for each subchannel, according to the normalized matrix corresponding to each Tilestrip in the subchannel, the normalized matrix averaging values corresponding to all Tilestrips in the subchannel are determined. Each antenna includes a plurality of subchannels, and an average of the normalized matrices of each of the antennas is determined according to the determined normalized matrices of all the Tilestris in each subchannel. The process of determining the average value of the normalized matrix of each antenna is similar to the process of determining the average value of the normalized matrix of each subchannel, that is, for each subchannel corresponding to the scale (channelldx), The minimum value of scale (channelldx), for the minimum value of the scale (channelldx), the unified matrix of each other subchannel relative to the minimum value of the scale (channelldx), unified for each subchannel A matrix, determining an average of the normalized matrix of the antenna, and taking the minimum value of the scale (channelldx) as the scale (AntIdx) of the antenna.

After determining the average of the normalized matrix of each antenna, the average of the normalized matrix of the base station is determined according to the average of the normalized matrix of each antenna. When determining the average value of the normalized matrix of the base station, determining the base station normalization according to the average value of the normalized matrix of each antenna and the scale (Antldx) of the antenna saved for each antenna The average of the matrix after the transformation, the specific process and the above-mentioned determination of the matrix of each subchannel normalized The mean, and the process of determining the average of the normalized matrix for each antenna are similar, and are not mentioned here.

Specifically, in the embodiment of the present invention, when the average value of the normalized matrix of the base station is determined for each antenna, it is difficult to divide the N-th power of the divisor by a fixed-point number, so the conversion is multiplied by Count down and save the countdown as a table. Since the base station supports up to 8 antennas, the reciprocal table is defined as: Q[8] = { 0x7fff, 0x3fff, 0x2aaa, Oxlfff, 0x1999, 0x1555, 0x1249, OxOfffj o

After determining the average value of the normalized matrix of the base station, performing fast Fourier transform according to the elements in the average value of the normalized matrix of the base station, and performing frequency offset according to the result of the fast Fourier transform. Correction.

When performing fast Fourier transform on the elements in the average value of the matrix normalized by the base station, first construct a column vector P, where the column vector is a column vector of 512*1, and assign values to different elements in the column vector. , where the result of the assignment is:

P ) = 0

N-k _ ' =° ( 0 < k < N )

^{P k 0} (k > N) where the xth row and the yth column in the matrix are represented, and the unassigned elements of the column vector P are assigned a value of 0 according to the above-mentioned assignment result. Perform fast Fourier transform (FFT) on the assigned column vector P, and perform real part operation on the column vector P after FFT operation, that is, Thelta = real (fft ( P ) ), where Thelta is 512* 1 The column vector, then set the element of Thelta's 128-383 to 0.

After the column vector Thelta is determined, the position corresponding to the maximum value in the column vector Thelta is determined, that is, when [max, Idx]=max (Thelta) is determined to determine the position Idx corresponding to the maximum value in the column vector Thelta, The following algorithm is implemented: If ldx >= 384

Idx = Idx - 512;

Else

Idx = Idx

End

After determining the position corresponding to the maximum value in the column vector Thelta, the normalized frequency offset is determined, wherein the normalized frequency offset D is. The calculation process of the normalized frequency offset D includes: ΑΘ = 2 ^■ (if 512) ■ (Idx - 1)/3 When the normalized frequency offset is determined, all carriers scheduled for the current user can be , phase

= p . υ

Rotate to correct the frequency offset, ie ' · _ '· , where i represents the i-th OFDM symbol,

And in the implementation of the fixed point, it is more difficult to perform the power operation. This complicated operation can be avoided by using the look-up table method. The table creation process is as follows: Since the value of Idx is -128~127. The uplink supports up to 21 symbols, where the symbol 0 does not need to be frequency offset corrected, then the size of the table is 256x20 = 5120B.

For symbol i: calculation

I = cos( (2*3.14*Idx/3/512) * i) * 0x7fff;

Q = -sin( (2*3.14*Idx/3/512) * i) * 0x7fff;

Where Idx is -128~127.

I and Q are stored as high 16bit and low 16bit of the table, respectively.

When performing the frequency offset correction, it is only necessary to find the corresponding frequency offset correction value by Idx, and then multiply each subcarrier by the symbol one by one.

FIG. 3 is a schematic structural diagram of an apparatus for frequency offset correction according to an embodiment of the present invention, where the apparatus includes:

a first normalization processing module 31, configured to receive data on an antenna of the base station, Obtaining an average value of responses of different pilot subcarriers according to responses of all pilot subcarriers of each Tilestrip in each subchannel, constructing a column vector according to the average of the responses of the obtained different pilot subcarriers, and constructing according to the The column vector constructs a matrix, performs scale processing on the constructed matrix, and normalizes the matrix according to the scale processed matrix and the trace of the matrix;

The second normalization processing module 32 is configured to determine, according to each Tilestrip normalized matrix, an average value of the normalized matrix of each antenna, and according to the determined matrix normalized by each antenna The average value of the average of the matrices of the base station;

The frequency offset correction module 33 is configured to perform fast Fourier transform according to the elements in the average value of the matrix normalized by the base station, and perform frequency offset correction according to the result of the fast Fourier transform.

The first normalization processing module 31 is specifically configured to determine, according to the number of OFDM symbols of the pilot subcarriers in the time domain, a sequence number corresponding to an average value of different pilot subcarriers; when different pilot subcarriers are acquired After the average of the responses, the average of the responses obtained is divided into two groups, wherein the average of the pilot subcarrier responses with odd sequence numbers is one group, and the average value of the pilot subcarrier responses with even serial numbers is one. Group, construct a column vector based on each group.

The first normalization processing module 31 is specifically configured to construct two matrices according to the constructed column vector ^ζ ι, ^ζ 2, and

Where 2, , '' and ' are the responses of the different pilot subcarriers; determine the sum of the two matrices ^R l and ^R ₂ constructed, and construct the sum as a matrix of column vectors.

The first normalization processing module 31 is specifically configured to determine an element M having the largest absolute value among the real and imaginary parts of all elements in the matrix, where M is a value greater than zero; determining that the element M is symbolic with respect to 32 bits The number of bits shifted up to the left by S1; according to the determined number of bits S1, the matrix is The real and virtual steps of all elements are shifted to the left by the S1 bit; the high 16 bits of the real and imaginary parts of all elements in the matrix after truncation are taken as the matrix after scale processing, where M is a value greater than zero.

The first normalization processing module 31 is specifically configured to determine a sum of real parts of diagonal elements of the matrix after the scale processing, and use the sum as a trace of the matrix; determine a quotient of the matrix and the trace of the matrix, The quotient is the result of normalization to the matrix.

Specifically, when determining the average value of the normalized matrix of the base station, since each subchannel includes six Tilestrips, each Tilestrip is processed according to the above process, and each is obtained.

The normalized matrix R corresponding to Tilestri and the value of scale(Tileldx) corresponding to the matrix are averaged for each subchannel corresponding to the normalized matrix of all Tilestrips in the subchannel.

Finding a minimum value of a scale (Tileldx) corresponding to the six Tilestrips according to a value of a scale (Tileldx) saved in the normalization process corresponding to each Tilestrip in the subchannel, and the minimum value i is a scaleMin; The normalized matrix R of the Tilestrip corresponding to the minimum value of scale(Tileldx) is not processed, but the real and imaginary parts of all elements in the normalized matrix corresponding to other Tilestrip are shifted to the right. a number, where the number of bits shifted to the right is the difference between the scale (Tileldx) of the other Tilestrip and the scaleMin, and the matrix after the right shift is used as the unified matrix corresponding to the other Tilestri;

Determining, according to the unified matrix corresponding to each Tilestrip, an average value R of the unified matrix of the subchannel, and using the scaleMin as the scale (channelldx) of the subchannel, that is, ⁰ i , ^Λ 'For the unified matrix corresponding to each Tilestrip, the normalized matrix of the Tilestrip corresponding to the scaleMin is the unified matrix corresponding to the Tilestrip.

In the above process, for each subchannel, according to the normalized matrix corresponding to each Tilestrip in the subchannel, the normalized matrix averaging values corresponding to all Tilestrips in the subchannel are determined. And each antenna includes multiple subchannels, according to all determined within each subchannel

The normalized matrix corresponding to Tilestri is averaged to determine the average of the normalized matrix for each antenna. The process of determining the average value of the normalized matrix of each antenna is similar to the process of determining the average value of the normalized matrix of each subchannel, that is, for each subchannel corresponding to the scale (channelldx), The minimum value of scale (channelldx), for the minimum value of the scale (channelldx), the unified matrix of each other subchannel relative to the minimum value of the scale (channelldx), unified for each subchannel A matrix, determining an average of the normalized matrix of the antenna, and taking the minimum value of the scale (channelldx) as the scale (AntIdx) of the antenna.

After determining the average of the normalized matrix of each antenna, the average of the normalized matrix of the base station is determined based on the average of the normalized matrix of each antenna. When determining the average value of the normalized matrix of the base station, determining the base station normalization according to the average value of the normalized matrix of each antenna and the scale (Antldx) of the antenna saved for each antenna The average value of the matrix after the normalization is similar to the above-mentioned method of determining the average value of the matrix normalized by each subchannel and determining the average value of the normalized matrix of each antenna, which will not be described here. .

Specifically, in the embodiment of the present invention, when the average value of the normalized matrix of the base station is determined for each antenna, it is difficult to divide the N-th power of the divisor by a fixed-point number, so the conversion is multiplied by Count down and save the countdown as a table. Since the base station supports up to 8 antennas, the reciprocal table is defined as: Q[8] = {0x7fff, 0x3fff, 0x2aaa, Oxlfff, 0x1999, 0x1555, 0x1249 , OxOfffj o

The device also includes:

a frequency offset estimation module 34, configured to determine a real part of the column vector constructed according to an element in the average value of the normalized matrix, and a column vector constructed by the real part of the column vector fast Fourier transform The position of the maximum value in the constructed column vector; based on the determined position of the maximum value, frequency offset estimation is performed to determine the normalized frequency offset. The frequency offset correction module 33 is specifically configured to perform frequency offset correction according to ^{_ 2} ', where i represents an ith OFDM symbol and D is a normalized frequency offset.

Embodiments of the present invention provide a method and apparatus for frequency offset correction, which constructs a corresponding matrix by using responses of all pilot subcarriers of each Tilestrip in each subchannel, and normalizes the constructed matrix. Determining an average value of the normalized matrix of each antenna, thereby determining an average value of the normalized matrix of each receiving antenna of the base station, and performing fast Fourier according to determining the average value of the normalized matrix of the base station The leaf transforms, thereby performing frequency offset estimation, and performs frequency offset correction based on the result of the frequency offset estimation. Thereby, the estimation and correction of the frequency offset in the OFDM system is realized.

The above description shows and describes a preferred embodiment of the present invention, but as described above, it should be understood that the present invention is not limited to the form disclosed herein, and should not be construed as being Other combinations, modifications, and environments are possible and can be modified by the teachings of the above teachings or related art within the scope of the inventive concept described herein. All changes and modifications made by those skilled in the art are intended to be within the scope of the appended claims.

Claims

1. A method for frequency offset correction, comprising:

2. The method of claim 1, wherein the process of constructing a column vector based on the average of the responses of the different pilot subcarriers obtained comprises:

3. The method of claim 1, wherein the process of constructing the matrix according to the constructed column vector comprises:

among them

, , '' and, '' are responses to different pilot subcarriers; a matrix constructed by combining the sum of the two matrices ^ and ^R 2 as a column vector.

4. The method of claim 3, wherein before the constructing the two matrices, the method further comprises:

Each of the two matrices and ^ is shifted right by one bit.

5. The method of claim 1, wherein the process of scale processing the constructed matrix comprises:

M;

Determining the number of bits of the element M relative to the 32-bit signed number of left shifts S1;

The method of claim 1, wherein the normalizing the matrix after the scale processing comprises:

7. The method according to claim 1, wherein the process of performing frequency offset correction according to the result of the fast Fourier transform comprises:

= p —j A i—l)

The method of claim 7, wherein before the performing the frequency offset correction, the method further comprises: Determining the maximum value of the column vector of the real part structure according to the column vector constructed by the element in the average value of the normalized matrix and the column vector constructed by the real part of the column vector fast Fourier transform Position, based on which frequency offset estimation is performed to determine the normalized frequency offset.

9. A frequency offset correction device, comprising:

a first normalization processing module, configured to obtain, according to a response of all pilot subcarriers of each Tilestrip in each subchannel, an average of responses of different pilot subcarriers for data received on one antenna of the base station According to this, the column vector is constructed, and the matrix is constructed according to the constructed column vector, the constructed matrix is scaled, and the scaled matrix is normalized; the second normalization processing module is used according to each a normalized matrix of Tilestrip, determining an average value of the normalized matrix of each antenna, and determining an average value of the normalized matrix of the base station;

10. The apparatus according to claim 9, wherein the first normalization processing module is configured to: when constructing a column vector according to an average of the responses of the different pilot subcarriers obtained:

11. The apparatus according to claim 9, wherein the first normalization processing module is configured to: when constructing a matrix according to the constructed column vector:

₇ = ( u + y ₄ /

Where '— 2 , , '' and , '' are responses of different pilot subcarriers;

12. The apparatus according to claim 9, wherein the first normalization processing module performs scale processing on the constructed matrix, and is used to:

M;

The apparatus according to claim 9, wherein the first normalization processing module is configured to: when normalizing the scale processed matrix:

The device according to claim 9, wherein the frequency offset correction module performs frequency offset correction according to the result of the fast Fourier transform, and is configured to: perform frequency offset correction according to ^! ' ^-3 , where i represents The i-th OFDM symbol, D is the normalized frequency offset, which is the value of the carrier before the frequency offset correction, and Si is the corrected value.

The device of claim 14, wherein the device further comprises: a frequency offset estimation module, configured to determine a real part structure according to a column vector constructed by an element in an average value of the normalized matrix and a column vector constructed by the real part of the column vector fast Fourier transform The position of the maximum value in the column vector, from which the frequency offset estimation is performed to determine the normalized frequency offset.