CN104980375A - Frequency offset estimation method and apparatus based on differential phase - Google Patents

Abstract

The invention discloses a frequency offset estimation method and apparatus based on a differential phase. The method comprises steps of updating a reception pilot frequency data buffer memory when a pilot symbol is received; extracting a currently-received pilot symbol from the reception pilot frequency data buffer memory to determine an initial frequency offset estimation and to further determine a phase difference value; performing pretreatment of two adjacent pilot symbols stored in the reception pilot frequency data buffer memory and computing corresponding phases of the two adjacent pilot symbols; correcting the corresponding phases of the two adjacent pilot symbols; updating two groups of phase sliding window buffer memories if the corrected phases are determined to be normal; calculating an average phase value according to phase values in the two groups of phase sliding window buffer memories to further determine the maximum likelihood frequency offset estimation; and updating a frequency offset estimation result sliding window buffer memory according to the maximum likelihood frequency offset estimation, and performing smooth filtering of the buffered maximum likelihood frequency offset estimation to obtain the current maximum likelihood frequency offset estimation of a current moment. The invention is advantaged by being precise in frequency offset estimation and small in estimation fluctuation range.

Description

A kind of frequency deviation estimating method based on differential phase and device

Technical field

The present invention relates to the communications field, particularly relate to a kind of MIMO-OFDM communication system and receive synchronously based on the frequency deviation estimating method of differential phase and device.

Background technology

For any digital communication system, be synchronously prerequisite and the important guarantee of reliable data transmission.The quality of net synchronization capability will directly affect the performance of whole communication system.But in a wireless communication system, due to the frequency difference between sending ending equipment and receiving device, and ustomer premises access equipment moves the impacts such as brought Doppler frequency-shift, makes to there is frequency deviation between the frequency of carrier frequency and local crystal oscillator.In order to ensure the transmitting of data, must carry out accurately estimating and being compensated to the frequency deviation of signal.In addition 3GPP(3rdGeneration Partnership Project, third generation partner program) LTE((Long TermEvolution, Long Term Evolution) descendingly have employed the higher OFDM(OrthogonalFrequency Division Multiplexing of the availability of frequency spectrum, orthogonal frequency division multiplexi) modulation technique, OFDM symbol is superposed by multiple sub-carrier signal to be formed, utilize the orthogonality between subcarrier to carry out demodulation at receiving terminal, the orthogonality thus between ofdm system sub-carrier proposes strict requirement.In actual transmissions, owing to not mating the frequency departure brought between Doppler frequency shift and transceiver local oscillator, the orthogonality between ofdm system subcarrier can be destroyed, cause inter-sub-carrier interference (ICI).Timing slip then can cause intersymbol interference (ISI), reduces the validity of Cyclic Prefix (CP).Therefore, synchronously very important for ofdm system.Around MIMO(multiple-input and multiple-output) frequency offset estimation technique under-OFDM or OFDM condition, expand a large amount of research work both at home and abroad, and achieved a lot of achievement in research, have the practical algorithm that can adopt in a large number, these achievements in research become the important leverage promoting and improve frequency deviation estimated performance.

Frequency synchronization method in ofdm system roughly can be divided into blind synchronized algorithm and synchronized algorithm two class based on training sequence.Blind synchronized algorithm mainly make use of Cyclic Prefix character specific to ofdm system to complete synchronous estimation.J.-J.van de Beek, M.Sandell, the people such as M.Isaksson document " MLestimation of Time and Frequency Offset in OFDM Systems; " IEEETrans.Commun., vol.43, propose a kind of maximum likelihood synchronized algorithm based on Cyclic Prefix in pp.761-766, August1997, this algorithm can go out timing slip and carrier frequency shift by Combined estimator.Ronghong Mo and Yong Huat Chew is at document " A New Blind Joint Timing andFrequency Offset Estimator for OFDM Systems Over Multipath FadingChannels, " IEEE Trans.Veh.Technol., vol.57, no.5, pp.2947-2957, the blind synchronization method of Combined estimator timing slip and carrier frequency shift under a kind of multidiameter fading channel is proposed in September2008, the time domain OFDM sampled point received is divided into the mutual incoherent several subset of sampling point by this algorithm, the advantage of this synchronizer is in high SNR(signal to noise ratio) there is reasonable performance in situation, but its shortcoming is that its performance in low SNR situation is not good.Wen Long Chin and Sau Gee Chen is at document " A Blind Synchronizer For OFDM Systems Based on SINR Maximization inMultipath Fading channels, " IEEE Trans.Veh.Technol., vol.58, no.2, pp.625-635, propose a kind of based on letter interference ratio (SINR) maximized blind synchronized algorithm in February2009, synchronous error introduces while ISI and ICI cause system loss, the SINR receiving data also declines thereupon, this algorithm utilizes this characteristic, by maximizing SINR metric Function Estimation timing offset and carrier frequency offset, emulation proves that this algorithm has good performance than traditional synchronized algorithm, but the shortcoming of this algorithm is that the implementation complexity of its algorithm is higher.

With blind synchronized algorithm unlike, based on training sequence synchronized algorithm need utilize insert known pilot symbols realize synchronously.T.M.Schmidl and Donald.C.Cox document " Robust Frequencyand Timing Synchronization for OFDM; " IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, propose in December1997 and utilize first training symbol to realize Timing Synchronization and fraction frequency offset estimation, utilize the differential relationship of former and later two training sequences to realize integer-times frequency offset and estimate.But there is timing platform problem in this algorithm, have impact on net synchronization capability.P.H.Moose is at document " A technique for orthogonal frequency division multiplexing frequency offsetcorrection, " IEEE Trans.Commun., vol.42, pp.2908-2914, propose the identical OFDM symbol of transmission two in October1994 to estimate to simplify frequency deviation, but its frequency offset estimation range is smaller, in addition because frequency deviation estimates it is complete at frequency domain, so need FFT(Fast Fourier Transformation, fast Fourier transform) conversion realize estimate at frequency domain, so the implementation complexity of algorithm is higher.

Be widely used all kinds of training sequence in view of comprising in the practical communication system such as LTE standard, the frequency synchronization algorithm therefore around pilot aided often more has actual using value in actual applications.Under the precondition of receiving terminal known training sequence, the frequency deviation of maximum likelihood is estimated often to obtain better frequency deviation estimated performance, and becomes the emphasis of all kinds of research institute concern.Wu Yan, Samir Attallah and J.W.M.Bergmans gives the MIMO-OFDM system frequency deviation estimation technique scheme of a class based on adjustable quadrature training sequence in document " Efficient Training Sequence for Joint Carrier Frequency Offsetand Channel Estimation for MIMO-OFDM Systems, " Proc of IEEE ICC2007.The great advantage of this frequency excursion algorithm makes full use of Naoki Suhiro and Mitsutoshi Hatori at document " Modulatable Orthogonal Sequences and Their Application to SSAMSystems, " IEEE Trans.Information Theory, vol.34, no.1, pp.93-100, the orthogonal property of the adjustable orthogonal sequence proposed in January1988, by inserting two groups of identical training sequences in time domain, maximum likelihood frequency deviation estimation problem is reduced to the related operation problem of the two adjacent groups frequency pilot sign received, have and calculate simply, frequency deviation estimated performance is excellent, time domain training symbol PAPR(PeaktoAverage Power Ratio, papr) be the advantages such as constant.But the shortcoming of this algorithm is to estimate channel simultaneously, need to increase training sequence length.If number of transmit antennas is Nt, channel path number is L, and the training sequence length required for system is NtL, and in addition, system postulation inserts pilot tone in time domain, and it is inconsistent that the frequency domain that this and LTE standard suppose inserts pilot tone.

In order to effectively support the Combined estimator of frequency deviation under MIMO-OFDM or ofdm system condition and channel, Tao Cui and Chintha Tellambura is at document Joint Frequency Offset and ChannelEstimation for OFDM Systems Using Pilot Symbols and Virtual Carriers, IEEETrans.Wireless Commun., Vol.6, No.4, the Joint iteration proposed in 2007 based on decision-aided (Decision-Directed) frequency deviation of pilot tone and virtual subnet carrier wave, channel and data symbol is estimated and detection algorithm.S.Salari, M.Ardebilipour and M.Ahmadian is at document Jointmaximum-likelihood frequency offset and channel estimation for multiple-inputmultiple-output – orthogon frequency division multiplexing systems, IETCommun., 2008, vol.2, No.8, propose based on EM(Expectation-maximization algorithm in pp.1069-1076, EM algorithm) frequency deviation and channel joint estimation method, on the basis of channel and frequency deviation combined likelihood function, channel estimating is upgraded by E step, upgrade frequency deviation by M step to estimate.Although the frequency deviation of iteration and channel Combined estimator algorithm are associating frequency deviation under MIMO-OFDM system condition and channel estimating provide a kind of effective method, the iterative computation complexity involved by iterative computation structure still has influence on the computational efficiency of synchronized algorithm to a great extent.In order to avoid the computation complexity that iterative computation is introduced, Yao-Jen Liang and Jin-FuChang is at document Non-iterative Joint Frequency Offset and Channel Estimation forMIMO OFDM Systems Using Cascaded Orthogonal Pilots, IEEE Trans.VehTechn., vol.59, no.8, give a class in Oct.2010 and be applicable to MIMO-OFDM system based on the non-iterative frequency deviation of cascade orthogonal pilot frequency sequence and channel joint Estimation scheme, by making full use of " cascade is orthogonal " character of training sequence, the frequency deviation simplified under MIMO-OFDM system condition is estimated and channel estimating.

Under MIMO-OFDM system or ofdm system condition, frequency deviation is estimated or frequency deviation and channel Combined estimator, also has many patent of invention achievements both at home and abroad.The comprehensive analysis at present domestic and international frequency deviation around under MIMO-OFDM and ofdm system condition is estimated or the achievement of frequency deviation and the research of channel joint Estimation, on the basis of given frequency deviation and channel joint likelihood function, existing a large amount of practicable frequency deviation and channel estimation methods can for using for reference at present, and these frequency deviations and channel estimation method angularly propose solution feasible in a large number from pilot frequency sequence design, OFDM symbol specificity analysis and utilization, Pilot OFDM symbols design, concrete frequency excursion algorithm.Due to internal relation existing between frequency deviation and adjacent pilot symbols phase place, existing many patents give effectively based on frequency offset estimation technique and the method for differential phase.But carefully analyze these correlation techniques, we are also not difficult to find, also seldom there is the maximum likelihood frequency deviation estimating method based on differential phase around MIMO-OFDM system, particularly almost do not have around the maximum likelihood frequency deviation estimating method be similar to based on differential phase under the specific pilot frequency format condition of the communication systems such as LTE standard.

Summary of the invention

The technical problem to be solved in the present invention is to provide a kind of frequency deviation estimating method based on differential phase and device, estimates with the maximum likelihood frequency deviation realized based on differential phase.

In order to solve the problems of the technologies described above, the invention provides a kind of frequency deviation estimating method based on differential phase, comprising:

When receiving frequency pilot sign, upgrade and receive pilot data buffer memory;

From described reception pilot data buffer memory, extract the frequency pilot sign of current reception, determine that initial frequency deviation is estimated, estimate to determine phase difference value according to described initial frequency deviation;

Preliminary treatment is carried out to adjacent two frequency pilot signs stored in described reception pilot data buffer memory, calculates the phase place that described two frequency pilot signs are corresponding respectively;

Estimate according to described initial frequency deviation, phase place that described phase difference value is corresponding to described two frequency pilot signs revises;

If the phase place after decision revision is normal value, then respectively the sliding window buffer memory of two groups of phase places is upgraded;

Average phase value is calculated, according to described average phase value determination maximum likelihood frequency offset estimation respectively according to phase value in the sliding window buffer memory of described two groups of phase places;

Upgrade the sliding window buffer memory of frequency offset estimation result according to described maximum likelihood frequency offset estimation, the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory is obtained to the maximum likelihood frequency offset estimation of current time.

Further, said method also has feature below: described renewal receives pilot data buffer memory and is achieved in the following ways:

$R_{-1}^{\left(n\right)}=R_0^{\left(n\right)},R_0^{\left(n\right)}=R^{\left(n\right)},$ Wherein,

for the frequency pilot sign that buffer memory current time receives a moment on buffer memory the frequency pilot sign of middle buffer memory;

$R^{\left(n\right)}={{[r_1^{\left(n\right)T},...,r_{N_R}^{\left(n\right)T}]}^T}_{(N\·N_R)\×1},r_q^{\left(n\right)}=\underset{p=1}{\overset{N_T}{\mathrm{\Σ}}}{\tilde{E}}^{\left(n\right)}{\tilde{F}}^H{D}_p^{\left(n\right)}\tilde{W}{g}_{q,p}^{\left(n\right)}+n_q,$ represent the n-th OFDM (OFDM) the time-domain symbol block that q root reception antenna receives, N _tfor the number of transmit antennas of multiple-input and multiple-output (MIMO) system, N _rfor the reception antenna number of mimo system,

represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding, represent carrier wave frequency deviation matrix, N is OFDM sub-carrier number,

ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time,

$D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R}$ For the pilot symbol transmitted of correspondence,

$D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends;

for Discrete Fourier transform, wherein (l, m) individual element is ${\tilde{F}}_{l,m}=\frac{1}{\sqrt{N}}{e}^{-j(2\mathrm{\π}(l-1)(m-1)/N)};$

represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},\tilde{V}\∈C^{N\×(N-L)},$ And have ${\tilde{W}}^H\tilde{V}=0,\tilde{W}{\tilde{W}}^H+\tilde{V}{\tilde{V}}^H=I.$

Further, said method also has feature below: describedly determine that initial frequency deviation is estimated be achieved in the following ways:

${\mathrm{\ϵ}}_0^{\left(n\right)}=\arg \underset{\mathrm{\ϵ}}{\max }\left[\mathrm{tr}\left(R^{\left(n\right)H}{E}^{\left(n\right)}{C}^{\left(n\right)H}{C}^{\left(n\right)}{E}^{\left(n\right)H}{R}^{\left(n\right)}\right)\right],$ Wherein,

$F=(I_{N_T{N}_R}\⊗\tilde{F}),W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R},$

Describedly to estimate according to described initial frequency deviation determine that phase difference value is realized by mode below:

$\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N},$

Wherein, L _prepresent adjacent pilot symbols gap length.

Further, said method also has feature below: to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, calculate the phase place angle1 that described two frequency pilot signs are corresponding respectively ⁽ⁿ⁾, angle2 ⁽ⁿ⁾, comprising:

To adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, obtain two revision measuring-signals with

If two adjacent frequency domain frequency pilot signs that transmitting terminal sends are identical, then

${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}\right)$

If the pilot tone symbol that transmitting terminal sends is the pilot tone pattern of Long Term Evolution prescribed by standard, then

${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0{B}^{*}\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}B\right)$

Wherein, B=vec (A), B ^*=vec (A ^h), vec (A _{n × N}) represent and get diagonal matrix A _{n × N}the row vector of element composition, A=diag [e ^{j α 0}..., e ^{j α (N-1)}], α=π.

Further, said method also has feature below: describedly to estimate according to described initial frequency deviation, phase place that described phase difference value is corresponding to described two frequency pilot signs revises, comprising:

The folding number of turns times of phase place is calculated according to following formula:

$\mathrm{times}=\mathrm{round}(\frac{L_p}{N}\·{\mathrm{\ϵ}}_0^{\left(n\right)})$

Wherein, round (.) represents the computing that rounds up;

Calculate angle1 by the following method ⁽ⁿ⁾, angle2 ⁽ⁿ⁾three groups of phase candidate correction values, if wherein initial frequency deviation estimate for on the occasion of, when calculated candidate correction value select addition, otherwise select subtraction:

${\mathrm{angle}1}_1^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}1}_2^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}1}_3^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π}$

${\mathrm{angle}2}_1^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}2}_2^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}2}_3^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π}$

Three groups of phase difference values are respectively calculated as follows according to three groups of candidate phases values:

${\mathrm{\Δ}}_i^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)}-{\mathrm{angle}2}_i^{\left(n\right)},i=\mathrm{1,2,3}$

Be calculated as follows absolute difference D _i:

$D_i=|{\mathrm{\Δ}}_i^{\left(n\right)}-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|$

Confirm the revision phase value that corresponding absolute difference is minimum as the phase value of final revision, even:

${\mathrm{angle}1}^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)},{\mathrm{angle}2}^{\left(n\right)}={\mathrm{angle}2}_i^{\left(n\right)}.$

Further, said method also has feature below: when judging to meet following formula, then the phase place after decision revision is normal value:

wherein, δ is the adjustable decision threshold judging that whether phase place is abnormal.

Further, said method also has feature below: described respectively according to phase value calculating average phase value ave_angle1 in the sliding window buffer memory of described two groups of phase places ⁽ⁿ⁾, ave_angle2 ⁽ⁿ⁾realized by following formula:

${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}2(n-k),$ Wherein M is the designated length of the sliding window buffer memory of described phase place;

Described according to described average phase value determination maximum likelihood frequency offset estimation ε ⁽ⁿ⁾realized by following formula:

${\mathrm{\ϵ}}^{\left(n\right)}=\frac{N\·[{\mathrm{ang}\_\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}.$

Further, said method also has feature below: the smoothing filtering process of the described maximum likelihood frequency offset estimation to buffer memory obtains the maximum likelihood frequency offset estimation of current time realized by following formula:

wherein, P is the designated length of the sliding window buffer memory of described frequency offset estimation result.

In order to solve the problem, present invention also offers the device that a kind of frequency deviation based on differential phase is estimated, wherein, comprising:

Update module, during for receiving frequency pilot sign, upgrading and receiving pilot data buffer memory;

First determination module, for extracting the frequency pilot sign of current reception from described reception pilot data buffer memory, determines that initial frequency deviation is estimated, estimates to determine phase difference value according to described initial frequency deviation;

Pretreatment module, for carrying out preliminary treatment to adjacent two frequency pilot signs stored in described reception pilot data buffer memory, calculates the phase place that described two frequency pilot signs are corresponding respectively;

Correcting module, revises for the phase place estimated according to described initial frequency deviation, described phase difference value is corresponding to described two frequency pilot signs;

Judging module is normal value for the phase place after such as decision revision, then upgrade the sliding window buffer memorys of two groups of phase places respectively;

Second determination module, for calculating average phase value, according to described average phase value determination maximum likelihood frequency offset estimation according to phase value in the sliding window buffer memory of described two groups of phase places respectively;

Processing module, for upgrading the sliding window buffer memory of frequency offset estimation result according to described maximum likelihood frequency offset estimation, obtains the maximum likelihood frequency offset estimation of current time to the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory.

Further, said apparatus also has feature below:

Described update module, upgrades reception pilot data buffer memory and is achieved in the following ways: wherein, for the frequency pilot sign R that current time described in buffer memory receives ⁽ⁿ⁾, a moment on buffer memory the frequency pilot sign of middle buffer memory, represent the n-th OFDM (OFDM) the time-domain symbol block that q root reception antenna receives, N _tfor the number of transmit antennas of multiple-input and multiple-output (MIMO) system, N _rfor the reception antenna number of mimo system, represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding, represent carrier wave frequency deviation matrix, N is OFDM sub-carrier number, ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time, $D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R},$ For the pilot symbol transmitted of correspondence, $D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends; for Discrete Fourier transform, wherein (l, m) individual element is ${\tilde{F}}_{l,m}=\frac{1}{\sqrt{N}}{e}^{-j(2\mathrm{\π}(l-1)(m-1)/N)};$ represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},$ $\tilde{V}\∈C^{N\×(N-L)},$ And have ${\tilde{W}}^H\tilde{V}=0,\tilde{W}{\tilde{W}}^H+\tilde{V}{\tilde{V}}^H=I.$

Further, said apparatus also has feature below:

Described first determination module, determines that initial frequency deviation is estimated be achieved in the following ways: ${\mathrm{\ϵ}}_0^{\left(n\right)}=\arg \underset{\mathrm{\ϵ}}{\max }\left[\mathrm{tr}\left(R^{\left(n\right)H}{E}^{\left(n\right)}{C}^{\left(n\right)H}{C}^{\left(n\right)}{E}^{\left(n\right)H}{R}^{\left(n\right)}\right)\right],$ Wherein, $F=(I_{N_T{N}_R}\⊗\tilde{F}),$ $W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R},$ describedly to estimate according to described initial frequency deviation determine that phase difference value is realized by mode below: wherein, L _prepresent adjacent pilot symbols gap length.

Further, said apparatus also has feature below:

Described pretreatment module, to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, calculate the phase place angle1 that described two frequency pilot signs are corresponding respectively ⁽ⁿ⁾, angle2 ⁽ⁿ⁾, comprising: to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, obtain two revision measuring-signals if two adjacent frequency domain frequency pilot signs that transmitting terminal sends are identical, then ${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}\right);$ If the pilot tone symbol that transmitting terminal sends is the pilot tone pattern of Long Term Evolution prescribed by standard, then ${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0{B}^{*}\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}B\right),$ Wherein, B=vec (A), B ^*=vec (A ^h), vec (A _{n × N}) represent and get diagonal matrix A _{n × N}the row vector of element composition, A=diag [e ^{j α 0}..., e ^{j α (N-1)}], α=π.

Further, said apparatus also has feature below:

Described correcting module, specifically for calculating the folding number of turns times of phase place according to following formula: wherein, round (.) represents the computing that rounds up; Calculate angle1 by the following method ⁽ⁿ⁾, angle2 ⁽ⁿ⁾three groups of phase candidate correction values, if wherein initial frequency deviation estimate for on the occasion of, when calculated candidate correction value select addition, otherwise select subtraction:

${\mathrm{angle}1}_1^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π},{\mathrm{angle}2}_1^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}1}_2^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π},{\mathrm{angle}2}_2^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}1}_3^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π},{\mathrm{angle}2}_3^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π},$

Three groups of phase difference values are respectively calculated as follows according to three groups of candidate phases values: ${\mathrm{\Δ}}_i^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)}-{\mathrm{angle}2}_i^{\left(n\right)},i=\mathrm{1,2,3},$ Be calculated as follows absolute difference D _i: $D_i=|{\mathrm{\Δ}}_i^{\left(n\right)}-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|,$ Confirm the revision phase value that corresponding absolute difference is minimum as the phase value of final revision, even: ${\mathrm{angle}1}^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)},{\mathrm{angle}2}^{\left(n\right)}={\mathrm{angle}2}_i^{\left(n\right)}.$

Further, said apparatus also has feature below:

Described judging module, when judging to meet following formula, then the phase place after decision revision is normal value: wherein, δ is the adjustable decision threshold judging that whether phase place is abnormal.

Further, said apparatus also has feature below:

Described second determination module, calculates average phase value ave_angle1 according to phase value in the sliding window buffer memory of described two groups of phase places respectively ⁽ⁿ⁾, ave_angle2 ⁽ⁿ⁾realized by following formula: ${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}2(n-k),$ Wherein M is the designated length of the sliding window buffer memory of described phase place; According to described average phase value determination maximum likelihood frequency offset estimation ε ⁽ⁿ⁾realized by following formula: ${\mathrm{\ϵ}}^{\left(n\right)}=\frac{N\·[{\mathrm{ang}\_\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}.$

Further, said apparatus also has feature below:

Described processing module, obtains the maximum likelihood frequency offset estimation of current time to the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory realized by following formula: wherein, P is the designated length of the sliding window buffer memory of described frequency offset estimation result.

To sum up, the invention provides a kind of frequency deviation estimating method based on differential phase and device, on the basis analyzing adjacent pilot frequencies phase difference and frequency deviation relation, give a kind of enhancing maximum likelihood frequency deviation based on differential phase to estimate, effectively to meet the requirement of the real systems such as LTE to high-performance frequency offset estimation technique solution.

Accompanying drawing explanation

Fig. 1 is the flow chart of MIMO-OFDM system based on the frequency deviation estimating method of differential phase of the embodiment of the present invention;

Fig. 2 is the LTE Uplink MIMO SC-FDMA system sending and receiving end schematic diagram of the embodiment of the present invention;

Fig. 3 is the LTE descending MIMO ofdm system sending and receiving end schematic block diagram of the embodiment of the present invention;

Fig. 4 is the schematic diagram of the phase value sliding window updating of the embodiment of the present invention;

Fig. 5 is the schematic diagram of the frequency offset estimation result Sliding window data buffer update of the embodiment of the present invention;

Fig. 6 be pilot tone identical time different frequency excursion algorithm MSE(Mean Square Error, mean square error) Performance comparision figure;

Fig. 7 be pilot tone identical time different frequency offset algorithm CDF Performance comparision figure;

Fig. 8 be pilot tone identical time different frequency excursion algorithm MSE Performance comparision figure;

Fig. 9 be pilot tone identical time different frequency excursion algorithm CDF Performance comparision figure;

Figure 10 is different frequency excursion algorithm MSE Performance comparision figure under LTE standard;

Figure 11 is different frequency offset algorithm CDF Performance comparision figure under LTE standard;

Figure 12 is different frequency excursion algorithm MSE Performance comparision figure under LTE standard;

Figure 13 is different frequency offset algorithm CDF Performance comparision figure under LTE standard;

Figure 14 is the schematic diagram of the device that the frequency deviation based on differential phase of the embodiment of the present invention is estimated.

Embodiment

For making the object, technical solutions and advantages of the present invention clearly understand, hereinafter will be described in detail to embodiments of the invention by reference to the accompanying drawings.It should be noted that, when not conflicting, the embodiment in the application and the feature in embodiment can combination in any mutually.

Fig. 1 is the flow chart of MIMO-OFDM system based on the frequency deviation estimating method of differential phase of the embodiment of the present invention, as shown in Figure 1, comprises the following steps:

Step 11, when receiving frequency pilot sign, upgrade and receive pilot data buffer memory;

Receiving terminal, first when receiving frequency pilot sign R (n), adopting following methods to upgrade and receiving pilot data buffer memory:

$R_{-1}^{\left(n\right)}=R_0^{\left(n\right)},R_0^{\left(n\right)}=R^{\left(n\right)}$

Wherein, for the frequency pilot sign that buffer memory current time receives a moment on buffer memory the frequency pilot sign of middle buffer memory.

$R^{\left(n\right)}={{[r_1^{\left(n\right)T},...,r_{N_R}^{\left(n\right)T}]}^T}_{(N\·N_R)\×1},r_q^{\left(n\right)}=\underset{p=1}{\overset{N_T}{\mathrm{\Σ}}}{\tilde{E}}^{\left(n\right)}{\tilde{F}}^H{D}_p^{\left(n\right)}\tilde{W}{g}_{q,p}^{\left(n\right)}+n_q,$ represent the n-th OFDM time-domain symbol block that q root reception antenna receives, N _tfor the number of transmit antennas of mimo system, N _rfor the reception antenna number of mimo system;

represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding;

represent carrier wave frequency deviation (CFO) matrix, N is OFDM sub-carrier number;

ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time;

$D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R}$ For the pilot symbol transmitted of correspondence;

$D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends;

for DFT(Discrete Fourier Transform, discrete Fourier transform) matrix, wherein (l, m) individual element is ${\tilde{F}}_{l,m}=\frac{1}{\sqrt{N}}{e}^{-j(2\mathrm{\π}(l-1)(m-1)/N)};$

represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},\tilde{V}\∈C^{N\×(N-L)},$ And have ${\tilde{W}}^H\tilde{V}=0,\tilde{W}{\tilde{W}}^H+\tilde{V}{\tilde{V}}^H=I.$

Step 12, from described reception pilot data buffer memory, extract the frequency pilot sign of current reception, determine that initial frequency deviation is estimated, estimate to determine phase difference value according to described initial frequency deviation;

From receiving the frequency pilot sign extracting current reception pilot data buffer memory calculate by the following method and determine frequency deviation initial estimation:

${\mathrm{\ϵ}}_0^{\left(n\right)}=\arg \underset{\mathrm{\ϵ}}{\max }\left[\mathrm{tr}\left(R^{\left(n\right)H}{E}^{\left(n\right)}{C}^{\left(n\right)H}{C}^{\left(n\right)}{E}^{\left(n\right)H}{R}^{\left(n\right)}\right)\right],$ Wherein,

$F=(I_{N_T{N}_R}\⊗\tilde{F}),W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R}.$

Then evaluation phase difference as follows on the basis that initial frequency deviation is estimated:

$\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N},$ Wherein,

L _prepresent adjacent pilot symbols gap length.

Step 13, preliminary treatment is carried out to adjacent two frequency pilot signs stored in described reception pilot data buffer memory, calculate the phase place that described two frequency pilot signs are corresponding respectively;

According to receiving adjacent two frequency pilot signs stored in pilot data buffer memory calculate two revision measuring-signals with and calculate phase place angle1 respectively according to two whether identical employing following methods of adjacent frequency domain frequency pilot sign that transmitting terminal sends ⁽ⁿ⁾, angle2 ⁽ⁿ⁾:

If two adjacent frequency domain frequency pilot signs that A transmitting terminal sends are identical:

${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}\right)$

If the pilot tone symbol that B transmitting terminal sends is the pilot tone pattern of LTE standard defined:

${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0{B}^{*}\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}B\right)$

Wherein, B=vec (A), B ^*=vec (A ^h), vec (A _{n × N}) represent and get diagonal matrix A _{n × N}the row vector of element composition, A=diag [e ^{j α 0}..., e ^{j α (N-1)}], α=π.

Step 14, estimate according to described initial frequency deviation, phase place that described phase difference value is corresponding to described two frequency pilot signs revises;

Estimate in initial frequency deviation basis on, calculate the folding number of turns of phase place according to following formula:

$\mathrm{times}=\mathrm{round}(\frac{L_p}{N}\·{\mathrm{\ϵ}}_0^{\left(n\right)})$

Wherein, round (.) represents the computing that rounds up.Calculate angle1 by the following method ⁽ⁿ⁾, angle2 ⁽ⁿ⁾three groups of phase candidate correction values, if wherein initial frequency deviation estimate for on the occasion of, when calculated candidate correction value select addition, otherwise select subtraction:

${\mathrm{angle}1}_1^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}1}_2^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}1}_3^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π}$

${\mathrm{angle}2}_1^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}2}_2^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}2}_3^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π}$

Three groups of phase difference values are respectively calculated as follows according to three groups of candidate phases values:

${\mathrm{\Δ}}_i^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)}-{\mathrm{angle}2}_i^{\left(n\right)},i=\mathrm{1,2,3}$

Be calculated as follows absolute difference:

$D_i=|{\mathrm{\Δ}}_i^{\left(n\right)}-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|$

The revision phase value that corresponding absolute difference is minimum namely as the phase value that final revision confirms, for subsequent calculations, even:

${\mathrm{angle}1}^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)},{\mathrm{angle}2}^{\left(n\right)}={\mathrm{angle}2}_i^{\left(n\right)}.$

Step 15, judgement abnormal phase;

Calculate in data prediction, phase difference calculating step and phase only pupil filter calculation procedure and determine phase place angle1 ⁽ⁿ⁾, angle2 ⁽ⁿ⁾basis on, judge whether current phase place is exceptional value according to following relation, wherein δ is the adjustable decision threshold judging that phase place is whether abnormal.

(1) if $|({\mathrm{angle}1}^{\left(n\right)}-{\mathrm{angle}2}^{\left(n\right)})-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|\≤\mathrm{\δ},$ Two phase values of current calculating are normal value;

(2) if $|({\mathrm{angle}1}^{\left(n\right)}-{\mathrm{angle}2}^{\left(n\right)})-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|>\mathrm{\δ},$ One is had at least for exceptional value in two phase values of current calculating.

Step 16, renewal phase data buffer memory;

According to the judgement conclusion that abnormal phase decision steps draws, determine two the phase value angle1 whether calculated by data prediction and phase calculation step ⁽ⁿ⁾, angle2 ⁽ⁿ⁾upgrade the sliding window buffer of two groups of phase places as shown in Figure 4.

(1) if two phase value angle1 calculating of the decision data preliminary treatment of abnormal phase decision steps and phase calculation step ⁽ⁿ⁾, angle2 ⁽ⁿ⁾not exceptional value, then upgrade the sliding window buffer of two groups of phase places by the following method:

angle1(n-M+i)=angle1(n-M+i+1)

angle2(n-M+i)=angle2(n-M+i+1),i=1,…,M

angle1(n)=angle1 ⁽ⁿ⁾,angle2(n)=angle2 ⁽ⁿ⁾

Wherein, M is the designated length of the sliding window buffer memory of described phase place.

(2) if two phase value angle1 calculating of the decision data preliminary treatment of abnormal phase decision steps and phase calculation step ⁽ⁿ⁾, angle2 ⁽ⁿ⁾in have at least one for exceptional value time, then keep buffer memory phase value in the sliding window buffers of two groups of phase places constant.

Step 17, calculate average phase value, according to described average phase value determination maximum likelihood frequency offset estimation according to phase value in the sliding window buffer memory of described two groups of phase places respectively;

The phase value of buffer memory from two groups of phase places sliding window buffer, calculates corresponding average phase value by the following method:

${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}2(n-k)$

Wherein, M is the designated length of the sliding window buffer memory of described phase place.

On the basis of average phase value result of calculation, calculate by the following method and determine the maximum likelihood frequency offset estimation based on average phase-difference:

${\mathrm{\ϵ}}^{\left(n\right)}=\frac{N\·[{\mathrm{ang}\_\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}.$

Step 18, upgrade the sliding window buffer memory of frequency offset estimation result according to described maximum likelihood frequency offset estimation, the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory is obtained to the maximum likelihood frequency offset estimation of current time;

By the frequency offset estimation ε that the maximum likelihood appraising frequency bias step estimation based on average phase-difference obtains ⁽ⁿ⁾length by the following method stored in correspondence is in the sliding window memory of frequency offset estimation result of the first-in first-out (FIFO) of P, as shown in Figure 5.

Then P the frequency offset estimation that the frequency offset estimation result being P according to length slides buffer memory in window memory calculates the final frequency offset estimation determining current time by the following method:

${\stackrel{\‾}{\mathrm{\ϵ}}}_{\mathrm{ML}}^{\left(n\right)}=\frac{1}{P}\underset{i=0}{\overset{P-1}{\mathrm{\Σ}}}{\mathrm{\ϵ}}^{(n-i)}.$

The embodiment of the present invention provide a kind of be applicable to MIMO-OFDM system acceptance synchronous in based on the frequency deviation estimating method of differential phase, the method, under the condition not revising transmitting terminal frequency pilot sign, pilot placement, significantly promotes and improves frequency deviation estimated performance.

Embodiment one

Below in conjunction with TD-LTE substandard Uplink MIMO SC-FDMA system and descending MIMO-OFDM system, specific embodiment of the invention step is described in detail.Be TD-LTE standard Uplink MIMO SC-FDMA(Single-carrier Frequency-DivisionMultiple Access respectively shown in Fig. 2 and Fig. 3, single-carrier frequency division multiple access) system and descending MIMO-OFDM system sending and receiving end schematic block diagram.Under the condition of hypothesis receiving terminal sign synchronization, for Uplink MIMO SC-FDMA system and descending MIMO-OFDM system, although LTE uplink and downlink number of pilot symbols, the regularity of distribution in each subframe, uplink and downlink transmitting terminal number of antennas is not quite similar, but the time-domain received signal characteristic that reception antenna is removed after Cyclic Prefix is identical, in order to simplified characterization, for MIMO SC-FDMA system, suppose N _t, N _rbe respectively transmitting antenna and the reception antenna number of mimo system, the sub-carrier number that N adopts for system.

Under the condition of system receiving terminal sign synchronization, the time-domain expression of the n-th symbol that q root reception antenna receives is:

$r_q^{\left(n\right)}=\underset{p=1}{\overset{N_T}{\mathrm{\Σ}}}{\tilde{E}}^{\left(n\right)}{\tilde{F}}^H{D}_p^{\left(n\right)}\tilde{W}{g}_{q,p}^{\left(n\right)}+n_q,---\left(1\right)$

Wherein, represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding, represent carrier wave frequency deviation (CFO) matrix, ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time;

for DFT matrix, wherein (l, m) individual element is

$D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends;

represent the L footpath time-domain channel gain of corresponding n-th frequency pilot sign between (q, p) antenna pair.

represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},\tilde{V}\∈C^{N\×(N-L)},$ And have n _qrepresent that zero-mean is tieed up in N × 1 that q root reception antenna receives, every one dimension variance is multiple Gaussian noise.Order:

$G^{\left(n\right)}={[g_1^{\left(n\right)T},g_2^{\left(n\right)T},...,g_{N_T}^{\left(n\right)T}]}_{L{N}_T{N}_R\×1}^T,g_q^{\left(n\right)}={[g_{1,q}^{\left(n\right)T},...,g_{N_R,q}^{\left(n\right)T}]}_{(L\·N_R)\×,}^T,$

$N=F{[n_1^T,...,n_{N_R}^T]}^T,F=(I_{N_T{N}_R}\⊗\tilde{F}),W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R},$

$E^{\left(n\right)}=\mathrm{blkdiag}[{\tilde{E}}_1^{\left(n\right)},...,{\tilde{E}}_{N_T{N}_R}^{\left(n\right)}],$

$D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R},R^{\left(n\right)}={{[r_1^{\left(n\right)T},...,r_{N_R}^{\left(n\right)T}]}^T}_{(N\·N_R)\×1},$

Then N _rthe matrix form that the Received signal strength of root reception antenna can be expressed as:

R ^（n)=E ⁽ⁿ⁾F ^HD ⁽ⁿ⁾WG ⁽ⁿ⁾+N (2)

Wherein, G ⁽ⁿ⁾represent time-domain channel gain, theoretical according to maximal possibility estimation, adopt the analytical method similar with mimo system, it be easy to show that ML channel estimator and frequency deviation are estimated to meet following relation:

Wherein represent pseudoinverse.

If ignore the change of the channel that adjacent pilot frequencies experiences, the time-domain received signal expression formula that can obtain adjacent two pilot tones according to formula (2) is respectively:

R ₁=EF ^HD ₁WG+N ₁(3)

$R_2=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{\mathrm{EF}}^H{D}_2\mathrm{WG}+N_2---\left(4\right)$

Wherein, L _prepresent adjacent pilot symbols gap length.Can be obtained by (3) and (4):

$R_2^H{R}_1=e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D_2}^H{D}_1\mathrm{WG}+\tilde{N}---\left(2\right)$

Theoretical according to maximal possibility estimation, be not difficult to obtain following channel and frequency deviation estimation likelihood function:

$\mathrm{\Λ}\left(R_2^H{R}_1\right|G,\mathrm{\ϵ})=\frac{1}{{\left(\mathrm{\π}{\mathrm{\σ}}_n^2\right)}^{N\×N_R}}\exp \{-\frac{1}{{\mathrm{\σ}}_n^2}\mathrm{tr}\left[\begin{array}{c}{(R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})}^H\\ (R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})\end{array}\·\right]\}---\left(6\right)$

Corresponding log-domain frequency deviation and channel estimating likelihood function are:

$\begin{array}{c}L\left(R_2^H{R}_1\right|G,\mathrm{\ϵ})=-\mathrm{tr}\left[{(R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})}^H(R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})\right]\\ =-\mathrm{tr}\left[\begin{array}{c}{R}_1^H{R}_2{R}_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG}-e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_1^H{D}_2\mathrm{WG}{R}_2^H{R}_1\\ +G^H{W}^H{D}_1^H{D}_2\mathrm{WG}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG}\end{array}\right]\end{array}---\left(7\right)$

Order $\frac{L\left(R_2^H{R}_1\right|G,\mathrm{\ϵ})}{\∂\mathrm{\ϵ}}=0$ Can obtain:

$e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG}=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_1{G}^H{W}^H{D}_1^H{D}_2\mathrm{WG}---\left(8\right)$

In order to be estimated by (8) formula maximum likelihood frequency deviation simplified of deriving further, we make a concrete analysis of the relativeness of the 4th and the 11st Pilot OFDM symbols in a subframe in LTE standard.According to LTE standard, in same subframe, between two frequency pilot signs, there is following relation:

D ₂=AD ₁，A=diag[e ^jα0,…,e ^jα(N-1)],α=π (9)

Formula (9) is substituted into (8) Shi Ke get:

$e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2{G}^H{W}^H{D}_1^H{A^{H}D}_1\mathrm{WG}=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_1{G}^H{W}^H{D}_1^{H}A{D}_2\mathrm{WG}---\left(10\right)$

Make vectorial B _{1 × N}=(diag (A _{n × N})) ^t, B ^*=(diag (A ^h)) ^t, wherein diag (A _{n × N}) represent and get diagonal matrix A _{n × N}diagonal element composition column vector, then formula (10) can be written as further:

$\begin{array}{c}{e}^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2\left[B^{*}\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)\right]D_1\mathrm{WG}=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_1[B\·\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)]D_1\mathrm{WG}\\ \⇒e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2{B}^{*}\·\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)D_1\mathrm{WG}=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_{1}B\·\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)D_1\mathrm{WG}\end{array}$

Obviously about fall the common factor of both members, can obtain:

$R_1^H{R}_2{B}^{*}=e^{\frac{j4\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_{1}B---\left(11\right)$

Here it is emphasized that to adjacent pilot symbols according to (5) formula process obtain on the basis of revision frequency pilot sign, use the maximum likelihood frequency deviation that obtains of maximal possibility estimation theory to estimate that restriction relation is irrelevant with channel estimating.In fact, this adopts (5) formula to carry out pretreated mainspring place to reception frequency pilot sign just.On the basis of (11) formula, the closed solutions showing that following maximum likelihood frequency deviation is estimated of easily deriving:

Here it is pointed out that with not not scalar but row vector, corresponding calculate as follows:

Order then have:

${\widehat{\mathrm{\ϵ}}}^{\left(n\right)}=\frac{N\·[{\mathrm{angle}1}^{\left(n\right)}-{\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}---\left(13\right)$

In order to improve frequency deviation estimated performance further, under the constant or slow scene change of frequency deviation, can consider to improve frequency deviation estimated performance to dependent phase in (13) formula after being averaged.Here can consider that adopting the average phase method based on sliding window to calculate upgrades phase value, if sliding window length is M, then corresponding average phase can calculate by the following method to be determined:

${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}{\mathrm{angle}2}^{(n-k)}---\left(14\right)$

Corresponding (13) formula can be rewritten as:

${\widetilde{\mathrm{\ϵ}}}^{\left(n\right)}=\frac{N\·[\mathrm{ave}\_{\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}---\left(15\right)$

Frequency deviation due to the overwhelming majority estimates that Output rusults is all similar to Gaussian distributed, can consider the dynamic fluctuation adopting smothing filtering, reduced frequency deviation estimation by time diversity further for this reason.Here we can adopt the smothing filtering improvement based on sliding window to estimate based on the maximum likelihood frequency deviation of differential phase equally, if sliding window length is P, then the final expression formula estimated based on the enhancing maximum likelihood frequency deviation of differential phase is

${\stackrel{\‾}{\mathrm{\ϵ}}}_{\mathrm{ML}}^{\left(n\right)}=\frac{1}{P}\underset{i=0}{\overset{P-1}{\mathrm{\Σ}}}{\widetilde{\mathrm{\ϵ}}}^{(n-i)}\frac{N\·\underset{i=0}{\overset{P-1}{\mathrm{\Σ}}}[\mathrm{ave}\_{\mathrm{angle}1}^{(n-i)}-{\mathrm{ave}\_\mathrm{angle}2}^{(n-i)}]}{4\mathrm{\π}{L}_{p}P}---\left(16\right)$

Embodiment two:

Similar with LTE system in embodiment one, here we are for the frequency offset estimation technique embodiment of general MIMO-OFDM System Discussion based on differential phase, with embodiment one difference be, we suppose that the pilot tone that MIMO-OFDM system transmitting terminal sends is identical, then two groups of adjacent reception frequency pilot signs are:

R ₁=EF ^HDWG+N ₁(1)

$R_2=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{\mathrm{EF}}^H{D}_2\mathrm{WG}+N_2---\left(2\right)$

Wherein, L _prepresent adjacent pilot symbols gap length.Can be obtained by (1) and (2)

$R_2^H{R}_1=e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D_2}^H{D}_1\mathrm{WG}+\tilde{N}---\left(3\right)$

Theoretical according to maximal possibility estimation, be not difficult to obtain following channel and frequency deviation estimation likelihood function:

$\mathrm{\Λ}\left(R_2^H{R}_1\right|G,\mathrm{\ϵ})=\frac{1}{{\left(\mathrm{\π}{\mathrm{\σ}}_n^2\right)}^{N\×N_R}}\exp \{-\frac{1}{{\mathrm{\σ}}_n^2}\mathrm{tr}\left[\begin{array}{c}{(R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})}^H\\ (R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})\end{array}\·\right]\}---\left(4\right)$

Corresponding log-domain frequency deviation and channel estimating likelihood function are:

$\begin{array}{c}L\left(R_2^H{R}_1\right|G,\mathrm{\ϵ})=-\mathrm{tr}\left[{(R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})}^H(R_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG})\right]\\ =-\mathrm{tr}\left[\begin{array}{c}{R}_1^H{R}_2{R}_2^H{R}_1-e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG}-e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{G}^H{W}^H{D}_1^H{D}_2\mathrm{WG}{R}_2^H{R}_1\\ +G^H{W}^H{D}_1^H{D}_2\mathrm{WG}{G}^H{W}^H{D}_2^H{D}_1\mathrm{WG}\end{array}\right]\end{array}---\left(5\right)$

Order $\frac{L\left(R_2^H{R}_1\right|G,\mathrm{\ϵ})}{\∂\mathrm{\ϵ}}=0$ Can obtain:

$\begin{array}{c}{e}^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2\left[B^{*}\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)\right]D_1\mathrm{WG}=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_1[B\·\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)]D_1\mathrm{WG}\\ \⇒e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2{B}^{*}\·\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)D_1\mathrm{WG}=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_{1}B\·\mathrm{diag}\left(G^H{W}^H{D}_1^H\right)D_1\mathrm{WG}\end{array}$

There is GHWHDHDWG on formula (6) both sides, and it is scalar, and namely it can directly about fall by both members.Then formula (6) can abbreviation be further:

$e^{\frac{-j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_1^H{R}_2=e^{\frac{j2\mathrm{\π\ϵ}{L}_p}{N}}{R}_2^H{R}_1---\left(7\right)$

Here it is emphasized that the revision measuring-signal of adjacent pilot symbols through obtaining such as formula the method process of (3), the maximum likelihood frequency deviation then using maximal possibility estimation theory to obtain on its basis again estimates that restriction relation and channel estimating have nothing to do! In fact, this just project team consider to adopt the method to revise the mainspring place of measuring-signal.Be not difficult to obtain by formula (7) closed solutions that following maximum likelihood frequency deviation estimates:

$\widehat{\mathrm{\ϵ}}=\frac{N\·[arg\left(R_1^H{R}_2\right)\text{-arg}\left(R_2^H{R}_1\right)]}{4\mathrm{\π}{L}_p}---\left(8\right)$

Wherein arg () expression is got dependent variable phase place.Order

${\widehat{\mathrm{\ϵ}}}^{\left(n\right)}=\frac{N\·[arg\left(R_1^{\left(n\right)H}{R}_2^{\left(n\right)}\right)-\arg \left(R_2^{\left(n\right)H}{R}_1^{\left(n\right)}\right)]}{4\mathrm{\π}{L}_p}\left]\frac{N\·[{\mathrm{angle}1}^{\left(n\right)}-{\mathrm{angle}2}^{\left(n\right)}}{4\mathrm{\π}{L}_p}\right]---\left(9\right)$

In order to improve frequency deviation estimated performance further, under the constant or slow conversion scene of frequency deviation, can consider to improve frequency deviation estimated performance to dependent phase in (9) formula after being averaged.Here can consider that adopting the average phase method based on sliding window to calculate upgrades phase value, if sliding window length is M, then corresponding average phase can calculate by the following method to be determined:

${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}{\mathrm{angle}2}^{(n-k)}---\left(11\right)$

Corresponding (9) formula can be rewritten as:

${\widetilde{\mathrm{\ϵ}}}^{\left(n\right)}=\frac{N\·[\mathrm{ave}\_{\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}---\left(12\right)$

Frequency deviation due to the overwhelming majority estimates that Output rusults is all similar to Gaussian distributed, and we can be considered employing smothing filtering, be reduced the dynamic fluctuation of frequency deviation estimation by time diversity further for this reason.Here we can adopt the smothing filtering improvement based on sliding window to estimate based on the maximum likelihood frequency deviation of differential phase equally, if sliding window length is P, then the final expression formula estimated based on the enhancing maximum likelihood frequency deviation of differential phase is:

${\stackrel{\‾}{\mathrm{\ϵ}}}_{\mathrm{ML}}^{\left(n\right)}=\frac{1}{P}\underset{i=0}{\overset{P-1}{\mathrm{\Σ}}}{\widetilde{\mathrm{\ϵ}}}^{(n-i)}\frac{N\·\underset{i=0}{\overset{P-1}{\mathrm{\Σ}}}[\mathrm{ave}\_{\mathrm{angle}1}^{(n-i)}-{\mathrm{ave}\_\mathrm{angle}2}^{(n-i)}]}{4\mathrm{\π}{L}_{p}P}---\left(13\right)$

In order to verify the maximum likelihood frequency offset estimation technique scheme performance based on differential phase, we consider to fix enhancing maximum likelihood frequency deviation estimated performance based on differential phase under Frequency Offset by relevant parameter simulating, verifying in table 1.

Simulation parameter under Frequency Offset fixed by table 1

(1), all pilot tones are all identical

When pilot resources is 20RB and 2RB, and all pilot tones all identical when, enhancing maximum likelihood frequency deviation based on differential phase estimates that frequency deviation under ETU300Hz channel condition estimates mean square error performance and CDF(Cumulative Distribution Function, cumulative distribution function) results of property is as shown in Fig. 6-Fig. 9.From relevant simulation result, the enhancing maximum likelihood frequency deviation based on differential phase under fixing Frequency Offset under two class channels estimates that (marking with PD in figure) performance is obviously better than traditional maximum likelihood frequency deviation and estimates.Such as, under ETU300Hz channel condition, adopt the enhancing maximum likelihood frequency deviation based on differential phase to estimate, the frequency deviation that can obtain 8 × 10-7 at-5dB estimates mean square deviation performance.Although along with the increase of signal to noise ratio is less based on the enhancing maximum likelihood frequency deviation estimated performance improvement of differential phase, the enhancing maximum likelihood frequency deviation high-performance frequency deviation be estimated as under severe channel conditions obviously based on differential phase is estimated to provide a kind of effective implementation.

(2), the pilot tone that specifies of LTE standard

When pilot resources is 20RB and 2RB, pilot tone is under LTE standard regulation pilot conditions, and the enhancing maximum likelihood frequency deviation based on differential phase estimates that the frequency deviation under ETU300Hz channel condition estimates that mean square error results of property is as shown in Figure 10-Figure 13.From relevant simulation result, the enhancing maximum likelihood frequency deviation based on differential phase under fixing Frequency Offset under two class channels estimates that (PD) performance is equally obviously better than traditional maximum likelihood frequency excursion algorithm.Such as, under ETU300Hz channel condition, when 15dB, traditional maximum likelihood frequency excursion algorithm also can only obtain 1 × 10 ^-5frequency deviation estimate mean square deviation performance, and adopt PD to estimate can obtain close to 6 × 10 when 15dB ^-7frequency deviation estimate mean square deviation performance, obviously, adopt the enhancing maximum likelihood frequency deviation based on differential phase to estimate to provide satisfied frequency deviation estimated performance under LTE framework.

Compared with prior art, the invention has the beneficial effects as follows:

One, innovative point: with conventional maximum likelihood frequency excursion algorithm directly from compared with the maximal possibility estimation receiving pilot signal R (n) and derive frequency deviation, main innovate point based on the enhancing maximum likelihood frequency excursion algorithm of differential phase is, it is from adjacent reception pilot signal that relevant maximum likelihood frequency deviation is estimated construct the two groups of revisions obtained and receive frequency pilot sign (under LTE standard be ) and (under LTE standard be ) phase difference in derive maximum likelihood frequency deviation estimate.

Two, in estimating based on the maximum likelihood frequency deviation revising measuring-signal phase difference, the present invention adopts the method for sliding window to carry out effective the disposal of gentle filter effectively to improve frequency deviation estimated performance to involved differential phase and middle frequency offset estimation.

Three, frequency deviation estimating method has calculating simply, the advantage that frequency offset estimation scope is large, frequency deviation estimated performance is good.

Figure 14 is the schematic diagram of the device that the frequency deviation based on differential phase of the embodiment of the present invention is estimated, as shown in figure 14, the device of the present embodiment comprises:

Update module, during for receiving frequency pilot sign, upgrading and receiving pilot data buffer memory;

First determination module, for extracting the frequency pilot sign of current reception from described reception pilot data buffer memory, determines that initial frequency deviation is estimated, estimates to determine phase difference value according to described initial frequency deviation;

Pretreatment module, for carrying out preliminary treatment to adjacent two frequency pilot signs stored in described reception pilot data buffer memory, calculates the phase place that described two frequency pilot signs are corresponding respectively;

Correcting module, revises for the phase place estimated according to described initial frequency deviation, described phase difference value is corresponding to described two frequency pilot signs;

Judging module is normal value for the phase place after such as decision revision, then upgrade the sliding window buffer memorys of two groups of phase places respectively;

Second determination module, for calculating average phase value, according to described average phase value determination maximum likelihood frequency offset estimation according to phase value in the sliding window buffer memory of described two groups of phase places respectively;

Processing module, for upgrading the sliding window buffer memory of frequency offset estimation result according to described maximum likelihood frequency offset estimation, obtains the maximum likelihood frequency offset estimation of current time to the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory.

Wherein, described update module, upgrading reception pilot data buffer memory can realize in the following manner: wherein, for the frequency pilot sign that current time described in buffer memory receives a moment on buffer memory the frequency pilot sign of middle buffer memory, $R^{\left(n\right)}={{[r_1^{\left(n\right)T},...,r_{N_R}^{\left(n\right)T}]}^T}_{(N\·N_R)\×1},r_q^{\left(n\right)}=\underset{p=1}{\overset{N_T}{\mathrm{\Σ}}}{\tilde{E}}^{\left(n\right)}{\tilde{F}}^H{D}_p^{\left(n\right)}\tilde{W}{g}_{q,p}^{\left(n\right)}+n_q,$ represent the n-th OFDM (OFDM) the time-domain symbol block that q root reception antenna receives, N _tfor the number of transmit antennas of multiple-input and multiple-output (MIMO) system, N _rfor the reception antenna number of mimo system, represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding, represent carrier wave frequency deviation matrix, N is OFDM sub-carrier number, ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time, $D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R},$ For the pilot symbol transmitted of correspondence, $D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends; for Discrete Fourier transform, wherein (l, m) individual element is ${\tilde{F}}_{l,m}=\frac{1}{\sqrt{N}}{e}^{-j(2\mathrm{\π}(l-1)(m-1)/N)};$ represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},$ $\tilde{V}\∈C^{N\×(N-L)},$ And have ${\tilde{W}}^H\tilde{V}=0,\tilde{W}{\tilde{W}}^H+\tilde{V}{\tilde{V}}^H=I.$

Wherein, described first determination module, determines that initial frequency deviation is estimated can be achieved in the following ways: ${\mathrm{\ϵ}}_0^{\left(n\right)}=\arg \underset{\mathrm{\ϵ}}{\max }\left[\mathrm{tr}\left(R^{\left(n\right)H}{E}^{\left(n\right)}{C}^{\left(n\right)H}{C}^{\left(n\right)}{E}^{\left(n\right)H}{R}^{\left(n\right)}\right)\right],$ Wherein, $F=(I_{N_T{N}_R}\⊗\tilde{F}),$ $W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R},$ describedly to estimate according to described initial frequency deviation determine that phase difference value is realized by mode below: wherein, L _prepresent adjacent pilot symbols gap length.

Wherein, described pretreatment module, to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, calculate the phase place angle1 that described two frequency pilot signs are corresponding respectively ⁽ⁿ⁾, angle2 ⁽ⁿ⁾, can comprise: to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, obtain two revision measuring-signals with if two adjacent frequency domain frequency pilot signs that transmitting terminal sends are identical, then ${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}\right);$ If the pilot tone symbol that transmitting terminal sends is the pilot tone pattern of Long Term Evolution prescribed by standard, then ${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0{B}^{*}\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}B\right),$ Wherein, B=vec (A), B ^*=vec (A ^h), vec (A _{n × N}) represent and get diagonal matrix A _{n × N}the row vector of element composition, A=diag [e ^{j α 0}..., e ^{j α (N-1)}], α=π.

Wherein, described correcting module, specifically may be used for calculating the folding number of turns times of phase place according to following formula: wherein, round (.) represents the computing that rounds up; Calculate angle1 by the following method ⁽ⁿ⁾, angle2 ⁽ⁿ⁾three groups of phase candidate correction values, if wherein initial frequency deviation estimate for on the occasion of, when calculated candidate correction value select addition, otherwise select subtraction:

${\mathrm{angle}1}_1^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π},{\mathrm{angle}2}_1^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}1}_2^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π},{\mathrm{angle}2}_2^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}1}_3^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π},{\mathrm{angle}2}_3^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π},$

Three groups of phase difference values are respectively calculated as follows according to three groups of candidate phases values: ${\mathrm{\Δ}}_i^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)}-{\mathrm{angle}2}_i^{\left(n\right)},i=\mathrm{1,2,3},$ Be calculated as follows absolute difference D _i: $D_i=|{\mathrm{\Δ}}_i^{\left(n\right)}-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|,$ Confirm the revision phase value that corresponding absolute difference is minimum as the phase value of final revision, even: ${\mathrm{angle}1}^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)},{\mathrm{angle}2}^{\left(n\right)}={\mathrm{angle}2}_i^{\left(n\right)}.$

Wherein, described judging module, when judging to meet following formula, then the phase place after decision revision is normal value: wherein, δ is the adjustable decision threshold judging that whether phase place is abnormal.

Wherein, described second determination module, can calculate average phase value ave_angle1 according to phase value in the sliding window buffer memory of described two groups of phase places respectively ⁽ⁿ⁾, ave_angle2 ⁽ⁿ⁾realized by following formula: ${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}2(n-k);$ According to described average phase value determination maximum likelihood frequency offset estimation ε ⁽ⁿ⁾realized by following formula:

${\mathrm{\ϵ}}^{\left(n\right)}=\frac{N\·[{\mathrm{ang}\_\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}.$

Wherein, described processing module, obtains the maximum likelihood frequency offset estimation of current time to the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory can be realized by following formula: wherein, P is the designated length of the sliding window buffer memory of described frequency offset estimation result.

The all or part of step that one of ordinary skill in the art will appreciate that in said method is carried out instruction related hardware by program and is completed, and described program can be stored in computer-readable recording medium, as read-only memory, disk or CD etc.Alternatively, all or part of step of above-described embodiment also can use one or more integrated circuit to realize.Correspondingly, each module/unit in above-described embodiment can adopt the form of hardware to realize, and the form of software function module also can be adopted to realize.The present invention is not restricted to the combination of the hardware and software of any particular form.

These are only the preferred embodiments of the present invention; certainly; the present invention also can have other various embodiments; when not deviating from the present invention's spirit and essence thereof; those of ordinary skill in the art are when making various corresponding change and distortion according to the present invention, but these change accordingly and are out of shape the protection range that all should belong to the claim appended by the present invention.

Claims (16)

1., based on a frequency deviation estimating method for differential phase, comprising:

When receiving frequency pilot sign, upgrade and receive pilot data buffer memory;

From described reception pilot data buffer memory, extract the frequency pilot sign of current reception, determine that initial frequency deviation is estimated, estimate to determine phase difference value according to described initial frequency deviation;

Preliminary treatment is carried out to adjacent two frequency pilot signs stored in described reception pilot data buffer memory, calculates the phase place that described two frequency pilot signs are corresponding respectively;

Estimate according to described initial frequency deviation, phase place that described phase difference value is corresponding to described two frequency pilot signs revises;

If the phase place after decision revision is normal value, then respectively the sliding window buffer memory of two groups of phase places is upgraded;

Average phase value is calculated, according to described average phase value determination maximum likelihood frequency offset estimation respectively according to phase value in the sliding window buffer memory of described two groups of phase places;

Upgrade the sliding window buffer memory of frequency offset estimation result according to described maximum likelihood frequency offset estimation, the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory is obtained to the maximum likelihood frequency offset estimation of current time.

2. the method for claim 1, is characterized in that: described renewal receives pilot data buffer memory and is achieved in the following ways:

$R_{-1}^{\left(n\right)}=R_0^{\left(n\right)},R_0^{\left(n\right)}=R^{\left(n\right)},$ Wherein,

for the frequency pilot sign R that buffer memory current time receives ⁽ⁿ⁾, a moment on buffer memory the frequency pilot sign of middle buffer memory;

$R^{\left(n\right)}={{[r_1^{\left(n\right)T},...,r_{N_R}^{\left(n\right)T}]}^T}_{(N\·N_R)\×1},r_q^{\left(n\right)}=\underset{p=1}{\overset{N_T}{\mathrm{\Σ}}}{\tilde{E}}^{\left(n\right)}{\tilde{F}}^H{D}_p^{\left(n\right)}\tilde{W}{g}_{q,p}^{\left(n\right)}+n_q,$ represent the n-th OFDM (OFDM) the time-domain symbol block that q root reception antenna receives, N _tfor the number of transmit antennas of multiple-input and multiple-output (MIMO) system, N _rfor the reception antenna number of mimo system,

represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding, represent carrier wave frequency deviation matrix, N is OFDM sub-carrier number,

ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time,

$D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R}$ For the pilot symbol transmitted of correspondence,

$D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends;

for Discrete Fourier transform, wherein (l, m) individual element is ${\tilde{F}}_{l,m}=\frac{1}{\sqrt{N}}{e}^{-j(2\mathrm{\π}(l-1)(m-1)/N)};$

represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},\tilde{V}\∈C^{N\×(N-L)},$ And have ${\tilde{W}}^H\tilde{V}=0,\tilde{W}{\tilde{W}}^H+\tilde{V}{\tilde{V}}^H=I.$

3. method as claimed in claim 2, is characterized in that: describedly determine that initial frequency deviation is estimated be achieved in the following ways:

${\mathrm{\ϵ}}_0^{\left(n\right)}=\arg \underset{\mathrm{\ϵ}}{\max }\left[\mathrm{tr}\left(R^{\left(n\right)H}{E}^{\left(n\right)}{C}^{\left(n\right)H}{C}^{\left(n\right)}{E}^{\left(n\right)H}{R}^{\left(n\right)}\right)\right],$ Wherein,

$F=(I_{N_T{N}_R}\⊗\tilde{F}),W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R},$

Describedly to estimate according to described initial frequency deviation determine that phase difference value is realized by mode below:

$\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N},$

Wherein, L _prepresent adjacent pilot symbols gap length.

4. method as claimed in claim 3, is characterized in that: to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, calculate the phase place angle1 that described two frequency pilot signs are corresponding respectively ⁽ⁿ⁾, angle2 ⁽ⁿ⁾, comprising:

To adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, obtain two revision measuring-signals with

If two adjacent frequency domain frequency pilot signs that transmitting terminal sends are identical, then

${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}\right)$

If the pilot tone symbol that transmitting terminal sends is the pilot tone pattern of Long Term Evolution prescribed by standard, then

${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0{B}^{*}\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}B\right)$

Wherein, B=vec (A), B ^*=vec (A ^h), vec (A _{n × N}) represent and get diagonal matrix A _{n × N}the row vector of element composition, A=diag [e ^{j α 0}..., e ^{j α (N-1)}], α=π.

5. method as claimed in claim 4, is characterized in that: describedly to estimate according to described initial frequency deviation, phase place that described phase difference value is corresponding to described two frequency pilot signs revises, comprising:

The folding number of turns times of phase place is calculated according to following formula:

$\mathrm{times}=\mathrm{round}(\frac{L_p}{N}\·{\mathrm{\ϵ}}_0^{\left(n\right)})$

Wherein, round (.) represents the computing that rounds up;

Calculate angle1 by the following method ⁽ⁿ⁾, angle2 ⁽ⁿ⁾three groups of phase candidate correction values, if wherein initial frequency deviation estimate for on the occasion of, when calculated candidate correction value select addition, otherwise select subtraction:

${\mathrm{angle}1}_1^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}1}_2^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}1}_3^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π}$

${\mathrm{angle}2}_1^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}2}_2^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}2}_3^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π}$

Three groups of phase difference values are respectively calculated as follows according to three groups of candidate phases values:

${\mathrm{\Δ}}_i^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)}-{\mathrm{angle}2}_i^{\left(n\right)},i=\mathrm{1,2,3}$

Be calculated as follows absolute difference D _i:

$D_i=|{\mathrm{\Δ}}_i^{\left(n\right)}-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|$

Confirm the revision phase value that corresponding absolute difference is minimum as the phase value of final revision, even:

${\mathrm{angle}1}^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)},{\mathrm{angle}2}^{\left(n\right)}={\mathrm{angle}2}_i^{\left(n\right)}.$

6. method as claimed in claim 5, is characterized in that: when judging to meet following formula, then the phase place after decision revision is normal value:

wherein, δ is the adjustable decision threshold judging that whether phase place is abnormal.

7. method as claimed in claim 6, is characterized in that: described respectively according to phase value calculating average phase value ave_angle1 in the sliding window buffer memory of described two groups of phase places ⁽ⁿ⁾, ave_angle2 ⁽ⁿ⁾realized by following formula:

${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}2(n-k),$ Wherein M is the designated length of the sliding window buffer memory of described phase place;

Describedly to be realized by following formula according to described average phase value determination maximum likelihood frequency offset estimation ε (n):

${\mathrm{\ϵ}}^{\left(n\right)}=\frac{N\·[{\mathrm{ang}\_\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}.$

8. method as claimed in claim 7, is characterized in that: the smoothing filtering process of the described maximum likelihood frequency offset estimation to buffer memory obtains the maximum likelihood frequency offset estimation of current time realized by following formula:

wherein, P is the designated length of the sliding window buffer memory of described frequency offset estimation result.

9., based on the device that the frequency deviation of differential phase is estimated, it is characterized in that, comprising:

Update module, during for receiving frequency pilot sign, upgrading and receiving pilot data buffer memory;

First determination module, for extracting the frequency pilot sign of current reception from described reception pilot data buffer memory, determines that initial frequency deviation is estimated, estimates to determine phase difference value according to described initial frequency deviation;

Pretreatment module, for carrying out preliminary treatment to adjacent two frequency pilot signs stored in described reception pilot data buffer memory, calculates the phase place that described two frequency pilot signs are corresponding respectively;

Correcting module, revises for the phase place estimated according to described initial frequency deviation, described phase difference value is corresponding to described two frequency pilot signs;

Judging module is normal value for the phase place after such as decision revision, then upgrade the sliding window buffer memorys of two groups of phase places respectively;

Second determination module, for calculating average phase value, according to described average phase value determination maximum likelihood frequency offset estimation according to phase value in the sliding window buffer memory of described two groups of phase places respectively;

Processing module, for upgrading the sliding window buffer memory of frequency offset estimation result according to described maximum likelihood frequency offset estimation, obtains the maximum likelihood frequency offset estimation of current time to the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory.

10. device as claimed in claim 9, is characterized in that:

Described update module, upgrades reception pilot data buffer memory and is achieved in the following ways: wherein, for the frequency pilot sign R that current time described in buffer memory receives ⁽ⁿ⁾, a moment on buffer memory the frequency pilot sign of middle buffer memory, represent the n-th OFDM (OFDM) the time-domain symbol block that q root reception antenna receives, N _tfor the number of transmit antennas of multiple-input and multiple-output (MIMO) system, N _rfor the reception antenna number of mimo system, represent the frequency deviation matrix that the n-th frequency pilot sign is corresponding, represent carrier wave frequency deviation matrix, N is OFDM sub-carrier number, ε is normalization frequency deviation value, L _nrepresent the sequence number corresponding to the n-th frequency pilot sign first sampling time, $D^{\left(n\right)}=\frac{1}{\sqrt{N_T}}\mathrm{blkdiag}{[D_1^{\left(n\right)},...,D_{N_T}^{\left(n\right)},...,D_{N_T{N}_R}^{\left(n\right)}]}_{N{N}_T{N}_R\×N{N}_T{N}_R}$ For the pilot symbol transmitted of correspondence, $D_p^{\left(n\right)}=\mathrm{diag}\left(d_p^{\left(n\right)}\right)=\mathrm{aiag}\left(\right[d_{p,0}^{\left(n\right)},...,d_{p,N-1}^{\left(n\right)}{]}^T)$ Represent the n-th pilot tone sign matrix that p root transmitting antenna sends; for Discrete Fourier transform, wherein (l, m) individual element is ${\tilde{F}}_{l,m}=\frac{1}{\sqrt{N}}{e}^{-j(2\mathrm{\π}(l-1)(m-1)/N)};$ represent DFT matrix front L row, namely $\tilde{F}=\left[\tilde{W}\right|\tilde{V}],\tilde{W}\∈C^{N\×L},$ $\tilde{V}\∈C^{N\×(N-L)},$ And have ${\tilde{W}}^H\tilde{V}=0,\tilde{W}{\tilde{W}}^H+\tilde{V}{\tilde{V}}^H=I.$

11. devices as claimed in claim 10, is characterized in that:

Described first determination module, determines that initial frequency deviation is estimated be achieved in the following ways: ${\mathrm{\ϵ}}_0^{\left(n\right)}=\arg \underset{\mathrm{\ϵ}}{\max }\left[\mathrm{tr}\left(R^{\left(n\right)H}{E}^{\left(n\right)}{C}^{\left(n\right)H}{C}^{\left(n\right)}{E}^{\left(n\right)H}{R}^{\left(n\right)}\right)\right],$ Wherein, $F=(I_{N_T{N}_R}\⊗\tilde{F}),$ $W={(I_{N_T{N}_R}\⊗\tilde{W})}_{N{N}_T{N}_R\×L{N}_T{N}_R},$ describedly to estimate according to described initial frequency deviation determine that phase difference value is realized by mode below: wherein, L _prepresent adjacent pilot symbols gap length.

12. devices as claimed in claim 11, is characterized in that:

Described pretreatment module, to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, calculate the phase place angle1 that described two frequency pilot signs are corresponding respectively ⁽ⁿ⁾, angle2 (n), comprising: to adjacent two frequency pilot signs stored in described reception pilot data buffer memory carry out preliminary treatment, obtain two revision measuring-signals with if two adjacent frequency domain frequency pilot signs that transmitting terminal sends are identical, then ${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}\right);$ If the pilot tone symbol that transmitting terminal sends is the pilot tone pattern of Long Term Evolution prescribed by standard, then ${\mathrm{angle}1}^{\left(n\right)}=\arg \left(R_{-1}^{\left(n\right)H}{R}_0{B}^{*}\right),{\mathrm{angle}2}^{\left(n\right)}=\arg \left(R_0^{\left(n\right)H}{R}_{-1}B\right),$ Wherein, B=vec (A), B ^*=vec (A ^h), vec (A _{n × N}) represent and get diagonal matrix A _{n × N}the row vector of element composition, A=diag [e ^{j α 0}..., e ^{j α (N-1)}], α=π.

13. devices as claimed in claim 12, is characterized in that:

Described correcting module, specifically for calculating the folding number of turns times of phase place according to following formula: wherein, round (.) represents the computing that rounds up; Calculate angle1 by the following method ⁽ⁿ⁾, angle2 ⁽ⁿ⁾three groups of phase candidate correction values, if wherein initial frequency deviation estimate for on the occasion of, when calculated candidate correction value select addition, otherwise select subtraction:

${\mathrm{angle}1}_1^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π},{\mathrm{angle}2}_1^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±\mathrm{times}\×2\mathrm{\π}$

${\mathrm{angle}1}_2^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π},{\mathrm{angle}2}_2^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}+1)\×2\mathrm{\π}$

${\mathrm{angle}1}_3^{\left(n\right)}={\mathrm{angle}1}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π},{\mathrm{angle}2}_3^{\left(n\right)}={\mathrm{angle}2}^{\left(n\right)}\±(\mathrm{times}-1)\×2\mathrm{\π},$

Three groups of phase difference values are respectively calculated as follows according to three groups of candidate phases values: ${\mathrm{\Δ}}_i^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)}-{\mathrm{angle}2}_i^{\left(n\right)},i=\mathrm{1,2,3},$ Be calculated as follows absolute difference D _i: $D_i=|{\mathrm{\Δ}}_i^{\left(n\right)}-\frac{4\mathrm{\π}{\mathrm{\ϵ}}_0^{\left(n\right)}{L}_p}{N}|,$ Confirm the revision phase value that corresponding absolute difference is minimum as the phase value of final revision, even: ${\mathrm{angle}1}^{\left(n\right)}={\mathrm{angle}1}_i^{\left(n\right)},{\mathrm{angle}2}^{\left(n\right)}={\mathrm{angle}2}_i^{\left(n\right)}.$

14. devices as claimed in claim 13, is characterized in that:

Described judging module, when judging to meet following formula, then the phase place after decision revision is normal value: wherein, δ is the adjustable decision threshold judging that whether phase place is abnormal.

15. devices as claimed in claim 14, is characterized in that:

Described second determination module, calculates average phase value ave_angle1 according to phase value in the sliding window buffer memory of described two groups of phase places respectively ⁽ⁿ⁾, ave_angle2 ⁽ⁿ⁾realized by following formula: ${\mathrm{ave}\_\mathrm{angle}1}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}1(n-k),{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}=\frac{1}{M}\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}\mathrm{angle}2(n-k),$ Wherein M is the designated length of the sliding window buffer memory of described phase place; Realized by following formula according to described average phase value determination maximum likelihood frequency offset estimation ε (n): ${\mathrm{\ϵ}}^{\left(n\right)}=\frac{N\·[{\mathrm{ang}\_\mathrm{angle}1}^{\left(n\right)}-{\mathrm{ave}\_\mathrm{angle}2}^{\left(n\right)}]}{4\mathrm{\π}{L}_p}.$

16. devices as claimed in claim 15, is characterized in that:

Described processing module, obtains the maximum likelihood frequency offset estimation of current time to the smoothing filtering process of the maximum likelihood frequency offset estimation of buffer memory realized by following formula: wherein, P is the designated length of the sliding window buffer memory of described frequency offset estimation result.