CN101014029B - Method for generating OFDM synchronous training sequence and synchronizing method based on the training sequence - Google Patents

Abstract

The invention provides one OFDM simultaneous train sequence generation method and one based on train sequence for communication, wherein, the OFDM simultaneous train sequence generation method comprises steps as: generating OFDM simultaneous train sequence by two OFDM characters with two characters for regular and irregular characters. The train sequence simultaneous method comprises steps of rough simultaneous timing; small number frequency estimation, integral frequency estimation and fine simultaneous timing.

Description

The generation method of OFDM synchronous training sequence and based on the method for synchronous of this training sequence

Technical field

The present invention relates to communication technical field, the generation method of particularly a kind of OFDM (OFDM) synchronous training sequence and based on the method for synchronous of this training sequence.

Background technology

At present, the OFDM technology is applied at increasing wired, wireless communication field, as: DAB (digital audio broadcasting), DVB (digital video broadcasting), IBEE802.11a, HIPERLAN/2, ADSL (asymmetric local loop) or the like.This has many advantages mainly due to the OFDM technology: effectively anti-multipath is disturbed and narrow band interference, and availability of frequency spectrum height, message transmission rate is high.

Ofdm system is very responsive for lacking deviation synchronously.Synchronism deviation mainly is divided into symbol timing offset and frequency departure.When the symbol timing offset was meant the demodulating ofdm symbol, FFT, window shifted to an earlier date or lag behind.The if symbol timing offset is excessive, makes the length of the length sum of side-play amount regularly and maximum delay expansion less than Cyclic Prefix, can produce intersymbol interference, destroys the integrality of OFDM symbol, reduces the performance of system.Frequency departure f _ΔThe modulated carriers that is meant the demodulation carrier frequency of receiving terminal and transmitting terminal is inconsistent, and this is mainly due to crystal oscillator instability or Doppler frequency shift.Frequency departure is divided into the decimal overtones band deviation f of subcarrier spacing again _FInteger multiple frequency deviation f with subcarrier spacing _I, abbreviate decimal frequency bias and integer frequency bias hereinafter respectively as.Wherein, decimal frequency bias can cause and disturb (ICI) between subcarrier; Integer frequency bias can not cause ICI, but can cause the cyclic shift that receives data symbol, makes that the error probability of the information symbol that demodulation is come out is 50%.Therefore, estimate symbol timing offset and frequency departure effectively are the key factors that can ofdm system operate as normal.

Fig. 1 is the block diagram of a general purpose O FDM baseband system.Transmitting terminal string among Fig. 1 and conversion, IDFT (Fourier inversion), transmitting terminal and string conversion, insertion Cyclic Prefix insert modules such as synchronizing information, represent OFDM baseband modulation process; Separation synchronizing information among Fig. 1 is removed Cyclic Prefix, receiving terminal string and conversion, and DFT (Fourier transform) and frequency domain equalization, modules such as receiving terminal and string conversion are corresponding with it OFDM base band demodulating processes; Digital-to-analogue conversion among Fig. 1 sends Filtering Processing, channel, and the processing that accepts filter, modules such as analog-to-digital conversion are analog transmissions and channel link; Lock unit module among Fig. 1 promptly is to realize the synchronous parts of OFDM.Now the modulation and demodulation process of ofdm system is briefly introduced in conjunction with Fig. 1:

If the baseband sampling time interval is T _s, the significant character of ofdm system is counted and is N (N=2 generally speaking, ^β), the Dui Ying significant character cycle is T=NT with it _sN _uIt is the number of effective subcarrier in N the OFDM subcarrier.At transmitting terminal, frequency domain data { a _{I, k}| (k=-N/2 ..., 0,1 ..., N/2-1) } (a _{I, k}Be i symbol, loaded data on k the subcarrier) is placed into transmitting terminal string and modular converter among Fig. 1; Through the IDFT module, obtain time domain data { b at the output of transmitting terminal parallel serial conversion module _{I, l}| (l=0,1,2 ..., N-1) } (b _{I, l}Be i symbol, the data of l sampled point); In order to resist intersymbol interference (ISI), add N _gThe Cyclic Prefix (the insertion cyclic prefix module among Fig. 1) of point is so each OFDM symbol contains to count and is N _Sym(N _Sym=N+N _g); Insert synchronous training sequence (the insertion synchronizing information module among Fig. 1) again, whole then signal arrives receiving terminal by channel; At receiving terminal, isolate synchronous training sequence (the separation synchronizing information module among Fig. 1) and send into the lock unit module among Fig. 1; Leave out Cyclic Prefix (the removal cyclic prefix module among Fig. 1) again, the receiving terminal string in Fig. 1 and the output of modular converter obtain time domain data { r _{I, m}| (m=0,1,2 ..., N-1) } (r _{I, m}Be i symbol, the data of m sampled point); Pass through DFT and frequency domain equalization module among Fig. 1 at last, the output of the receiving terminal parallel serial conversion module in Fig. 1 must obtain the frequency domain data { z of demodulation _{I, n}| (n=-N/2 ..., 0,1 ..., N/2-1) } (z _{I, n}Be i symbol, demodulated data on n the subcarrier).

Common OFDM synchronization scenario has two kinds:

(1) based on the synchronous training sequence of separation structures such as specific frequency domain sequence and 2, this method is by the auto-correlation of the half symbols time-delay of calculation training symbol, seek correlation peak and obtain symbol regularly, ask the phase angle at correlation peak place to estimate decimal frequency bias, after more synchronous training symbol being carried out the decimal frequency bias compensation, do fast Fourier transform (FFT), to make cyclic shift relevant with known frequency domain sequence then, estimates integer frequency bias by seeking relevant peaks.The frequency deviation region that this method can be estimated increases along with the expansion of relevant search scope, and still, its computation complexity also can constantly rise.Therefore, because computation complexity, this method is difficult to be applied in the real system.Referring to document: Schmidl, T.M. etc. " Low-overhead; low-complexity[burst] synchronization for OFDM ", IEEE International Conference onCommunications, Volume3, June 1996, Page (s): 1301-1306 (" the OFDM method for synchronous of low accessing cost for data, low complex degree " IEEE international communication technical conference).

(2) based on L five equilibrium (L=2 ^v) the synchronous training symbol structure of 0FDM, the auto-correlation of the specific time-delay by the calculation training symbol is sought correlation peak and is obtained symbol regularly, asks the phase angle at correlation peak place to estimate frequency deviation again.Though this method computation complexity is low, its time net synchronization capability is relatively poor, and frequency offset estimation range is less, only be [L/2, L/2).In practice, the value of L generally can not be too big.Because excessive L can cause the peak-to-average power ratio of ofdm system too high, and produce bigger frequency offset estimating error, and the sparse property of the frequency domain data of training symbol causes the ability drop of its opposing frequency selective fading.Referring to document: Heiskala J, Terry J:OFDM Wireless LANs-A Theoretical and Practical Guide.[M] .Indianapolis USA:Pearson Education Inc, 2002.70-73 (" guidance of OFDM WLAN (wireless local area network)--theory and practice ").

Therefore, method (1) belongs to frequency domain algorithm, need make FFT to whole synchronous training symbol, and its major defect is the complexity height; Method (2) belongs to Time-Domain algorithm, though do not need whole synchronous training symbol is made FFT, has lower complexity, and its major defect is a poor-performing, and this is restricted above-mentioned two kinds of methods in actual applications.

Summary of the invention

The objective of the invention is at the deficiencies in the prior art, provide a kind of generation method of OFDM synchronous training sequence and based on the method for synchronous of this training sequence.The generation method of OFDM synchronous training sequence of the present invention is to generate the OFDM synchronous training sequence by 2 OFDM symbol cascades, and described 2 OFDM symbols are respectively non-canonical OFDM symbol and canonical OFDM symbol.Method for synchronous of the present invention may further comprise the steps based on the OFDM synchronous training sequence that the generation method according to described OFDM synchronous training sequence generates: thick regularly synchronous, decimal frequency bias estimates that integer frequency bias is estimated, and is thin regularly synchronous.Wherein adopt double sign cascading judgement algorithm to obtain more excellent thick timing and estimate synchronously, adopt the remainder theorem algorithm that frequency deviation is estimated again.With respect to existing scheme, advantage such as the present invention has that computation complexity is lower, frequency offset estimation range is big, timing synchronization and Frequency Synchronization are comparatively accurate.

The present invention is achieved by the following technical solutions:

The generation method of OFDM synchronous training sequence of the present invention, be to generate the OFDM synchronous training sequence by 2 OFDM symbol cascades, and more existing synchronous training sequence structures often need more symbols (as 4 OFDM symbols of WLAN (wireless local area network) needs), therefore, the training sequence that is generated with the generation method of OFDM synchronous training sequence of the present invention has the low advantage of accessing cost for data.Being constructed as follows of this sequence:

The 1st symbol: this is a non-positive then symbol (irregular symbol), is designated as S _a, S _aCounting of significant character part be N _a(N _aBe integer, and contain odd prime factor q, N _a≠ N unless N is the canonical outer symbol, comprises that the significant character of other OFDM symbols of positive then symbol is counted), it is divided into L _aPart (L _aBe integer, also contain odd prime factor q), usually, each part signal is called a slot.Each slot contains M _aPoint (M _aBe integer, and M _a=N _a/ L _a), S _aCyclic Prefix partly contain to count and be N _Ag(N _Ag=N _Sym-N _a), therefore, S _aalways count still and to be N _Sym

The 2nd symbol: this is a positive then symbol (regular symbol), is designated as S _n, counting of its live part is N, it is divided into L part, and (L is an integer, and L=2 ^v), each slot contains M point (M is an integer, and M=N/L), and some representative values of L are 4,8,16 etc.

Below the generation method of OFDM synchronous training sequence of the present invention is further described:

If S _a, S _nCorresponding OFDM symbolic label is respectively 1,2.

S _nBe the synchronous training symbol of canonical OFDM of a L five equilibrium, this is a kind of quite common sequential structure, in being usually used in timing synchronization algorithm and frequency offset estimating algorithm.For example, get L=4 in the IEEE802.11a standard.Its generation method is at first to insert frequency domain data according to formula (1)

$\left\{\begin{array}{ccc}{a}_{2,k}\&NotEqual;0& k& \left(\begin{array}{cc}\mathrm{mod}& L\end{array}\right)=0\\ a_{2,k}=0& k& \left(\begin{array}{cc}\mathrm{mod}& L\end{array}\right)\&NotEqual;0\end{array}\right.(0<|k|\&le;\frac{N_u}{2})---\left(1\right)$

Carry out N point Fourier inversion then and produce corresponding time-domain signal, the front end at time-domain signal adds Cyclic Prefix afterwards, can generate S _nUnder the situation of L five equilibrium, S _nContained the counting of each slot be M (M=N/L).

Symbol S _aContain L _aIndividual slot establishes L _aContain odd prime factor q, and have 2 ^α-1＜L _a≤ 2 ^α, (α is a positive integer), N _a(mod L _a)=0 and N (mod L _aN is got in) ≠ 0 _aBe L _aInteger multiple in the immediate natural number of N, that is:

$N_a=\underset{\mathrm{\&mu;}}{\arg \min }\left\{\right|\mathrm{\&mu;}-N\left|\right|\mathrm{\&mu;}\left(\mathrm{mod}{L}_a\right)=0\}---\left(2\right)$

S _aThe generation method be: with L _aAnd N _a-2 ^α+1L and N in the replacement formula (1) _u:

$\left\{\begin{array}{ccc}{a}_{1,k}\&NotEqual;0& k& \left(\mathrm{mod}{L}_a\right)=0\\ a_{1,k}=0& k& \left(\mathrm{mod}{L}_a\right)\&NotEqual;0\end{array}\right.(0<|k|\&le;\frac{N_u-2^{\mathrm{\&alpha;}+1}}{2})---\left(3\right)$

Insert frequency domain data according to formula (3), carry out N then _aCan get b after the some Fourier inversion _{1, l}:

$b_{1,l}=\frac{1}{N_a}\underset{k=0}{\overset{N_a-1}{\mathrm{\&Sigma;}}}{a}_{1,k}\exp \left(\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}\frac{\mathrm{kl}}{N_a}\right)(l=\mathrm{0,1,2},...,N_a-1)---\left(4\right)$

Add N again _AgThe dot cycle prefix just can obtain S _aNeed to prove the N of formula (3) _u, lack 2 than formula (1) ^α+1Individual, this is the frequency spectrum distortion that causes for compensation (2).

The present invention also provides the method for synchronous based on above-mentioned synchronous training sequence, and is specific as follows:

Step 1: thick regularly synchronous, to S _aDo time-delay and be M _aPoint, window is long to be (N _Sym-M _a) point auto-correlation, P _a(d).To S _nDo time-delay and be the M point, window is long to be (N _Sym-M) auto-correlation, P _n(d).To S _nDo time-delay and be the N/2 point, window is long to be (N _Sym-N/2) auto-correlation, P _f(d), again to P _a(d), P _n(d) and P _f(d) ask the energy sum,, get average autocorrelation P (d), seek the peak value of P (d), go out the original position of OFDM symbol according to a preliminary estimate divided by the signal gross energy in the associated window

This algorithm is called " double sign cascading judgement algorithm ".This algorithm can improve the thick regularly synchronization performance of system greatly, and this algorithm has lower computation complexity.

Step 2: decimal frequency bias estimation, the i.e. P that obtains according to step 1 _f(d), calculate it in the position

The phase angle at place, thereby the fractional part of estimation frequency deviation

Parity η with integer frequency bias.

Step 3: integer frequency bias estimation, the i.e. P that obtains according to step 1 _a(d) and P _n(d), calculate it in the position

The phase angle at place utilizes the remainder theorem in the number theory, estimates integer frequency bias This algorithm is called " remainder theorem algorithm ".Step 2 is obtained

With

Addition obtains the frequency offset estimating value

, its estimation range can reach [L _x/ 2, L _x/ 2) (L _xBe L and L _aLeast common multiple).This algorithm has the low and big advantage of frequency offset estimation range of computation complexity simultaneously.

Step 4: thin regularly synchronous, promptly obtain according to step 1

, extract S _nThe 2nd slot, obtain according to step 3 then

It is carried out compensate of frequency deviation, and it is relevant again itself and local slot sample to be made cyclic shift, seeks relevant peaks, thereby estimates the accurate original position of OFDM symbol

Below method for synchronous of the present invention is further described:

(1) thick regularly synchronous

Thick timing synchronized algorithm provided by the invention is called " double sign cascading judgement method ".

At first, calculate P _a(d), P _n(d) and P _f(d).As follows:

To S _aDo time-delay and be M _aPoint, window is long to be (N _Sym-M _a) point auto-correlation, P _a(d):

$P_a\left(d\right)=\underset{m=d}{\overset{d+N_{\mathrm{sym}}-M_a}{\mathrm{\&Sigma;}}}{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="H06130331520060904E000013.png,H06130331520060904E000031.png"\; state="\{\{state\}\}">r</figure-callout>}}_{1,m}^{*}{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="H06130331520060904E000013.png,H06130331520060904E000031.png"\; state="\{\{state\}\}">r</figure-callout>}}_{1,m+M_a}---\left(5\right)$

To S _nDo time-delay and be the M point, window is long to be (N _Sym-M) put auto-correlation, get P _n(d):

$P_n\left(d\right)=\underset{m=d}{\overset{d+N_{\mathrm{sym}}-M}{\mathrm{\&Sigma;}}}{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2,m}^{*}{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2,m+M}---\left(6\right)$

To S _nDo time-delay and be the N/2 point, window is long to be (N _Sym-N/2) auto-correlation, P _f(d):

$P_f\left(d\right)=\underset{m=d}{\overset{d+N_{\mathrm{sym}}-N/2}{\mathrm{\&Sigma;}}}{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2,m}^{*}{r}_{2,m+N/2}---\left(7\right)$

Then formula (5), formula (6) are asked the energy sum with formula (7),, get average autocorrelation P (d) again divided by the signal gross energy in the associated window,

$P\left(d\right)=\frac{{\left|P_a\left(d\right)\right|}^2+{\left|P_n\left(d\right)\right|}^2+{\left|P_f\left(d\right)\right|}^2}{{\left(\underset{m=d}{\overset{d+N_{\mathrm{sym}}-M_a}{\mathrm{\&Sigma;}}}{\left|{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="H06130331520060904E000013.png,H06130331520060904E000031.png"\; state="\{\{state\}\}">r</figure-callout>}}_{1,m+M_a}\right|}^2\right)}^2+{\left(\underset{m=d}{\overset{d+N_{\mathrm{sym}}-M}{\mathrm{\&Sigma;}}}{\left|{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2,m+M}\right|}^2\right)}^2+{\left(\underset{m=d}{\overset{d+N_{\mathrm{sym}}-N/2}{\mathrm{\&Sigma;}}}{\left|{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2,m+N/2}\right|}^2\right)}^2}---\left(8\right)$

Seek the peak value of P (d) then, thereby estimate the thick timing starting point of OFDM symbol

$\hat{d}=\underset{d}{\arg \max }\left\{P\left(d\right)\right\}---\left(9\right)$

In the formula (8), | P _a(d) | ²With | P _n(d) | ²Relevant peaks not sharp-pointed, when low signal-to-noise ratio, have stronger noiseproof feature; | P _f(d) | ²Relevant peaks more sharp-pointed, when high s/n ratio, can make comparatively accurately and estimating timing synchronization.Therefore, " double sign cascading judgement method " provided by the invention uses a plurality of relevant peaks to merge, and improved the thick regularly synchronization performance of system greatly.And formula (5) to the calculating of formula (8) can recurrence be carried out, and is to take advantage of again and several times plural number plus-minus for 9 times to the amount of calculation equivalence of each sampled point.

It is pointed out that " double sign cascading judgement method " only is applicable to the synchronizing sequence structure that the present invention designs.If adopt another positive then symbol S _nReplace S _a, suppose that it also is divided into L part, so, " double sign cascading judgement method " can be at d=N _SymAnd d=-N _Sy(supposing desirable symbol original position d=0) locates to have occurred respectively a relevant peaks, and this can the serious judgement of disturbing timing synchronization.

(2) decimal frequency bias is estimated

The decimal frequency bias algorithm that the present invention adopts is a kind of traditional decimal frequency bias algorithm for estimating.This method can be referring to document: Schmidl, T.M. etc. " Low-overhead; low-complexity[burst] synchronization for OFDM ", IEEE International Conference onCommunications, Volume 3, June 1996, Page (s): 1301-1306 (" the OFDM method for synchronous of low accessing cost for data, low complex degree ", IEEE international communication technical conference).The specific implementation method is that N point OFDM symbol is divided into 2 parts, and it is made auto-correlation, asks phase angle again, thereby estimates decimal frequency bias.

Because the S among the present invention _nBe a positive then symbol, L is an even number, so S _nThe OFDM symbol that also can regard 2 five equilibriums as can utilize P in the formula (7) _f(d) phase angle is estimated decimal frequency bias:

$\mathrm{\&xi;}=\frac{1}{\mathrm{\&pi;}}\mathrm{angle}\left\{P_f\left(\hat{d}\right)\right\}---\left(10\right)$

The scope of ξ be [1,1), and the scope of decimal frequency bias be [0.5,0.5), so need classification discussion, obtain the decimal frequency bias estimated value

With the integer frequency bias estimated value

Parity η:

Wherein, η=1 expression Be odd number; η=0 expression

Be even number.In this programme,

With η be important supplementary, this information helps integer frequency bias estimation module to realize large-scale frequency offset estimating.

(3) integer frequency bias is estimated

In the ofdm system in future, carrier frequency will be up to 1GHz to 100GHz, and the interval of subcarrier is less relatively, has only a few kHz to tens kHz, when the crystal oscillator of receiving terminal is not when stablizing very much, just require the frequency synchronization module of ofdm system can correct in a big way frequency deviation.With the carrier frequency is 10GHz, and the crystal oscillator frequency deviation is 20ppm, and subcarrier spacing is that the system of 5KHz is an example, its maximum possible f _ΔBe ± 40.At present, the technology of correcting long-range frequency deviation mainly contains 2 kinds:

● based on the synchronous training sequence of specific frequency domain sequence, this method need be carried out the decimal frequency bias compensation to synchronous training symbol, remakes FFT, needs at least

Inferior complex multiplication, to make cyclic shift relevant with known frequency domain sequence then, estimates integer frequency bias by seeking relevant peaks.The frequency deviation region that this method can be estimated increases along with the expansion of relevant search scope, and still, its computation complexity also can constantly rise.Frequency of every search just needs approximately

Inferiorly take advantage of again.Therefore, because computation complexity, this method is difficult to be applied in the real system.Referring to document: Schmidl, T.M. etc. " Low-overhead; low-complexity[burst] synchronization for OFDM ", IEEE International Conference onCommunications, Volume3, June 1996, Page (s): 1301-1306 (" the OFDM method for synchronous of low accessing cost for data, low complex degree " IEEE international communication technical conference).

● based on L five equilibrium (L=2 ^v) the synchronous training symbol structure of OFDM, the auto-correlation of the specific time-delay by the calculation training symbol asks phase angle to estimate integer frequency bias again.The maximum estimated scope of this method is that [L/2, L/2), still, the actual value of L generally can not be too big.Because excessive L can cause the peak-to-average power ratio of ofdm system too high, and produce bigger frequency offset estimating error, and the sparse property of the frequency domain data of training symbol causes the ability drop of its opposing frequency selective fading.Referring to document: Heiskala J, Terry j:OFDM Wireless LANs-A Theoretical and Practical Guide.[M] .Indianapolis USA:Pearson Education Inc, 2002.70-73 (" guidance of OFDM WLAN (wireless local area network)--theory and practice ").

Based on the OFDM synchronous training sequence that the generation method of OFDM synchronous training sequence of the present invention generates, it is bigger to design a kind of frequency offset estimation range, and the lower integer frequency bias algorithm for estimating of complexity, is referred to as " remainder theorem algorithm ".This algorithm at first utilizes formula (11) to obtain And η, obtain the P in the formula (5) _a(d) and the P in the formula (6) _n(d) in the position The phase angle at place is estimated (f thereby integer frequency bias is made 2 _aWith f _n):

$f_a=\left\{\begin{array}{c}\underset{\mathrm{\&lambda;}}{\arg \min }\left\{\right|\mathrm{\&lambda;}-\left(\frac{L_a}{2\mathrm{\&pi;}}\mathrm{angle}\right\{P_a\left(\hat{d}\right)\}-{\hat{f}}_F)\left|\right\}\\ \underset{\mathrm{\&lambda;}}{\arg \min }\left\{\right|\mathrm{\&lambda;}-\left(\frac{L_a}{2\mathrm{\&pi;}}\mathrm{angle}\right\{P_a\left(\hat{d}\right)\}-{\hat{f}}_F)\left|\right|\mathrm{\&lambda;}\left(\mathrm{mod}2\right)=\mathrm{\&eta;}\}\end{array}\right.---\left(12\right)$

$f_n=\underset{\mathrm{\&lambda;}}{\arg \min }\left\{\right|\mathrm{\&lambda;}-\left(\frac{L_a}{2\mathrm{\&pi;}}\mathrm{angle}\right\{P_n\left(\hat{d}\right)\}-{\hat{f}}_F)\left|\right|\mathrm{\&lambda;}\left(\mathrm{mod}2\right)=\mathrm{\&eta;}\}---\left(13\right)$

f _a(modL _a) L arranged _aPlant possible value: 0,1 ..., L _a-1}; f _n(mod L) has the possible value of L kind: and 0,1 .., L-1}.By f _a(mod L _a) and f _n(modL), can derive an Indeterminate Equation Group:

$\left\{\begin{array}{c}{\hat{f}}_I\left(\mathrm{mod}{L}_a\right)=f_a\left(\mathrm{mod}{L}_a\right)\\ {\hat{f}}_I\left(\mathrm{mod}L\right)=f_n\left(\mathrm{mod}L\right)\end{array}\right.---\left(14\right)$

According to the remainder theorem in the number theory, by f _a(mod L _a) and f _nThe various combination of (mod L) can solve It has L _xPlant possible integer solution: { L _x/ 2 ,-L _x/ 2+1 ..., L _x/ 2-1} (L _xBe L and L _aLeast common multiple).Therefore, the frequency deviation region that can estimate of the present invention reaches [L _x/ 2, L _x/ 2).For convenience hardware is realized, can use the look-up table group (14) of solving an equation: first to L _xIndividual feasible solution

Enumerate, calculate its correspondence

(mod L _a) and

(mod L), and these two the value as

The entry address of form, again will

Be filled into the memory cell of entry address correspondence, so just made the table of finding the solution of equation group (14).In estimating the integer frequency bias process, only need use f _a(mod L _a) and f _n(modL), table look-up and obtain as table index

Get final product.

" remainder theorem algorithm " provided by the invention has the low and big advantage of frequency offset estimation range of computation complexity simultaneously, specifically, formula of this algorithm (12) and formula (13) only need be asked the computing of phase angle for 2 times, can separate Indeterminate Equation Group (14) with look-up table then, just can finish estimation integer frequency bias:

${\hat{f}}_I=\mathrm{LUT}(f_a\left(\mathrm{mod}{L}_a\right),f_n\left(\mathrm{mod}L\right))---\left(15\right)$

Function LUT in the formula (15) represents look-up-table function, and its computation complexity can be ignored, yet the frequency offset estimation range of this algorithm is but very big.With L _a=7, L=8 is an example, because L _x=56, thus frequency offset estimation range be [28,28), again with L _a=15, L=8 is an example, because L _x=120, thus frequency offset estimation range reach [60,60), big like this frequency offset estimation range can satisfy the needs of following ofdm system.

In addition, the present invention has flexibility, chooses different L _aAnd L, just can satisfy the requirement of the integer frequency bias estimation of different occasions.Such as, get L _a=3, L=2, then L _x=6, its frequency offset estimation range have only [3,3), can be used for frequency offset estimating among a small circle.And equation group (14) is used, and to find the solution table be quite simple, and only needing a size be (L _a* L) two-dimensional array of Byte is stored L _xIndividual integer solution gets final product.With L _a=7, L=8 is an example, equation group (14) find the solution the table as shown in the table:

Formula (11) is obtained

Obtain with formula (15)

Addition, make complete frequency offset estimating:

${\hat{f}}_{\mathrm{\&Delta;}}={\hat{f}}_I+{\hat{f}}_F---\left(16\right)$

It needs to be noted, because S _aIrregularity, in the formula (12)

It is the approximate formula of frequency offset estimating.In fact, utilize

Phase angle carry out right formula that frequency deviation do to estimate as the formula (17):

${\mathrm{\&xi;}}_a=\frac{\mathrm{<figure-callout\; id="2"\; label="N"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">N</figure-callout>}}{2\mathrm{\&pi;}(N_a/L_a)}\mathrm{angle}\left\{P_a\left(\hat{d}\right)\right\}---\left(17\right)$

As the definite N of formula (2) _aWith N very near the time:

${\mathrm{\&xi;}}_{\text{a}}\text{\&ap;}\frac{L_a}{2\mathrm{\&pi;}}\mathrm{angle}\left\{P_a\left(\hat{d}\right)\right\}---\left(18\right)$

Obviously, the estimation range of formula (18) is [L _a/ 2, L _a/ 2), satisfy the requirement of remainder theorem, and the estimation range of formula (17) is [NL _a/ 2N _a, NL _a/ 2N _a), require not to be inconsistent with the inherence of remainder theorem, therefore, among the present invention, employing formula (18) is come replacement formula (17), and the error factor of Yin Ruing is thus:

$\mathrm{error}\left({\mathrm{\&xi;}}_a\right)=|\frac{N_a}{N}-1|---\left(19\right)$

Because error (ξ _a) be one and take advantage of sex factor, so its when frequency deviation is big to influence meeting serious slightly, the maximum frequency deviation error that it causes is For example, work as N _a=1022, L _a=7, N=1024, during L=8, error (ξ _a)=0.00195, $\frac{L_a}{2}\mathrm{error}\left({\mathrm{\&xi;}}_a\right)=0.006825.$ Consider ξ _aJust be used for integer frequency bias and estimate, therefore, error (ξ _a) hardly can be to f _aEstimation exert an influence.

(4) thin regularly synchronous

In order to reach regularly synchronous preferably estimated performance, the present invention is adopting " double sign cascading judgement method " to obtain thick preferably regularly estimation synchronously

After, adopt thin synchronized algorithm that timing synchronization is made fine estimation again, concrete grammar is:

Extract positive then symbol S _nIn the 2nd slot, and obtain with formula (16)

It is carried out phase compensation, get x=(x ₀, x ₁..., x _m..., x _M-1), wherein,

$x_m={\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2,\hat{d}+M+m}\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;m}{\hat{f}}_{\mathrm{\&Delta;}}/N)(m=\mathrm{0,1,2},...,M-1)---\left(20\right)$

The cyclic shift that x and local slot sample are made time domain is relevant, seeks relevant peaks, thereby reaches the synchronous purpose of thin timing:

$\widehat{\mathrm{\&epsiv;}}=\hat{d}+\underset{u}{\arg \mathrm{ma}}x\{R\left(u\right)={\left|\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">b</figure-callout>}}_{2,m+u\left(\mathrm{mod}M\right)}^{*}{x}_m\right|}^2|u=-\frac{\mathrm{<figure-callout\; id="2"\; label="M"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">M</figure-callout>}}{2},-\frac{\mathrm{<figure-callout\; id="2"\; label="M"\; filenames="H06130331520060904E000061.png,H06130331520060904E000062.png"\; state="\{\{state\}\}">M</figure-callout>}}{2}+1,...,\frac{M}{2}-1\}---\left(21\right)$

As long as

Evaluated error in ± M/2, formula (21) can be made quite accurately the initial moment of OFDM symbol and be estimated that its main amount of calculation is the inferior complex multiplication of M (M+1).

Innovation part of the present invention is: designed non-canonical OFDM symbol S _aWith canonical OFDM symbol S _nThe generation method of the synchronous training sequence of cascade, and based on this training sequence, propose " double sign cascading judgement method ", 2 relevant peaks can not appear, prevented interference to the judgement of timing synchronization, improved the synchronous estimated performance of thick timing, proposed " remainder theorem algorithm " simultaneously again, frequency offset estimation range has been extended to [L _x/ 2, L _x/ 2), L _xBe S _aIsodisperse L _aWith S _nThe product of isodisperse L, the L that some are commonly used _aAs shown in the table with the combination parameter of L and frequency offset estimation range thereof and error:

N＝1024	N a	M a	L x	f ΔEstimation range	error(ξ a)
						L a＝3，L＝2	1023	341	6	[-3，3)	0.00098
L a＝5，L＝2	1025	205	10	[-5，5)	0.00098
						L a＝7，L＝2	1022	146	14	[-7，7)	0.00195
L a＝9，L＝2	1026	114	18	[-9，9)	0.00195
						L a＝11，L＝2	1023	93	22	[-11，11)	0.00098
L a＝3，L＝1	1023	341	12	[-6，6)	0.00089
						L a＝5，L＝4	1025	205	20	[-10，10)	0.00098
L a＝7，L＝4	1022	146	28	[-14，14)	0.00195
						L a＝9，L＝4	1026	114	36	[-18，18)	0.00195
L a＝11，L＝4	1023	93	44	[-22，22)	0.00098
						L a＝3，L＝8	1023	341	24	[-12，12)	0.00098
L a＝5，L＝8	1025	205	40	[-20，20)	0.00098
						L a＝7，L＝8	1022	146	56	[-28，28)	0.00195
L a＝9，L＝8	1026	114	72	[-36，36)	0.00195

N＝1024	N a	M a	L x	f ΔEstimation range	error(ξ a)
						L a＝11，L＝8	1023	93	88	[-44，44)	0.00098

Special needs to be pointed out is that computation complexity of the present invention is very low, and is as shown in the table:

Synchronization module	Computation complexity in each frame
		Thick regularly synchronous	Each sampled point, 9 CMs and several times plural number plus-minus
Decimal frequency bias is estimated	Ask phase angle 1 time
		Integer frequency bias is estimated	Ask phase angle 2 times, look into a L aThe integer form of * L dimension
Thin regularly synchronous	The inferior complex multiplication of M (M+1)

And timing synchronization error rate of the present invention and frequency offset estimating error rate are lower, and therefore, the present invention has very high using value in ofdm system.

Description of drawings

Fig. 1 is an OFDM baseband modulation and demodulation block diagram

Fig. 2 is the schematic diagram of synchronous training sequence of the present invention

Fig. 3 is N _a=1022, L _a=7 o'clock, symbol S _aThe generation schematic diagram

Fig. 4 is N=1024, during L=8, and symbol S _nThe generation schematic diagram

Fig. 5 is a method for synchronous block diagram of the present invention

Fig. 6 is the concrete implementing procedure figure of method for synchronous of the present invention

Fig. 7 is that " double sign cascading judgement algorithm " provided by the invention is at N _a=1022, L _a=7, N=1024, L=8, N _Sym=1088 o'clock, the schematic diagram of related operation and merging

Fig. 8 is that " double sign cascading judgement algorithm " provided by the invention is at N _a=1024, L _a=8, N=1024, L=8, N _Sym=1088 o'clock, the schematic diagram of related operation and merging

Fig. 9 for the present invention under 0dB white Gaussian noise situation, the numerical value figure of P (d) during certain emulation

Figure 10 for the present invention under 0dB white Gaussian noise situation, angle{P during certain emulation _f(d) }/the numerical value figure of π

Figure 11 for the present invention under 0dB white Gaussian noise situation, the numerical value figure of R (u) during certain emulation

Figure 12 for the present invention and traditional canonical symbol training sequence thick regularly synchronously the side performance compare

Figure 13 is the present invention and the traditional performance comparison of canonical symbol training sequence aspect regularly synchronous estimated standard deviation

Figure 14 is the present invention and the traditional performance comparison of canonical symbol training sequence aspect the integer frequency bias estimated error rate.

Performance aspect the integer frequency bias estimated error rate compares Figure 15 for the present invention and frequency domain algorithm for estimating

Figure 16 is SNR=2dB, adopts the error rate flatness of integer frequency bias algorithm for estimating of the present invention in estimation range

Embodiment

Provide a concrete OFDM parameter configuration embodiment below, set forth performing step of the present invention.Need to prove that the parameter in the following example does not influence generality of the present invention.

The document of 3GPP tissue: TR 25.892 V6.0.0, " Feasibility Study for OrthogonalFrequency Division Multiplexing (OFDM) for UTRAN enhancement (Release 6) " (adopting the reliability consideration (the 6th edition) of OFDM scheme in the technology upgrading that universal mobile telecommunications system and continental rise radio insert), the one group of OFDM parameter that provides is as follows:

Carrier frequency fc 3.5GHz

System bandwidth B 6.528MHz

Sub-carrier number N 1024

Effective sub-carrier number N 705

Effective bandwidth 4.495MHz

Subcarrier spacing Δ f 6.375kHz

Cyclic extensions CP 64 (9.803us)

Symbol period Ts 156.85+9.81=166.66us

Protection carrier number 319

Under the above-mentioned parameter condition, get N _a=1024, L _a=7, N=1024, L=8, N _Sym=1088, during emulation, the system frequency deviation of setting up departments f _Δ=19.3, the channel of employing is 8 road rayleigh fading channels:

	Time-delay (us)	Relative power (dB)
			Path 1	0	0
Path 2	153	-6.65
			Path 3	306	-13.31
Path 4	459	-19.95
			Path 5	612	-26.61
Path 6	765	-33.26
			Path 7	919	-39.92
Path 8	1072	-46.57

Performing step of the present invention is as follows:

(1) generates the synchronous training symbol of OFDM, the schematic diagram of the synchronous training sequence that Fig. 2 is generated for the generation method by OFDM synchronous training sequence of the present invention.Its concrete generative process is as follows: get N according to formula (2) _a=1022, is 170 according to formula (3) in the subcarrier sequence number, 177,184,191,198,205,212,219,226,233,240,247,254,261,268,275,282,289,296,303,310,317,324,331,338,345,352,359,366,373,380,387,394,401,408,415,422,429,436,443,450,457,464,471,478,485,492,499,506,520,527,534,541,548,555,562,569,576,583,590,597,604,611,618,625,632,639,646,653,660,667,674,681,688,695,702,709,716,723,730,737,744,751,758,765,772,779,786,793,800,807,814,821,828,835,842,849, load the permanent envelope frequency domain data that generates at random on 856 the frequency, on other subcarriers, place remainder certificate, with of transmitting terminal string and the conversion of these frequency domain datas by Fig. 1, IDFT, transmitting terminal and string conversion, insert modules such as Cyclic Prefix, obtain the synchronous training symbol S of pairing 7 five equilibriums of formula (4) _aFig. 3 is N _a=1022, L _a=7 o'clock, S _aThe generation schematic diagram.Then, is 160 according to formula (1) in the subcarrier sequence number, 168,176,184,192,200,208,216,224,232,240,248,256,264,272,280,288,296,304,312,320,328,336,344,352,360,368,376,384,392,400,408,416,424,432,440,448,456,464,472,480,488,496,504,520,528,536,544,552,560,568,576,584,592,600,608,61 6,624,632,640,648,656,664,672,680,688,696,704,712,720,728,736,744,752,760,768,776,784,792,800,808,816,824,832,840,848,856, load the permanent envelope frequency domain data that generates at random on 864 the frequency, place remainder certificate on other subcarriers, with of transmitting terminal string and the conversion of such frequency domain data by Fig. 1, IDFT, transmitting terminal and string conversion, insert modules such as Cyclic Prefix, just can generate the synchronous training symbol S of 8 five equilibriums _nFig. 4 is N=1024, during L=8, and S _nThe generation schematic diagram.

(2) utilization " double sign cascading judgement method " carries out that thick this process is positioned at the thick timing synchronization module of Fig. 5 regularly synchronously, and its detailed process is positioned at the thick timing synchronization module of Fig. 6, promptly according to formula (5) to formula (8), estimate thick timing position Fig. 7 is that " double sign cascading judgement algorithm " is at N _a=1022, L _a=7, N=1024, L=8, N _Sym=1088 o'clock, the schematic diagram of related operation and merging (supposing desirable symbol original position d=0).| P _a(d) | ²With | P _n(d) | ²Relevant peaks not sharp-pointed, when low signal-to-noise ratio, have stronger noiseproof feature; | P _f(d) | ²Relevant peaks more sharp-pointed, when high s/n ratio, can make comparatively accurately and estimating timing synchronization.Therefore, " double sign cascading judgement method " provided by the invention uses a plurality of relevant peaks to merge, and improved the thick regularly synchronization performance of system greatly.If adopt another positive then symbol S _nReplace S _a, suppose that it also is divided into L=8 part, so, the related operation of " double sign cascading judgement method " and the schematic diagram of merging as shown in Figure 8, at this moment, at d=N _SymAnd d=N _SymA relevant peaks has appearred respectively in the place, and this can the serious judgement of disturbing timing synchronization.Fig. 9 be the present invention under 0dB white Gaussian noise situation, the numerical value figure of P (d) during certain emulation, as seen it peak value occurred near correct timing point.In Fig. 9, d=0 is desirable symbol initial point, this moment according to formula (8) obtain thick estimated value synchronously be $\hat{d}=-5.$

(3) decimal frequency bias is estimated, this process is positioned at the decimal frequency bias estimation module of Fig. 5, and its detailed process is positioned at the decimal frequency bias estimation module of Fig. 6, promptly according to formula (10) and formula (11), estimates decimal frequency bias

With the integer frequency bias estimated value

Parity η.Figure 10 be the present invention under 0dB white Gaussian noise situation, angle{P during certain emulation _f(d) }/and the numerical value figure of π, calculate by formula (10)

$\mathrm{\&xi;}=\mathrm{angle}\left\{P_f\left(\hat{d}\right)\right\}/\mathrm{\&pi;}=\mathrm{angle}\left\{P_f(-5)\right\}/\mathrm{\&pi;}=-0.6772$

Get by formula (11) again

${\hat{f}}_F=\mathrm{\&xi;}-\mathrm{\&xi;}/\left|\mathrm{\&xi;}\right|=-0.6772-(-0.6772)/\mathrm{abs}(-0.6772)=0.3228$

$\mathrm{\&eta;}=\left\{(\mathrm{\&xi;}-{\hat{f}}_F)\left(\mathrm{mod}2\right)\right\}=\left\{(0.6772-0.3228)\left(\mathrm{mod}2\right)\right\}=1$

(4) the integer frequency bias estimation is carried out in utilization " remainder theorem algorithm ", and this process is positioned at the integer frequency bias estimation module of Fig. 5, and its detailed process is positioned at the integer frequency bias estimation module of Fig. 6, promptly according to formula (12) and formula (13), obtains 2 integer frequency bias f _aAnd f _n, substitution formula (14) is f _a(mod 7) and f _n(mod 8) are inquired about the table of finding the solution of indeterminate equation, thereby integer frequency bias are made an estimate as the entry address

Obtain complete frequency offset estimating value by formula (16) again

The present invention under 0dB white Gaussian noise situation, during certain emulation, $\hat{d}=-5,$ ${\hat{f}}_F=0.3228,$ η=1 gets f by formula (12) _a=5, get f by formula (13) _n=3, the inquiry indeterminate equation find the solution table, be 5 at ordinate, abscissa is that 3 places obtain ${\hat{F}}_I=19,$ Get by formula (16) again ${\hat{f}}_{\mathrm{\&Delta;}}={\hat{f}}_I+{\hat{f}}_F=19.3228,$ The frequency offset estimating error only is 19.3228-19.3=0.0228.

(5) thin this process is positioned at the thin timing synchronization module of Fig. 5 regularly synchronously, and its detailed process is positioned at the thin timing synchronization module of Fig. 6, i.e. basis

Information, extract positive then symbol S _nIn the 2nd slot, utilization formula (20) is carried out compensate of frequency deviation to it, through type (21) is made fine estimation to the symbol timing position again

The present invention under 0dB white Gaussian noise situation, during certain emulation, $\hat{d}=-5,$ ${\hat{f}}_{\mathrm{\&Delta;}}=19.3228,$ The S of utilization formula (20) to receiving _nIn after the 2nd slot carry out compensate of frequency deviation, the R (u) in the calculating formula (21) again, the numerical value figure of R (u) as shown in figure 11, it peak value occurs at u=5, so, according to formula (21)

$\widehat{\mathrm{\&epsiv;}}=\hat{d}+\underset{u}{\arg \max }\left\{R\left(u\right)\right\}=-5+5=0$

As seen, Ci Shi timing estimation error is 0.

Below investigate performance performance of the present invention:

Figure 12 is the present invention and traditional canonical symbol training sequence in the performance of thick regularly synchronization aspects relatively.The parameter of traditional canonical symbol training sequence is N _a=1024, L _a=64, N=1 024, L=64, N _Sym=1088.The thick regularly timing error rate of definition is a probability $\mathrm{Pr}\left(\right|\hat{d}|\&le;\frac{M}{2}),$ M＝128。Figure 12 shows, thick timing net synchronization capability of the present invention is better than traditional scheme and surpasses 4dB, this " double sign cascading judgement algorithm " effect of being produced that is of the present invention.

Figure 13 is the present invention and the traditional performance comparison of canonical symbol training sequence aspect regularly synchronous estimated standard deviation.The parameter of traditional canonical symbol training sequence still is N _a=1024, L _a=64, N=1024, L=64, N _Sym=1088.Its thin synchronous thin method for synchronous identical that also adopt with the present invention.Figure 13 shows, timing net synchronization capability of the present invention is better than traditional scheme and surpasses 3dB, and this mainly is of the present invention " double sign cascading judgement algorithm " makes the thick comparatively accurate effect that produces synchronously.

Figure 14 is the present invention and the traditional performance comparison of canonical symbol training sequence aspect the integer frequency bias estimated error rate.The parameter of traditional canonical symbol training sequence still is N _a=1024, L _a=64, N=1024, L=64, N _Sym=1088, its decimal frequency bias algorithm for estimating and the present invention are identical, its integer frequency bias algorithm for estimating is also similar with the present invention, thereby computation complexity is also basic identical, difference is that traditional canonical symbol training sequence can't adopt " remainder theorem algorithm ", and its through type (12) and formula (13) obtain f _aWith f _nAfterwards, form the equation group of formula (14), then directly with f _aWith f _nDo on average to obtain the estimated value of integer frequency bias.Figure 14 shows, integer frequency bias estimated performance of the present invention is better than traditional scheme and surpasses 6dB, this mainly be of the present invention " remainder theorem algorithm " that integer frequency bias is estimated is comparatively accurate.

Figure 15 is that the present invention and the performance of frequency domain algorithm for estimating aspect the integer frequency bias estimated error rate compare.The specific implementation method of frequency domain algorithm for estimating is referring to document: Schmidl, T.M. etc. " Low-overhead; low-complexity[burst] synchronization for OFDM ", IEEE InternationalConference on Communications, VolUme 3, June 1996, Page (s): 1301-1306 (" the OFDM method for synchronous of low accessing cost for data, low complex degree " IEEE international communication technical conference).Owing to only investigate the frequency offset estimating performance, so, make the time synchronized of two kinds of methods entirely true, and it is identical to obtain the method for decimal frequency bias estimated value and frequency offset estimation range and the present invention in this frequency domain algorithm for estimating.Figure 15 shows, integer frequency bias estimated performance of the present invention is almost identical with the frequency domain algorithm for estimating, and computation complexity of the present invention is well below this frequency domain algorithm for estimating, and under current parameter condition, computation complexity of the present invention is having only 5% of this frequency domain algorithm for estimating approximately.

Figure 16 be the present invention when SNR=2dB, the error rate flatness simulation result in frequency offset estimation range, this figure show, although non-positive then symbol S _aIntroduced evaluated error error (ξ _a), but it estimates almost not influence to integer frequency bias.

Simulation result shows, it is lower that the present invention has a computation complexity, and the lower advantage of estimated error rate, has very high using value in ofdm system.

Claims (5)

1. method for synchronous based on the OFDM synchronous training sequence, this OFDM synchronous training sequence is generated by 2 OFDM symbol cascades, and described 2 OFDM symbols are respectively non-canonical OFDM symbol S _aWith canonical OFDM symbol S _n, described non-positive then symbol S _aCounting of significant character part be N _a, N _a≠ N, N _aContain odd prime factor q, described non-positive then symbol S _aBe divided into L _aFive equilibrium, L _aContain odd prime factor q, described each five equilibrium contains M _aPoint, wherein M _a=N _a/ L _a, and N _a, L _a, M _aBe integer, S _aCyclic Prefix partly contain to count and be N _Ag, N _Ag=N _Sym-N _a, N _SymBe always counting of OFDM symbol; Described positive then symbol S _nBe divided into the L five equilibrium, L=2 ^v, described each five equilibrium contains the M point, and wherein M=N/L unless N is the canonical outer symbol, comprises that the significant character of other OFDM symbols of positive then symbol is counted, and N, L, M be integer, it is characterized in that:

Described method for synchronous is synchronous by thick timing successively, and decimal frequency bias estimates that integer frequency bias estimates that thin timing is synchronous, finishes the estimation to the accurate original position of OFDM symbol, comprises the steps:

Step 1: to non-positive then symbol S _aDoing time-delay is the auto-correlation of a five equilibrium signal length, gets P _a(d); Align then symbol S _nDoing time-delay is the auto-correlation of a five equilibrium signal length, gets P _n(d); Align then symbol S _nDoing time-delay is the auto-correlation of half significant character length, gets P _f(d), again to P _a(d), P _n(d) and P _f(d) ask the energy sum,, get average autocorrelation P (d), seek the peak value of P (d), go out the original position of OFDM symbol according to a preliminary estimate divided by the signal gross energy in the associated window

Step 2: the P that obtains according to step 1 _f(d) and

Calculate Phase angle, thereby estimate the fractional part of frequency deviation

Parity η with integer frequency bias;

Step 3: the P that obtains according to step 1 _a(d), P _n(d) and Calculate

With

Phase angle, integrating step two obtains again

And η, estimate integer frequency bias

Step 4: obtain according to step 1

Extract positive then symbol S _nThe 2nd wait sub-signal, then it is carried out compensate of frequency deviation, again itself and local positive then symbol S _nThe 2nd five equilibrium sample of signal to make cyclic shift relevant, seek relevant peaks, thereby estimate the accurate original position of OFDM symbol

2. the method for synchronous based on the OFDM synchronous training sequence according to claim 1 is characterized in that: described non-positive then symbol S _a, its generation method is to choose N _aBe L _aInteger multiple in the immediate natural number of N

Then, every interval L on frequency domain _aIndividual subcarrier inserts data, carries out N _aThe point Fourier inversion is added N again _AgThe dot cycle prefix.

3. the method for synchronous based on the OFDM synchronous training sequence according to claim 1 is characterized in that described step 1 is to S _aDo time-delay and be M _aPoint, window is long to be (N _Sym-M _a) point auto-correlation, P _a(d); To S _nDo time-delay and be the M point, window is long to be (N _Sym-M) auto-correlation, P _n(d); To S _nDo time-delay and be the LM/2 point, window is long to be (N _Sym-LM/2) auto-correlation, P _f(d).

4. the method for synchronous based on the OFDM synchronous training sequence according to claim 1 is characterized in that described step 1 is to P _a(d), P _n(d) and P _f(d) ask the energy sum,, get average autocorrelation P (d) again divided by the signal gross energy in the associated window,

Wherein, r _{I, m}Represent i symbol, the data of m sampled point.

5. the method for synchronous based on the OFDM synchronous training sequence according to claim 1 is characterized in that, described step 3 is estimated integer frequency bias

The time, earlier by To the integer frequency bias f that makes an estimate _a,

Again by

To the integer frequency bias f that makes an estimate _n,

Obtain Indeterminate Equation Group about integer frequency bias:

At last, the remainder theorem in the utilization number theory is separated this Indeterminate Equation Group, thereby is estimated integer frequency bias

Images