CN102882670A - Synchronous processing method based on CMMB signals - Google Patents

Abstract

The invention provides a synchronous processing method based on new CMMB (China Mobile Multimedia Broadcasting) synchronous signals. The CMMB synchronous signals comprise first training sequences and second training sequences; the first training sequences comprise CAZAC (Constant Amplitude Zero Auto Correlation) sequences, and the second training sequences comprise PN (Pseudo-Noise) sequences; in the synchronous processing method, the CAZAC sequences are utilized to achieve coarse symbol timing offset estimation and decimal frequency offset coarse estimation; the PN sequences are utilized to achieve the estimation of strongest path time delay in multipath, and the multipath is taken as coarse symbol timing positioning; the CAZAC sequences and the PN sequences are utilized to achieve the integer frequency offset estimation; the fast Fourier transform (FFT) is performed on the PN sequences to achieve the channel response estimation so as to estimate a first path time delay and achieve the fine symbol timing position estimation; and a maximum likelihood (ML) criterion is utilized to process the PN sequences in the second training sequences to obtain the decimal frequency offset fine estimation. The synchronous processing method can effectively improve the synchronous accuracy and can achieve better synchronization performance in the mobile communication environment with low signal-to-noise ratio.

Description

A kind of synchronization processing method based on the CMMB signal

Technical field

The present invention relates to the radio communication simultaneous techniques, relate in particular to the orthogonal frequency division multiplex OFDM communication system in the radio communication.

Technical background

Orthogonal frequency division multiplex OFDM is a kind of high speed transmission technology, be characterized in single high-speed data-flow parallel transmission on one group of low speed orthogonal sub-carriers, it has availability of frequency spectrum height, and anti-frequency selective fading is fit to high speed data transfers and realizes simple characteristics.Compare with single-carrier system, it is very sensitive to time and frequency deviation, and therefore in order to eliminate intersymbol interference and inter-carrier interference, the synchronous and accurate Frequency Synchronization of correct time is the prerequisite of ofdm system normal operation.The frequency deviation of receiving terminal is carried out normalization according to the interval of subcarrier, fractional part of frequency offset and integer frequency offset have just occurred.For ofdm system, require frequency offseting value to be not more than 4% of subcarrier spacing at awgn channel; Require frequency offseting value to be not more than 2 ~ 3% of subcarrier spacing for fading channel, require the Symbol Timing position to be in the ISI-FREE interval.

The simultaneous techniques of ofdm system is divided into two large classes at present: the method for the method of data auxiliary (DA) and non-data auxiliary (NDA).The method of estimation of data auxiliary (DA) is to utilize the method for synchronizing sequence (training sequence), and computational complexity is lower, and estimation range is larger, but can increase overhead.The method of non-data auxiliary (NDA) is to utilize the CMMB(China Mobile multimedia broadcasting) self character of signal and the statistical property of receive data finish synchronous estimation, estimate comprising blind estimation, half-blindness, but its computational complexity is higher, and estimation range is limited.

The synchronizing signal that the CMMB signal has utilized the PN sequence to consist of finish regularly and frequency difference synchronous, estimation range is wider, and computational complexity is low, but its PAR(papr) higher, synchronous effect is not good in the situation of low signal-to-noise ratio, and the frequency deviation estimated performance is bad.

In mobile telecommunication channel, the time selective fading that the frequency selective fading that multipath effect is brought and doppler spread bring has all proposed certain requirement to synchronous.Because fractional part of frequency offset can cause the ICI(inter-carrier interference), in order to guarantee synchronous effect, we need to guarantee to finish in the first step estimation of fractional part of frequency offset, and guarantee certain precision.

Existing synchronization signal designs and corresponding synchronized algorithm mainly contain following several:

1, SCA algorithm

The SCA algorithm is that estimation and the frequency deviation of utilizing two training sequences to finish Symbol Timing estimate that its frequency deviation estimated performance is better, but because the platform effect of SCA algorithm, variance is very large aspect timing, and timing estimation results easily is in beyond the ISE-FREE interval.Frequency offset estimation range is

Wherein Δ f is the subcarrier spacing of CMMB signal.

2, Morelli algorithm

The Morelli algorithm is to utilize J identical part of transmission in a CMMB symbol, and frequency offset estimation accuracy is improved to some extent, but its frequency offset estimation range is limited, and estimation range is In order to increase estimation range, can increase the J value.It is too large that but the value of J can not be got, because reliable in order to guarantee estimated result, must guarantee that the length of each identical data block of estimation must be expanded greater than channel delay, and the timing variance of this algorithm is bigger than normal in addition.

3, Ren algorithm

Utilize a CAZAC sequence to finish the synchronous of timing offset and frequency deviation in time domain, training symbol comprises two identical CAZAC sequences, and second portion adopts the weighting of PN sequence.Because it utilizes the difference cross-correlation of PN sequence to finish regularly, so its timing effect in the ISI of slow fading channel is fine, in addition, its PAR is very low, but under separate fading channel, because the decline of difference correlation performance can reduce estimated performance; And because the utilization of weighting PN sequence causes the distortion of synchronizing signal structural nonlinear, cause the fractional part of frequency offset estimated performance to descend.

4, Adegbenga B algorithm

Utilize a training symbol, consisted of by two identical parts.Whole synchronizing process is finished in time domain.Utilize cross-correlation and the combination of auto-correlation, band limit to finish the synchronous of whole timing and frequency based on the detection of thresholding, operand is lower.But the estimation of fractional part of frequency offset is in the situation of low signal-to-noise ratio, and estimation effect is relatively poor.In fading channel, there is reduction to a certain degree in its effect.The algorithm of this paper just is based on the improvement of this algorithm, and the synchronizing signal of CMMB signal is exactly an example of Adegbenga B training symbol, and the structure of this CMMB synchronizing signal is seen accompanying drawing 1.

Summary of the invention

Technical problem to be solved by this invention is for guaranteeing CMMB signal Timing Synchronization precision and Frequency Synchronization precision under the fading channel of low signal-to-noise ratio, to provide a kind of synchronization processing method based on new CMMB synchronizing signal.

The present invention is that the technical scheme that the above-mentioned technology of solution adopts is, a kind of synchronization processing method based on the CMMB signal, utilize the CMMB synchronizing signal to the fraction frequency offset of CMMB signal, thereby the zero-time of integer-times frequency offset and CMMB signal estimates to finish synchronous processing, described CMMB synchronizing signal is comprised of the first training sequence and the second training sequence, described the first training sequence comprises that length is that permanent envelope zero auto-correlation CAZAC sequence and the length of N is the Cyclic Prefix of GI, and described the second training sequence comprises that length is that PN sequence and the length of N is the Cyclic Prefix of GI;

The concrete steps of processing synchronously are as follows:

1) utilizing in the first training sequence the CAZAC sequence to finish thick timing slip estimates and the fractional part of frequency offset rough estimate

2) use the fractional part of frequency offset rough estimate

Receiving sequence is carried out the fractional part of frequency offset compensation, utilize the PN sequence after fractional part of frequency offset compensates to finish the strongest path delay of time in the multipath

Estimation, will the strongest path delay of time As thick timing position;

3) use thick timing position And fractional part of frequency offset rough estimate

Receiving sequence is compensated, utilize CAZAC sequence and PN sequence after compensating to finish integer-times frequency offset

Estimation, with integer-times frequency offset Integer-times frequency offset as the CMMB signal;

4) use integer-times frequency offset

Receiving sequence after the compensation is carried out the integer multiple frequency compensation again, PN sequence in the second training sequence after the integer multiple frequency compensation is carried out fast Fourier transform FFT finish the channel response estimation, estimate the time delay in article one path

And finish the estimation of smart timing position θ,

With the zero-time of smart timing position θ as the CMMB signal;

5) the smart timing position θ of use upgrades the receiving sequence after integer multiple frequency compensates, and the PN sequence utilizes maximum likelihood ML criterion to obtain the smart estimation of fraction frequency offset in the second training sequence after essence is regularly upgraded With the smart estimation of fraction frequency offset Fraction frequency offset as the CMMB signal;

11) utilize the zero-time of fraction frequency offset, integer-times frequency offset and the CMMB signal of the CMMB signal of determining to finish synchronous processing.

The present invention is in order to lower the PAR of synchronizing signal, the synchronizing signal that the PN sequence of employing CAZAC sequence and mould value consists of the CMMB signal reaches less PAR, simultaneously better in order to guarantee the frequency deviation estimation effect, utilized the ML criterion in certain scope, to obtain preferably estimated value by search in the rear end.

Concrete, CAZAC sequence s in the first training sequence _T1Be expressed as:

s _t1=[c(n),c(n)]n=0,1,...,(N/2)-1；

Wherein, 2 identical sequence c (n) of formation CAZAC sequence are c (n)=exp (j π n ²/ (N/2)), the exponential function of exp () expression take natural logrithm e the end of as, N is the number of subcarrier.

Concrete, PN sequence s in the second training sequence _T2Be expressed as:

$s_{t2}=[\mathrm{pn}\left(0\right),...,\mathrm{pn}(N-1)],\mathrm{pn}\left(i\right)\∈\{\frac{1}{\sqrt{2}}(1+i),\frac{1}{\sqrt{2}}(1-i),\frac{1}{\sqrt{2}}(-1-i),\frac{1}{\sqrt{2}}(-1+i)\}$

Wherein, i=0 ..., N-1.

Concrete, the concrete grammar of step 1) is:

Utilize the auto-correlation of the identical repeating part of CAZAC sequence to carry out the search of peak value platform, all are called thick timing range greater than the set that 0.95 times timing position of maximum related value consists of, get the thick timing estimation value of the average conduct θ of all values in the thick timing range _Opt

Utilize thick timing estimation value θ _OptThe phase estimation of the correlation of correspondence position goes out thick fractional part of frequency offset

Finish thick Symbol Timing and thick fractional part of frequency offset estimation

Wherein, P () is phase function, and angle () is angle function.

Concrete, step 2) utilize the PN sequence after the fractional part of frequency offset compensation to finish the strongest path delay of time in the multipath in

Estimation, concrete grammar is: utilize the PN sequence after the fractional part of frequency offset compensation of known PN synchronizing sequence and reception to finish cross-correlation, the position that the peak value of search cross correlation results occurs obtains the strongest path delay of time

Concrete, step 3) utilize CAZAC sequence and PN sequence after the compensation to finish integer-times frequency offset in

Estimation, concrete grammar is:

Utilize to receive and compensation after the CAZAC sequence and the frequency domain data of PN sequence and the frequency domain data of known CAZAC sequence and PN sequence slide relevant, thereby the acquisition integer-times frequency offset

Concrete, the concrete grammar that utilizes PN sequence in the second training sequence after the integer multiple frequency compensation to obtain smart timing position θ in the step 4) is:

PN sequence in the second training sequence after the integer multiple frequency compensation is carried out fast Fourier transform FFT obtain the channel response estimation, obtain the channel time domain response through fast Fourier transform IFFT again, utilize the energy criteria of channel as first path delay of time in the judgement amount estimation multipath

Finish the estimation of smart timing position θ,

The invention has the beneficial effects as follows that when the low signal-to-noise ratio mobile communication environment energy Effective Raise synchronization accuracy reaches better net synchronization capability.

Description of drawings

Fig. 1 is the PN synchronizing signal structure chart of CMMB;

Fig. 2 is the structure chart of CMMB synchronizing signal of the present invention;

Fig. 3 is embodiment of the invention flow chart;

Fig. 4 is the timing estimation performance comparison figure of the PN synchronizing signal of the existing CMMB of CMMB synchronizing signal of the present invention.

Fig. 5 is the frequency deviation estimated performance comparison diagram of the PN synchronizing signal of CMMB synchronizing signal of the present invention and existing CMMB.

Embodiment

The structure of CMMB synchronizing signal of the present invention as shown in Figure 2, formed by the first training sequence and the second training sequence, described the first training sequence comprises that length is that permanent envelope zero auto-correlation CAZAC sequence and the length of N is the cyclic prefix CP of GI, and described the second training sequence comprises that length is that PN sequence and the length of N is the cyclic prefix CP of GI.It is also identical that the part of same grayscale represents its value.

The embodiment flow process as shown in Figure 3, the method for specific embodiment may further comprise the steps:

1. utilize the character of CAZAC Sequence in the first training sequence to finish thick timing estimation and decimal times frequency difference rough estimate:

If a frame signal series of discrete that receives is expressed as r (n), n=1,2 ..., N.

Utilize the auto-correlation of the identical repeating part of CAZAC sequence to carry out the search of peak value platform, all are called thick timing range greater than the set that 0.95 times timing position of maximum related value consists of, get the thick timing estimation value of the average conduct θ of all values in the thick timing range _Opt:

θ _opt=mean(U)

Wherein, mean () expression is averaged, and U is the peak value flat roof area;

U={d|abs (M (d)) 〉=0.95C ₁, d ∈ { 1,2 ..., Q } }; Wherein, Q is the length of receiving sequence, and amplitude function is asked in abs () expression;

${\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">C</figure-callout>}}_1=\arg \underset{d}{\max }\left(M\left(d\right)\right),$ M(d)=｜P(d)｜ ²/(R(d)) ²， $R\left(d\right)=\frac{1}{2}\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{\left|r(d+k)\right|}^2,$ $P\left(d\right)=\underset{k=0}{\overset{N/2-1}{\mathrm{\&Sigma;}}}r(d+k)r^{*}(d+k+N/2);$

Represent d value corresponding when returning M (d) is maximum;

Utilize thick timing estimation value θ _OptThe phase estimation of the correlation of correspondence position goes out thick fractional part of frequency offset

${\widehat{\mathrm{\&epsiv;}}}_F=\mathrm{angle}\left(P\left({\mathrm{\&theta;}}_{\mathrm{opt}}\right)\right)/\mathrm{\&pi;}$

Wherein, angle () is for asking angle function, and P () is for asking phase function.

2. utilize the character of PN Sequence in the second training sequence to finish the strongest path delay of time in the multipath Estimation:

Use the fractional part of frequency offset rough estimate

Receiving sequence r (n) is carried out the fractional part of frequency offset compensation, obtain the sequence r after fractional part of frequency offset compensates _Comp(n):

$t_{\mathrm{comp}}\left(n\right)=r\left(n\right)\&CenterDot;e^{-j2\mathrm{\&pi;}{\widehat{\mathrm{\&epsiv;}}}_{F}n},n=\{\mathrm{1,2}...,Q\}$

Utilize the PN sequence after the fractional part of frequency offset compensation of known PN synchronizing sequence and reception to finish cross-correlation, the position that the peak value of search cross correlation results occurs obtains the strongest path delay of time

1) choose a value k from peak value flat roof area U, namely k ∈ U tries to achieve W (k, n) and E (k).W (k, n) be illustrated in timing position k constantly the carrying out that receive of n the correlation of the PN sequence after the fractional part of frequency offset compensation with known PN sequence, the carrying out that n received when E (k) was illustrated in timing position k the PN sequence of fractional part of frequency offset after compensating and the correlation of known PN sequence:

$W(k,n)=r_{\mathrm{comp}}(k+\mathrm{GI}+N+n)\&times;s_{t_2}\left(n\right),n=\{\mathrm{0,1},...,N-1\}$

Wherein, N is the PN sequence length, and GI is the length of cyclic prefix CP,

Be the second known training sequence, * expression data dot product;

$E\left(k\right)=\underset{n=0}{\overset{N-2}{\mathrm{\&Sigma;}}}{W}^{*}(k,n)\&times;W(k,n+1);$

2) utilize correlation W (k, n) and E (k) to obtain the strongest path delay of time in the multipath Estimation as thick sync bit: $\widehat{\mathrm{\&theta;}}=\arg \underset{k}{\max }\left({\left|E\left(k\right)\right|}^{2}M\left(k\right)\right).$

3. utilize CAZAC and PN synchronizing sequence to finish integer-times frequency offset

Estimate:

Thick sync bit is moved forward λ _cIndividual position:

$\widehat{\mathrm{\&theta;}}=\widehat{\mathrm{\&theta;}}-{\mathrm{\&lambda;}}_c$

λ herein _cValue be an empirical value, determine according to the relation between maximum multipath time delay and the circulating prefix-length.Generally get 1/4 of length of the cycle.

Use thick timing position And fractional part of frequency offset rough estimate

Receiving sequence is compensated, obtain the time domain PN sequence after the decimal overtones band compensates With the CAZAC sequence

$r_{{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}=r_{\mathrm{comp}}(n+\widehat{\mathrm{\&theta;}});$ $r_{t_2}=r_{\mathrm{comp}}(n+\widehat{\mathrm{\&theta;}}+N+\mathrm{GI}),n=\mathrm{0,1},...N-1;$

To time domain CAZAC sequence

With the PN sequence

And known CAZAC sequence

With the PN sequence Carry out FFT:

${\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">T</figure-callout>}}_1=\mathrm{FFT}\left(s_{t_1}\right)$ $T_2=\mathrm{FFT}\left(s_{t_2}\right)$

${\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">T</figure-callout>}}_1^{+}=\mathrm{FFT}\left(r_{t_1}\right)$ $T_2^{+}=\mathrm{FFT}\left(r_{t_2}\right)$

CAZAC sequence after receiving and compensating

With the PN sequence

Frequency domain data and known CAZAC sequence

With the PN sequence

Frequency domain data slide relevant, thereby obtain integer-times frequency offset

$G\left(m\right)=\frac{{\left|\underset{k\&Element;C_{\mathrm{EVEN}}}{\mathrm{\&Sigma;}}{\left({\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">T</figure-callout>}}_1^{+}(k+2m)\right)}^{*}{\left(V\left(k\right)\right)}^{*}{T}_2^{+}(k+2m)\right|}^2}{2{\left(\underset{k\&Element;C_{\mathrm{EVEN}}}{\mathrm{\&Sigma;}}{\left|T_2^{+}(k+2m)\right|}^2\right)}^2},m=[0,\frac{N}{2}]$

Wherein, V (k) is known CAZAC sequence and the relation between the PN sequence frequency domain

$V\left(k\right)=\left\{\begin{array}{c}\frac{T_2\left(k\right)}{T_1\left(k\right)},k\&Element;C_{\mathrm{EVEN}},C_{\mathrm{EVEN}}=\{\mathrm{0,2,4},...,N-2\}\\ 0,k\&Element;C_{\mathrm{Odd}},C_{\mathrm{Odd}}=\{\mathrm{1,3,5},...,N-1\}\end{array}\right..$

Final integer frequency offset estimation result is:

${\widehat{\mathrm{\&epsiv;}}}_c=2\arg \underset{m}{\max }G\left(m\right)$

Obtaining thick Frequency Estimation result is

4. utilize the second training sequence FFT to finish channel response and estimate, estimate the time delay in article one path and finish regularly smart.

Use integer-times frequency offset

Receiving sequence after the compensation is carried out the integer multiple frequency compensation again obtain sequence r _{Comp_new}:

$r_{\mathrm{comp}\_\mathrm{new}}=r_{\mathrm{comp}}\left(n\right)\&times;e^{-j2\mathrm{\&pi;}{\widehat{\mathrm{\&epsiv;}}}_{c}n},n=\mathrm{1,2},...,Q;$

The PN sequence table is shown:

$r_{t_{2\mathrm{new}}}=r_{\mathrm{comp}\_\mathrm{new}}(n+\widehat{\mathrm{\&theta;}}+N+\mathrm{GI}),n=\mathrm{0,1},...N-1$

PN sequence in the second training sequence after the integer multiple frequency compensation is carried out fast Fourier transform FFT obtain the channel response estimation;

$Y=\mathrm{FFT}\left(r_{t_{2\mathrm{new}}}\right);$

Channel frequency domain estimate vector is:

$\hat{H}=\frac{Y}{T_2}=[H\left(0\right)e^{j2\mathrm{\&pi;}{\mathrm{<figure-callout\; id="0"\; label="f"\; filenames="HDA00002131764100031.png,HDA00002131764100032.png"\; state="\{\{state\}\}">f</figure-callout>}}_0\mathrm{\&tau;}},...,H\left(k\right)e^{j2\mathrm{\&pi;}{f}_k\mathrm{\&tau;}},....H(N-1)e^{j2\mathrm{\&pi;}{f}_{N-1}\mathrm{\&tau;}}];$

Again through fast Fourier transform IFFT,

The response of acquisition channel time domain:

IFFT obtains time domain channel shock response and is estimated as:

$\hat{h}\left(i\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\hat{H}\left(k\right)e^{-j2\mathrm{\&pi;}\frac{\mathrm{kn}}{N}}i=\mathrm{1,2}...,N$

Estimate to obtain the strongest path

Amplitude, utilize certain window long channel energy to determine the time delay in article one path as decision rule.Article one, the estimation function in path can be expressed as:

$\widehat{\mathrm{\&tau;}}=\arg \underset{l}{\max }\{E_h\left(l\right):l=\mathrm{0,1},...,\mathrm{GI}-K^{+}\}$

In the formula $E_h\left(l\right)=\left\{\begin{array}{cc}\underset{k=0}{\overset{K^{+}-1}{\mathrm{\&Sigma;}}}{\left|{\hat{h}}_{l+k}\right|}^2,& \mathrm{if}\left|{\hat{h}}_l\right|>\mathrm{\&sigma;}\&CenterDot;\left|{\hat{h}}_{\max }\right|\\ 0,& \mathrm{otherwise}\end{array}\right.$

E in the formula _h(l) expression is long from the window of l beginning is K ⁺Channel energy, channel delay expansion L＜K ⁺σ is for selecting the thresholding in article one path.

In the estimated position that obtains article one path

After, we can obtain smart timing position θ and are:

$\mathrm{\&theta;}=\widehat{\mathrm{\&theta;}}+\widehat{\mathrm{\&tau;}}.$

5. utilize the ML criterion to obtain the smart estimation of fractional part of frequency offset:

The θ that was obtained by the upper step upgrades receive data.With

The data of the PN sequence after representing to upgrade:

$r_{t_{2\mathrm{new}}}=r_{\mathrm{comp}\_\mathrm{new}}(n+\mathrm{\&theta;}+N+\mathrm{GI}),n=\mathrm{0,1},...N-1$

At this moment, the PN sequence after the renewal that obtains of receiving terminal is

The PN sequence that known transmitting terminal sends is

The PN sequence utilizes maximum likelihood ML criterion to obtain that fraction frequency offset is smart to be estimated in the second training sequence after essence regularly upgraded

Obtaining the smart cost function of estimating of fractional part of frequency offset is:

$\widehat{\mathrm{\&epsiv;}}=\arg \underset{\widehat{\mathrm{\&epsiv;}}}{\max }\{r_{t_{2\mathrm{new}}}W\left(\widehat{\mathrm{\&epsiv;}}\right){\mathrm{BW}}^H\left(\widehat{\mathrm{\&epsiv;}}\right){\left(r_{t_{2\mathrm{new}}}\right)}^H:\widehat{\mathrm{\&epsiv;}}={\widehat{\mathrm{\&epsiv;}}}_1+\mathrm{<figure-callout\; id="10"\; label="m\; F"\; filenames="HDA00002131764100031.png,HDA00002131764100032.png"\; state="\{\{state\}\}">m</figure-callout>}\frac{F}{}\frac{}{10},m=\{-10,-9,...\mathrm{9,10}\left\}\right\}$ In the following formula: Be the rough estimate that obtains in the step before, B=S (S ^HS) ^-1S ^H

W(ε)=diag{1,e ^j2πε/N,e ^j2π2ε/N,…,e ^{j2π(N-1)ε/N}}

Diag () is the diagonal matrix function, and F represents decimal overtones band hunting zone, can suitably adjust according to the precision that thick frequency deviation is estimated the size of F value, is taken as generally speaking 0.005.

Emulation experiment is considered the applied environment of CMMB signal take Chengdu ambulatory handheld TV as example, and the parameter of signal is: carrier frequency is f _c=506MHz, emulation 1 frame CMMB, 1 frame is divided into 40 time slots, the length 25ms of each time slot.CMMB synchronizing signal time span is 204.8 μ s, and the length of each CMMB symbol is 460.8us.The length of channel maximum delay expansion is L=50, protection interval GI=128, carrier wave number N=1024, frequency deviation e=6.4.The Simulated movable communication channel adopts ITU-M.1225 Vehicle Channel B channel (to consider speed v ₂During=60km/h) condition, signal length is P=(N+GI) * 80=46080.Channel condition sees the following form:

The method of specific embodiment may further comprise the steps:

A) utilize the character of CAZAC Sequence the first training sequence to finish thick timing estimation and decimal times frequency difference rough estimate from receiving signal.The thick timing slip of symbol estimates that expression formula is:

θ _opt=mean(U)

Wherein: the average of all values among the U is asked in mean (U) expression, and U is the peak value flat roof area.

U={d?|abs(M(d))≥0.95C ₁,d∈{1,2…,P}}

M(d)=｜P(d)｜ ²/(R(d)) ² ${\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">C</figure-callout>}}_1=\arg \underset{d}{\max }\left(M\left(d\right)\right)$

$P\left(d\right)=\underset{k=0}{\overset{511}{\mathrm{\&Sigma;}}}r(d+k)r^{*}(d+k+512)$ $R\left(d\right)=\frac{1}{2}\underset{k=0}{\overset{1023}{\mathrm{\&Sigma;}}}{\left|r(d+k)\right|}^2$

The rough estimate of decimal overtones band deviation Utilize Symbol Timing position θ _OptP (θ _Opt) obtain, be shown below.

${\widehat{\mathrm{\&epsiv;}}}_F=\mathrm{angle}\left(P\left({\mathrm{\&theta;}}_{\mathrm{opt}}\right)\right)/\mathrm{\&pi;}$

B) use the fractional part of frequency offset rough estimate

Receiving sequence r (n) is carried out the fractional part of frequency offset compensation, obtain the sequence r after fractional part of frequency offset compensates _Comp(n).

Utilize the PN sequence after fractional part of frequency offset compensates to finish the strongest path delay of time in the multipath

Estimation, concrete grammar is: utilize the PN sequence after the fractional part of frequency offset compensation of known PN synchronizing sequence and reception to finish cross-correlation, the position that the peak value of search cross correlation results occurs obtains the strongest path delay of time Utilize the character of PN Sequence in the second training sequence to finish the estimation in the strong path delay of time in the multipath

To receive signal r (n) and finish decimal times frequency difference compensation, obtain sequence r _Comp(n).

$r_{\mathrm{comp}}\left(n\right)=r\left(n\right)\&CenterDot;e^{-j2\mathrm{\&pi;}{\widehat{\mathrm{\&epsiv;}}}_{F}n},n=\{\mathrm{1,2}...,P\}$

Time delay in strong path in the multipath so

Can obtain according to the following step.

1) from scope U, chooses a value k, i.e. k ∈ U.

$W(k,n)=r_{\mathrm{comp}}(k+128+1024+n)\&times;s_{t_2}\left(n\right),n=\{\mathrm{0,1},...,1023\}$

2) $E\left(k\right)=\underset{n=0}{\overset{1022}{\mathrm{\&Sigma;}}}{W}^{*}(k,n)\&times;W(k,n+1)$

3) $\widehat{\mathrm{\&theta;}}=\arg \underset{k}{\max }\left({\left|E\left(k\right)\right|}^{2}M\left(k\right)\right)$

C) utilize CAZAC and PN synchronizing sequence to finish integer-times frequency offset

Estimate.

Thick sync bit is moved forward λ _c=20 positions, namely Carry out the integral multiple frequency difference according to the method that is similar to SCA

Estimation.

$r_{{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}=r_{\mathrm{comp}}(n+\widehat{\mathrm{\&theta;}});r_{t_2}=r_{\mathrm{comp}}(n+\widehat{\mathrm{\&theta;}}+1024+128),n=\mathrm{0,1},...1023$

${\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">T</figure-callout>}}_1=\mathrm{FFT}\left(s_{t_1}\right)$ $T_2=\mathrm{FFT}\left(s_{t_2}\right)$

${\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">T</figure-callout>}}_1^{+}=\mathrm{FFT}\left(r_{t_1}\right)$ $T_2^{+}=\mathrm{FFT}\left(r_{t_2}\right)$

$V\left(k\right)=\left\{\begin{array}{c}\frac{T_2\left(k\right)}{T_1\left(k\right)},k\&Element;C_{\mathrm{EVEN}},C_{\mathrm{EVEN}}=\{\mathrm{0,2,4},...,1022\}\\ 0,k\&Element;C_{\mathrm{Odd}},C_{\mathrm{Odd}}=\{\mathrm{1,3,5},...,1023\}\end{array}\right.$

Obtaining integer multiple frequency estimation of deviation expression is:

$G\left(m\right)=\frac{{\left|\underset{k\&Element;C_{\mathrm{EVEN}}}{\mathrm{\&Sigma;}}{\left({\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">T</figure-callout>}}_1^{+}(k+2m)\right)}^{*}{\left(V\left(k\right)\right)}^{*}{T}_2^{+}(k+2m)\right|}^2}{2{\left(\underset{k\&Element;C_{\mathrm{EVEN}}}{\mathrm{\&Sigma;}}{\left|T_2^{+}(k+2m)\right|}^2\right)}^2},$ m=[0,512],C _EVEN={0,2,4,…,1022}

Obtaining final integer multiple frequency estimation of deviation result is:

${\widehat{\mathrm{\&epsiv;}}}_c=2\arg \underset{m}{\max }G\left(m\right)$

D) utilize FFT to finish channel response and estimate, estimate the time delay in article one path and finish regularly smart.

Utilize FFT to obtain to need the value of suitable selection σ after the channel time domain impulse response estimation.σ is for selecting the thresholding in article one path, and σ can not be too large, may cause undetected the first path too greatly.The value of σ can not be too little, and too little can the increase noise be the possibility of article one of multipath by flase drop.Choosing of σ value can be carried out according to following formula:

${\mathrm{\&sigma;}}_2=\sqrt{\frac{{\mathrm{<figure-callout\; id="1"\; label="SNR"\; filenames="HDA00002131764100032.png"\; state="\{\{state\}\}">SNR</figure-callout>}}_1}{{\mathrm{SNR}}_2}}{\mathrm{\&sigma;}}_1$

σ ₁To be SNR in signal to noise ratio ₁The time emulation reference value that obtains, σ ₂For being SNR in signal to noise ratio ₂Situation under the threshold value that obtains.

Obtain at SNR by emulation ₁σ during=10dB ₁=0.05, can get

Consider the characteristics of actual mobile telecommunication channel, get the long K of window ⁺=54, step 4 can estimate smart timing position to specifications.

E) utilize the ML criterion to obtain the smart estimation of decimal overtones band deviation.

The θ that was obtained by the upper step upgrades the data of PN sequence

$r_{t_{2\mathrm{new}}}=r_{\mathrm{comp}\_\mathrm{new}}(n+\mathrm{\&theta;}+1024+128),n=\mathrm{0,1},...1023$

The PN sequence of known transmitting terminal emission is

The value of frequency difference hunting zone is set to F=0.005, utilizes so the ML criterion can obtain the smart cost function of estimating of decimal overtones band deviation to be:

$\widehat{\mathrm{\&epsiv;}}=\arg \underset{\widehat{\mathrm{\&epsiv;}}}{\max }\{r_{t_{2\mathrm{new}}}W\left(\widehat{\mathrm{\&epsiv;}}\right){\mathrm{BW}}^H\left(\widehat{\mathrm{\&epsiv;}}\right){\left(r_{t_{2\mathrm{new}}}\right)}^H:\widehat{\mathrm{\&epsiv;}}={\widehat{\mathrm{\&epsiv;}}}_1+m\frac{0.005}{10},m=\{-10,-9,...\mathrm{9,10}\left\}\right\}$ By emulation in the situation of low signal-to-noise ratio, the Symbol Timing of two kinds of synchronizing signals and frequency difference estimation performance, and the result is compared to weigh the performance of new synchronizing signal and synchronization mechanism.In comparison procedure, with the SCA algorithm also as a relatively reference of performance.

From accompanying drawing 4, we can obtain in the situation of low signal-to-noise ratio, and new synchronizing signal can reach better timing.From accompanying drawing 5, the CMMB-PN curve is frequency difference estimation MSE curve take the PN sequence as synchronizing signal, and new way fine curve is the frequency difference estimation MSE curve of new synchronizing signal.The SCA curve is the frequency difference estimation MSE curve of SCA algorithm.We can find that in the situation of low signal-to-noise ratio synchronizing signal of the present invention can reach better timing estimation and frequency deviation is estimated.

Claims (7)

1. synchronization processing method based on the CMMB signal, utilize the CMMB synchronizing signal that thereby the zero-time of fraction frequency offset, integer-times frequency offset and the CMMB signal of CMMB signal is estimated to finish synchronous processing, it is characterized in that, described CMMB synchronizing signal is comprised of the first training sequence and the second training sequence, described the first training sequence comprises that length is that permanent envelope zero auto-correlation CAZAC sequence and the length of N is the Cyclic Prefix of GI, and described the second training sequence comprises that length is that PN sequence and the length of N is the Cyclic Prefix of GI;

The concrete steps of processing synchronously are as follows:

1) utilizing in the first training sequence the CAZAC sequence to finish thick timing slip estimates and the fractional part of frequency offset rough estimate

2) use the fractional part of frequency offset rough estimate

Receiving sequence is carried out the fractional part of frequency offset compensation, utilize the PN sequence after fractional part of frequency offset compensates to finish the strongest path delay of time in the multipath

Estimation, will the strongest path delay of time As thick timing position;

3) use thick timing position

And fractional part of frequency offset rough estimate

Receiving sequence is compensated, utilize CAZAC sequence and PN sequence after compensating to finish integer-times frequency offset

Estimation, with integer-times frequency offset

Integer-times frequency offset as the CMMB signal;

4) use integer-times frequency offset

Receiving sequence after the compensation is carried out the integer multiple frequency compensation again, PN sequence in the second training sequence after the integer multiple frequency compensation is carried out fast Fourier transform FFT finish the channel response estimation, estimate the time delay in article one path

And finish the estimation of smart timing position θ,

With the zero-time of smart timing position θ as the CMMB signal;

5) the smart timing position θ of use upgrades the receiving sequence after integer multiple frequency compensates, and the PN sequence utilizes maximum likelihood ML criterion to obtain the smart estimation of fraction frequency offset in the second training sequence after essence is regularly upgraded

With the smart estimation of fraction frequency offset

Fraction frequency offset as the CMMB signal;

6) utilize the zero-time of fraction frequency offset, integer-times frequency offset and the CMMB signal of the CMMB signal of determining to finish synchronous processing.

2. a kind of synchronization processing method based on the CMMB signal as claimed in claim 1 is characterized in that CAZAC sequence s in the first training sequence _T1Be expressed as:

s _t1=[c(n),c(n)]n=0,1,...,(N/2)-1；

Wherein, 2 identical sequence c (n) of formation CAZAC sequence are c (n)=exp (j π n ²/ (N/2)), the exponential function of exp () expression take natural logrithm e the end of as, N is the number of subcarrier.

3. a kind of synchronization processing method based on the CMMB signal as claimed in claim 1 is characterized in that PN sequence s in the second training sequence _T2Be expressed as:

$s_{t2}=[\mathrm{pn}\left(0\right),...,\mathrm{pn}(N-1)],\mathrm{pn}\left(i\right)\&Element;\{\frac{1}{\sqrt{2}}(1+i),\frac{1}{\sqrt{2}}(1-i),\frac{1}{\sqrt{2}}(-1-i),\frac{1}{\sqrt{2}}(-1+i)\}$

Wherein, i=0 ..., N-1.

4. a kind of synchronization processing method based on the CMMB signal as claimed in claim 2 is characterized in that the concrete grammar of step 1) is:

Utilize the auto-correlation of the identical repeating part of CAZAC sequence to carry out the search of peak value platform, all are called thick timing range greater than the set that 0.95 times timing position of maximum related value consists of, get the thick timing estimation value of the average conduct θ of all values in the thick timing range _Opt

Utilize thick timing estimation value θ _OptThe phase estimation of the correlation of correspondence position goes out thick fractional part of frequency offset

Finish thick Symbol Timing and thick fractional part of frequency offset estimation

Wherein, P () is phase function, and angle () is angle function.

5. a kind of synchronization processing method based on the CMMB signal as claimed in claim 3 is characterized in that step 2) in utilize the PN sequence after the fractional part of frequency offset compensation to finish the strongest path delay of time in the multipath

Estimation, concrete grammar is: utilize the PN sequence after the fractional part of frequency offset compensation of known PN synchronizing sequence and reception to finish cross-correlation, the position that the peak value of search cross correlation results occurs obtains the strongest path delay of time

6. a kind of synchronization processing method based on the CMMB signal as claimed in claim 1 is characterized in that step 3) in utilize CAZAC sequence and PN sequence after the compensation to finish integer-times frequency offset

Estimation, concrete grammar is:

Utilize to receive and compensation after the CAZAC sequence and the frequency domain data of PN sequence and the frequency domain data of known CAZAC sequence and PN sequence slide relevant, thereby the acquisition integer-times frequency offset

7. a kind of synchronization processing method based on the CMMB signal as claimed in claim 3 is characterized in that, the concrete grammar that utilizes PN sequence in the second training sequence after the integer multiple frequency compensation to obtain smart timing position θ in the step 4) is:

PN sequence in the second training sequence after the integer multiple frequency compensation is carried out fast Fourier transform FFT obtain the channel response estimation, obtain the channel time domain response through fast Fourier transform IFFT again, utilize the energy criteria of channel as first path delay of time in the judgement amount estimation multipath

Finish the estimation of smart timing position θ,

Images