Blind frequency-offset estimating method based on maximum likelihood two-dimensional search
Technical field
The invention belongs to communication technical field, it is related to multiple-input and multiple-output space-time block code MIMO-STBC technologies, more enters
One step is related to a kind of blind frequency-offset estimating method based on maximum likelihood two-dimensional search.The present invention can be used for signal of communication blind recognition
Blind offset estimation is carried out to MIMO-STBC signals in scape.
Background technology
Next-Generation Wireless Communication Systems will provide transmission rate and service quality higher, in the background that frequency spectrum resource is nervous
Under, MIMO-STBC technologies can provide technical support for the realization of this target., it is necessary to right in signal of communication blind recognition scene
Non-cooperative communication signal carries out blind estimate and identification, and blind offset estimation is one of key technology of Signal blind recognition.
Existing MIMO-STBC frequency excursion algorithms are broadly divided into two classes:Data-aided frequency excursion algorithm and blind frequency
Deviation estimation algorithm.Data-aided frequency excursion algorithm includes the algorithm based on training sequence and the algorithm based on pilot tone, this kind of
The complexity of method is relatively low, and estimated accuracy is higher, but the expense of assistance data causes the availability of frequency spectrum relatively low, and cannot answer
In signal of communication blind recognition scene.Blind frequency excursion algorithm is special including the algorithm based on Cyclic Prefix, based on cyclo-stationary
The algorithm of property, ESPRIT, the method based on cost function and the method based on orthogonal space-time block codes (OSTBC) etc., this
Class method does not need extra assistance data expense, and the availability of frequency spectrum is high, but estimated accuracy is relatively low.
Document 1 [Mody A N, Stuber G L.Synchronization for MIMO OFDM systems [C] //
Global Telecommunications Conference,2001.GLOBECOM'01.IEEE.IEEE,2001,1:509-
513.] a kind of frequency excursion algorithm based on training sequence design is proposed.The algorithm is direct modulation orthogonal more by designing
Phase training sequence improves performance of the algorithm under low signal-to-noise ratio, and algorithm complex is low.The method is disadvantageous in that instruction
The expense for practicing sequence causes the availability of frequency spectrum to reduce, and cannot apply in signal of communication blind recognition scene.
[Shahbazpanahi S, Gershman A B, the Giannakis G B.Joint blind channel of document 2
and carrier frequency offset estimation in orthogonally space-time block
coded MIMOsystems[C]//Signal Processing Advances in Wireless Communications,
2005 IEEE 6th Workshop on.IEEE,2005:363-367] and document [3 Shahbazpanahi S, Gershman
A B,Giannakis G B.Blind and semiblind channel and carrier frequency-offset
estimation in orthogonally space-time block coded MIMO systems[J].Signal
Processing,IEEE Transactions on,2008,56(2):702-711.] propose based on OSTBC coding blind frequency
Deviation estimation algorithm completes offset estimation using the orthogonality construction likelihood function of OSTBC encoder matrixs, and the algorithm need not be trained
Sequence or pilot tone, but offset estimation is carried out using the data division that OSTBC is encoded, the availability of frequency spectrum is higher, and estimated accuracy
It is higher than general blind frequency excursion algorithm, but the mimo system of orthogonal and nonopiate space-time block code cannot be directed at had
The offset estimation of effect, is of limited application.
The content of the invention
Shortcoming it is an object of the invention to overcome above-mentioned prior art, for MIMO-STBC systems, there is provided one kind is based on
The blind frequency-offset estimating method of maximum likelihood two-dimensional search is higher compared to the data-aided frequency excursion algorithm availability of frequency spectrum and estimate
Meter precision is higher;Compared to the blind frequency excursion algorithm based on OSTBC codeword structures, the present invention can be used not only for OSTBC codings
Mimo system, and go for the mimo system of nonopiate space-time block code and space-time code, i.e. this hair
It is bright suitable for MIMO-STBC systems, range of application is wider.
To achieve the above object, the concrete thought of present invention realization is:According to the signal message that communication control processor is received,
Obtain receipt signal matrix.Initialization frequency deviation search listing and code parameter search list.Received according to the compensation of frequency deviation search listing
Signal matrix obtains frequency deviation thermal compensation signal matrix.According to code parameter search list, each frequency deviation thermal compensation signal matrix is counted respectively
Calculate frequency deviation-code parameter maximum likelihood function value.Frequency in judgement frequency deviation-code parameter maximum likelihood function value corresponding to maximum
Partially.
It is as follows that the present invention implements step:
(1) signal is received.Communication control processor receives the MIMO signal that transmitting terminal sends by reception antenna, obtains receiving letter
Number matrix Y;
(2) initialize.
Frequency deviation search listing 2a) is set:Frequency deviation step-size in search is set according to the required precision of offset estimation, according to system
Frequency deviation region sets frequency deviation hunting zone, and frequency deviation search listing is obtained by frequency deviation step-size in search and frequency deviation hunting zone, and will row
First frequency deviation is used as current frequency offset in table;
Code parameter search list 2b) is set:According to transmitting antenna number, by the code parameter corresponding to be possible to STBC, i.e.,
The information symbol number that includes in block length and each packet, enters row-column list, and using first group of code parameter in list as current
Code parameter;
(3) receipt signal matrix are compensated:Frequency deviation compensation is carried out to receipt signal matrix using current frequency offset, frequency deviation benefit is obtained
Repay signal matrix.
(4) pre -whitening processing is carried out to frequency deviation thermal compensation signal matrix, obtains prewhitening signal matrix.
(5) frequency deviation-code parameter maximum likelihood function is calculated:
It is 5a) that the real part of prewhitening signal matrix and imaginary part is in parallel, and carry out packet combining treatment, i.e., according to when former code ginseng
Number, will belong in parallel with the reception signal column vector of packet, and calculate its correlation matrix;
5b) to step 5a) in correlation matrix do Eigenvalues Decomposition, obtain eigenvalue matrix;
5c) frequency deviation-code parameter maximum likelihood function is calculated according to when former code parameter:
Wherein NrIt is reception antenna number, ρkStep 5a) in the eigenvalue matrix that obtains, l is when corresponding to former code parameter
Block length, n is the information symbol number included in each packet corresponding to the former code parameter.
(6) current frequency offset is recorded and when the frequency deviation corresponding to former code parameter-code parameter maximum likelihood function value.
(7) decide whether to have traveled through a yard parameter search list, if meeting, be transferred to step (8), otherwise, searched according to code parameter
Rope list update works as former code parameter, then is transferred to step (5).
(8) decide whether to have traveled through frequency deviation search listing, if meeting, be transferred to step (9), otherwise, row are searched for according to frequency deviation
Table updates current frequency offset, then is transferred to step (4);
(9) it is judgement frequency deviation to take the frequency deviation corresponding to record set frequency deviation-code parameter maximum likelihood function maximum.
The present invention has advantages below compared with prior art:
1. because the present invention docking collection of letters number is using the Two-dimensional Maximum likelihood search of frequency deviation-code parameter so that this hair
The estimated accuracy of bright method is high.
2. because the present invention estimates frequency deviation merely with the codeword structure of STBC, it is not necessary to training sequence or pilot tone.So that being
System can reduce the expense of training sequence or pilot tone, therefore the availability of frequency spectrum is higher;
3. due to the frequency deviation-code parameter maximum likelihood function of the system rely only on block length and group separator number, make
Present invention may apply to orthogonal, nonopiate and standard is orthogonal, and the multiple situation such as layered space-time group code LSTBC, is applicable
Scope is wide;
Because the present invention docking collection of letters number takes packet combining treatment so that frequency deviation-code parameter maximum likelihood function
Result does not rely on data start, has for incomplete property data and estimates performance well, therefore this method is to synchronous mistake
Difference has certain tolerance;
Brief description of the drawings
Fig. 1 is the system model schematic diagram that the present invention is used;
Fig. 2 is flow chart of the invention;
Fig. 3 be it is of the invention with it is existing based on training sequence periodically blind frequency excursion algorithm offset estimation mean square error
Compare figure.
Specific embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.
Reference picture 1, the system model that the present invention is used includes 1 emitter and 1 receiver, wherein, the day of emitter
Line number mesh is Nt, the number of antennas of receiver is Nr, and Nr≥Nt;In emitter, information sequence is by space-time block code
(STBC) it is, parallel data stream by converting serial data streams, then passes through N respectivelytRoot antenna sends.Transmission signal is by flat
Smooth Rayleigh quasistatic channel reaches receiver, receives signal and is expressed as Y=HX+B, wherein, Y is receipt signal matrix, and H is element
The channel matrix of independent Gaussian distributed, X is transmission signal matrix, and B is white Gaussian noise matrix.
2 the present invention is described in further detail below in conjunction with the accompanying drawings.
Step 1, receiver receives the MIMO signal that transmitting terminal sends by reception antenna, obtains receipt signal matrix Y:
Wherein,Represent receiver n-thrThe signal that root antenna is received in l-th time slot, 1≤nr≤Nr, NrTo receive day
Line number, 1≤l≤L, L are the timeslot number that receiver receives signal.
Step 2, initialization.
Frequency deviation search listing 2a) is set:Frequency deviation step-size in search is set according to the required precision of offset estimation, according to system
Frequency deviation region sets frequency deviation hunting zone, and frequency deviation search listing is obtained by frequency deviation step-size in search and frequency deviation hunting zone, and will row
First frequency deviation is used as current frequency offset f in table;
Code parameter search list 2b) is set:According to transmitting antenna number, by the code parameter corresponding to be possible to STBC, i.e.,
The information symbol number that includes in block length and each packet, enters row-column list, and using first group of code parameter in list as current
Code parameter (n, l);
Step 3, compensates receipt signal matrix.
Frequency deviation compensation is carried out to receipt signal matrix using current frequency offset, frequency deviation thermal compensation signal matrix is obtained
Wherein, j is imaginary unit, and f is current frequency offset, and L is the timeslot number that receiver receives signal;
Step 4, pre -whitening processing.
4a) calculate the correlation matrix of frequency deviation thermal compensation signal matrix, and Eigenvalues Decomposition done to correlation matrix, obtain feature to
Moment matrix U and eigenvalue matrix Λ2:
Wherein, λ1≥λ2≥…λk,
4b) whitening matrix W is calculated using eigenvectors matrix U and eigenvalue matrix:
Wherein, ()TRepresent transposition computing;
Pre -whitening processing 4c) is carried out to frequency deviation thermal compensation signal matrix using whitening matrix W, prewhitening signal matrix is obtained.
Step 5, calculates frequency deviation-code parameter maximum likelihood function.
It is 5a) that the real part of prewhitening signal matrix and imaginary part is in parallel, and carry out packet combining treatment, i.e., according to code parameter,
To belong in parallel with the reception signal column vector of packet, and calculate its correlation matrix.
5b) to step 5a) in correlation matrix do Eigenvalues Decomposition, obtain eigenvalue matrix;
5c) frequency deviation-code parameter maximum likelihood function is calculated according to when former code parameter:
Wherein, NrIt is reception antenna number, ρkStep 5b) in the eigenvalue matrix that obtains, l is when corresponding to former code parameter
Block length, n is the information symbol number included in each packet corresponding to the former code parameter.
Step 6, record current frequency offset with when the corresponding frequency deviation-code parameter maximum likelihood function value of former code parameter.
Step 7, decides whether to have traveled through a yard parameter search list, if meeting, is transferred to step 8, otherwise, is searched according to code parameter
Rope list update works as former code parameter, then is transferred to step 5.
Step 8, decides whether to have traveled through frequency deviation search listing, if meeting, is transferred to step 9, otherwise, row is searched for according to frequency deviation
Table updates current frequency offset, then is transferred to step 4.
Step 9, it is judgement frequency deviation to take the frequency deviation corresponding to record set frequency deviation-code parameter maximum likelihood function maximum.
Effect of the invention is described in detail with reference to Fig. 3.
1. simulated conditions
Analogue system is transmitting antenna number Nt=4, reception antenna Nr=8 MIMO-STBC systems.Transmission signal frame length is
96, data are modulated using QKSP and use nonopiate Space Time Coding.The element independent same distribution of channel matrix H, obeying average is
0, variance is 1 multiple Gauss distribution, and noise is white Gaussian noise, and average is 0, varianceDetermined by Normalized Signal/Noise Ratio.Sample
Symbol lengths are 1000.Emulation SNR ranges are 0~25dB, every 5dB emulation once, carry out Monte Carlo simulation, cover special
Carlow simulation times are 1000.
It is made up of training sequence and data per frame signal in analogue system 1, training sequence can be direct for document [1] proposition
The orthogonal leggy training sequence of modulation, training sequence a length of 32, data length is 64.Per frame signal without instruction in analogue system 2
Practice sequence, be only made up of data, data length is 96.The availability of frequency spectrum of analogue system 2 is higher than analogue system 1.
2. emulation content and simulation result
The method based on training sequence design that [document 1] proposition is respectively adopted is carried out with the present invention to system 1 and system 2
Simulation comparison, as shown in Figure 3.The solid line of wherein circle mark represents analogue system 1 using the frequency deviation based on training sequence design
The offset estimation normalized mean squared error of algorithm for estimating, the solid line of square mark represents analogue system 2 and uses frequency of the invention
Normalized mean squared error is estimated partially.Can be seen that detection method of the invention by the figure can be to nonopiate Space Time Coding
MIMO signal carries out effective offset estimation, and estimated accuracy higher than the frequency excursion algorithm based on training sequence.
It should be appreciated that for those of ordinary skills, can according to the above description be improved or converted,
And all these modifications and variations should all belong to the protection domain of appended claims of the present invention.