CN101179291B - Condition maximum likelihood estimation based ultra-wideband communication system synchronization method - Google Patents

Abstract

The invention belongs to the field of wireless communications technology and specifically relates to a synchronous method of an ultra wideband (UWB) communications system based on a maximal likelihood estimation of condition. Circulation stability existing in a UWB symbol structure is used for restraining frame level noise for received waveform and constructing a noise template required synchronously; then the invented noise template is used for estimating and obtaining a parameter of A<SUB>Xi</SUB> which represents the energy in a frame, and the parameter is user for estimating and represents a timing deviation of n<SUB>f</SUB> of the frame level; lastly, provided sliding correlative search is used for obtaining the timing deviation inside the frame of Xi. The preceding two items provided by the invention can effectively restrain the effect on noise component during parameter estimating, improve the synchronous deviation performance in mean square under the condition of a low signal-to-noise ratio and decrease the lower limit of the synchronous deviation performance in the mean square under the condition of a high signal-to-noise ratio.

Description

Ultra-wideband communication system synchronization method based on the condition maximum likelihood estimation

Technical field

The invention belongs to wireless communication technology field, be specifically related to a kind of method for synchronous that is applied to pulse system ultra wide band (IR-UWB) communication system.

Background technology

(Ultra-wideband, UWB) technology has become one of wireless communication technology that academia and industrial quarters in recent years very pay close attention to ultra broadband.Wherein, IR-UWB (Impulse Radio UWB) scheme adopts the pulse train of subnanosecond or picosecond directly to carry out baseband transmission; Avoided carrier wave frequency deviation (carrier frequency offset; CFO) caused decreased performance, simultaneity factor complexity also reduce because of need not to make carrier processing greatly.In addition, good multipath differentiability has weakened the caused energy loss of indoor multipath fading, and making the UWB signal propagate and share frequency spectrum with other communication modes with low transmitting power becomes possibility.It will be further appreciated that,, promised to undertake high channel capacity, for the short distance high-speed radiocommunication provides a new approach as the ultra-narrow pulse of orthodox ultra-broadband signal through the calculating of shannon formula.Yet just owing to adopted ultra-narrow pulse to transmit, the UWB design of communication system is faced with great challenge.Wherein, regularly the design of synchronism link is even more important.Research shows, descends even minimum timing error all might cause big bit error rate performance.Traditional U WB generates template based on the parameter that channel estimating obtains in this locality synchronously, and the slip through signal and template is relevant obtains symbol synchronization.Assessing this type algorithm from hard-wired angle can find, accomplishes precise channels by DSP and estimates that system will introduce the ADC of sample rate up to number GHz, make it carry out high-speed sampling to pulsewidth less than the pulse of 1 nanosecond; Simultaneously, a large amount of relevant operational capabilities to DSP of slip also are no small tests.Having researcher [1] to propose a kind of data, auxiliary (Data-aided, DA) mode estimates (Conditional MaximumLikelihood, CML) algorithm based on condition maximum likelihood.The significant advantage of this algorithm is to slide that relevant only to carry out obtaining the timing of frame aspect in the symbol aspect synchronous.By receive signal form NT (Noise Template, NT), and based on the criterion of general likelihood ratio test (Generalized likelihood ratio tests GLRT) carries out the estimation of CML, thereby obtains the timing offset amount of frame level.Yet there is the problem of two aspects in former algorithm.At first, this algorithm is under the situation of low signal-to-noise ratio, because noise suppressed is not very effectively, thereby all square estimated performance is unsatisfactory; Secondly, under the situation of high s/n ratio,,, there is lower limit (low bound) in theory so mean square error can not further reduce with the raising of signal to noise ratio because algorithm is not introduced smart synchronism link.

Former CML estimates that there is the shortcoming of several respects in method for synchronous:

1. the cycle of its wave-average filtering operation of former CML algorithm for estimating is a symbol period, does not make full use of limited training sequence, has influence on the formation of NT thus, receives noise effect bigger.

2. obtain the estimated value of a frame energy by NT and reception signal correction, receive signal then and do not carry out the noise suppressed processing, thereby have influence on the estimation of frame level timing offset amount.

3. former CML algorithm for estimating only comprises thick synchronism link, does not consider smart stationary problem.

Summary of the invention

The object of the present invention is to provide a kind of novel synchronization method for ultra-broadband system of estimating based on CML,, and provide one-level smart synchronism link, further dwindle timing error so that more effectively suppress noise effect.

The synchronization method for ultra-broadband system of estimating based on CML provided by the invention belongs to data supplementary mode (Data-aided).Therefore, at first will construct one group of training sequence, this training sequence is formed (shown in Fig. 3 (b)) by preamble sequence and back synchronizing sequence two parts, wherein preamble sequence (Preamble) M ₁+ M ₂(M ₁=M ₂) be used for noise template (Noise Template, generation NT), back synchronizing sequence (Post-amble) M ₃Be used for estimated frame level timing offset amount n _f

Synchronizing process is divided following step:

(1) utilizes preceding M in the preamble sequence ₁Individual symbol sebolic addressing structure noise template NT ₁, suppress the influence that noise produces through the wave-average filtering computing of frame level.Noise template NT ₁Structural formula is following:

$p_{r{f}_1}\left(t\right)=\frac{1}{M_1{N}_f}\underset{k=0}{\overset{M_1{N}_f-1}{\mathrm{\&Sigma;}}}r(t+k{T}_f)W_f\left(t\right),t\&Element;[0,T_f)$

Wherein $W_f\left(t\right)=\left\{\begin{array}{cc}1& t\&Element;[0,T_f)\\ 0& t\&Element;(-\&infin;,0)\&cup;[T_f,+\&infin;)\end{array}\right.,$ N _fBe the frame number in the single symbol, T _fBe frame period, r (t) is for being written into the reception waveform of timing offset, and t is a time variable.

Then will

With T _fFor the cycle is carried out continuation, being extended to length is T _sNT ₁The NT of neotectonics ₁As shown in the formula:

$p_{N{T}_1}\left(t\right)=p_{r{f}_1}\left(t\mathrm{mod}{T}_f\right),t\&Element;[0,T_s)$

T wherein _s=N _f* T _f, represent single mark space.

(2) utilize back M in the preamble sequence again ₂(common M ₁=M ₂) individual symbol sebolic addressing structure noise template NT ₂, suppress the influence that noise produces through the wave-average filtering computing of frame level.Template NT ₂Structural formula is following:

$p_{r{f}_2}\left(t\right)=\frac{1}{M_2{N}_f}\underset{k=M_1{N}_f}{\overset{(M_1+M_2)N_f-1}{\mathrm{\&Sigma;}}}r(t+k{T}_f)W_f\left(t\right),t\&Element;[0,T_f)$

Wherein $W_f\left(t\right)=\left\{\begin{array}{cc}1& t\&Element;[0,T_f)\\ 0& t\&Element;(-\&infin;,0)\&cup;[T_f,+\&infin;)\end{array}\right.$

Then with p _Rf2 (t) are with T _fFor the cycle is carried out continuation, being extended to length is T _sNT ₂Construct new NT ₂As shown in the formula:

$p_{N{T}_2}\left(t\right)=p_{r{f}_2}\left(t\mathrm{mod}{T}_f\right),t\&Element;[0,T_s)$

(3) NT to generating ₁And NT ₂Thereby, average and obtain noise module NT, be shown below:

$p_{\mathrm{NT}}\left(t\right)=\frac{p_{N{T}_1}\left(t\right)+p_{N{T}_2}\left(t\right)}{2}$

(4) NT to generating ₁And NT ₂Carry out noise template cross-correlation and estimate, obtain to characterize the parameter A of a frame energy _ξ, be shown below:

${\hat{A}}_{\mathrm{\&xi;}}=\frac{1}{N_f}\underset{0}{\overset{T_S}{\&Integral;}}{p}_{N{T}_1}\left(t\right)p_{N{T}_2}\left(t\right)\mathrm{dt}$

(5), utilize and estimate gained A according to the CML algorithm _ξ, NT, back synchronizing sequence (Post-amble) M3 and other known parameters of being constructed calculate frame level timing offset amount n _f, be shown below:

Wherein $y\left[n\right]:=\underset{0}{\overset{T_s}{\&Integral;}}r(t+{\mathrm{NT}}_s)p_{\mathrm{NT}}\left(t\right)\mathrm{Dt};$

(6) on the basis that obtains frame level timing offset amount

, increase regional slip relevant search (Regional SlideCorrelation Search; RSCS); Estimate ξ through catching the corresponding τ value of relevant peaks, that is:

$\widehat{\mathrm{\&xi;}}=\arg \underset{\mathrm{\&tau;}\&Element;[-T_f,T_f)}{\max }\underset{0}{\overset{T_s}{\&Integral;}}r(t+({\mathrm{<figure-callout\; id="1"\; label="M"\; filenames="435696DEST\_PATH\_H200710047406520080202E000012.png,496846DEST\_PATH\_H200710047406520080202E000011.png"\; state="\{\{state\}\}">M</figure-callout>}}_1+M_2+M_3)T_s+\mathrm{\&tau;}+{\widetilde{\mathrm{\&tau;}}}_0)p_{\mathrm{NT}}(t+\mathrm{\&tau;}{+\widehat{\mathrm{\&tau;}}}_0)\mathrm{dt}$

Wherein τ exists Front and back span be 2T _fStep-searching in the zone, and step size Δ τ (Δ τ ∈ (0, T _f]) adjust according to the requirement of precision; And for noise template NT, adopt step cycle to be shifted and cooperate the slip of correlation window on r (t), to realize the slip related operation.

(7) estimation of finally always being delayed time ${\widehat{\mathrm{\&tau;}}}_{\mathrm{Total}}=T_f{\hat{n}}_f+\widehat{\mathrm{\&xi;}}.$

Among the present invention, preamble sequence can be made up of one group of complete "+1 " sequence: "+1 ,+1 ... ", the sequence formation that back synchronizing sequence is alternately arranged by a group "+1 and-1 ": "+1 ,-1 ,+1 ,-1 ... "

Beneficial effect

The novel synchronization method for ultra-broadband system of estimating based on CML that method for synchronous provided by the invention adopts has made full use of the estimation that limited training sequence carries out energy; Adopt frame level noise suppressed means; Improve the parameter Estimation precision under the low signal-to-noise ratio situation to a great extent, thereby improved the mean square error performance effectively; In addition, it is synchronously smart to increase one-level, greatly reduces the lower limit of mean square error performance.

Description of drawings

Fig. 1: signal to noise ratio E _b/ N ₀The time-domain signal waveform of=6dB relatively.(a) make wave-average filtering with 800 symbols; (b) make wave-average filtering with 100 symbols; (c) noiseless suppresses.

Fig. 2: signal to noise ratio E _b/ N ₀The time-domain signal waveform of=6dB compares, M ₁=50, N _f=20.(a) pass through the not signal waveform of plus noise of channel; (b) signal waveform after the frame level is average; (c) signal waveform after symbol level is average; (d) noiseless suppresses.

Fig. 3: the comparison of training sequence allocation plan.(a) former training sequence allocation plan; (b) the training sequence allocation plan of noise template cross-correlation estimation.

Fig. 4: what provide is under the CM1 of IEEE802.15.3a channel, and the normalization mean square error performance of add up three kinds of different technologies and former algorithm relatively.

Embodiment

In the IR-UWB of PAM-TH system, each information bit is transmitted by a symbol (symbol), and each symbol comprises N _fFrame (frame), wherein every frame comprises a pulse, and interframe is divided into T _f, pulse width T _p＜＜T _fTherefore mark space is T _s:=N _fT _fJump time-code (Time-hopping code, TH code) when introducing, each frame is divided into N again _cIndividual chip, chip-spaced are T _cThe TH code table is shown c _j∈ [0, N _c-1],

J ∈ [1, N _f], so single symbol waveform does

$p_s\left(t\right)=\underset{j=0}{\overset{N_f-1}{\mathrm{\&Sigma;}}}p(t-j{T}_f-c{T}_c)---\left(1\right)$

Wherein p (t) is that pulsewidth is T _pThe ultra-narrow pulse waveform.

The present invention is primarily aimed at binary system PAM modulation, and symbolic representation is that { energy is ε to s [n] ∈ for ± 1}, independent equiprobability appearance _sSo the UWB transmitted waveform is:

$u\left(t\right)=\sqrt{{\mathrm{\&epsiv;}}_s}\underset{n=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}s\left[n\right]p_s(t-n{T}_s)---\left(2\right)$

Signal u (t) is through the multidiameter fading channel in L footpath, and the attenuation in every footpath is expressed as { α respectively with time-delay _lAnd { τ _l, τ wherein ₀＜τ ₁＜...＜τ _L-1And the parameter of regularly being concerned about the most synchronously is exactly the τ time of advent in first footpath ₀Thereby the expression formula of channel can be write as:

$h\left(t\right)=\underset{l=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_l\mathrm{\&delta;}(t-{\mathrm{\&tau;}}_{l,0})---\left(3\right)$

τ wherein _{L, 0}:=τ _l-τ ₀Be defined as the relative time delay of every multipath.

Thus, signal u (t) through the waveform behind the channel is:

$g_s\left(t\right)=\underset{l=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_l{p}_s(t-{\mathrm{\&tau;}}_{l,0})---\left(4\right)$

Thereby complete receiving end signal waveform is:

$r\left(t\right)=\sqrt{{\mathrm{\&epsiv;}}_s}\underset{n=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}s\left[n\right]g_s(t-n{T}_s-{\mathrm{\&tau;}}_0)+w\left(t\right)---\left(5\right)$

Wherein, w (t) obeys that average is 0, variance is σ ²Additive white Gaussian noise (Additive White GaussianNoise, AWGN).

Suppose τ with being without loss of generality ₀∈ [0, T _s), thereby can define τ ₀:=n _fT _f+ ξ, wherein n _f:=

τ ₀/ T _f

, and surplus ξ ∈ [0, T _f).

So be written into the receiving terminal waveform of delayed data be:

$r\left(t\right)=\sqrt{{\mathrm{\&epsiv;}}_s}\underset{n=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}s\left[n\right]g_s(t-n{T}_s-n_f{T}_f-\mathrm{\&xi;})+w\left(t\right)---\left(6\right)$

Through collect training sequence (therefrom calculate get timing error, find the original position of a frame in essence by Training Sequence, energy TS) for CML algorithm [1].Thereby collection of energy will directly influence the accuracy of estimation.Through the relevant energy that obtain symbol of reception signal, promptly with template

$y\left[n\right]:=\underset{0}{\overset{T_s}{\&Integral;}}r(t+{\mathrm{nT}}_s)p_{\mathrm{rs}}\left(t\right)\mathrm{dt}---\left(7\right)$

Definition by energy can know that desirable template is the waveform that matees fully with r (t).Conventional synchronization generates correlate template through channel estimating, but multi-path dense condition lower channel is estimated very expensive, replaces channel estimating to accomplish the structure of template so propose NT.Particularly, send one group of complete "+1 " sequence in the first of training sequence, because there is not the symbol-modulated problem in complete "+1 " sequence, so $E\left\{r\left(t\right)\right\}=\sqrt{{\mathrm{\&epsiv;}}_s}{\mathrm{\&Sigma;}}_{n=-\&infin;}^{\&infin;}{g}_s(t-n{T}_s-{\mathrm{\&tau;}}_0),$ Thus to continuous N ₁Individual length is T _sTraining sequence average, the desirable template that can obtain to be similar to, as shown in the formula

${\tilde{p}}_{\mathrm{rs}}\left(t\right)=\frac{1}{{\mathrm{<figure-callout\; id="1"\; label="M"\; filenames="435696DEST\_PATH\_H200710047406520080202E000012.png,496846DEST\_PATH\_H200710047406520080202E000011.png"\; state="\{\{state\}\}">M</figure-callout>}}_1}\underset{k=0}{\overset{M_1-1}{\mathrm{\&Sigma;}}}r(t+k{T}_s)W_s\left(t\right),t\&Element;[0,T_s)---\left(8\right)$

$=x\left(t\right)+\frac{1}{{\mathrm{<figure-callout\; id="1"\; label="M"\; filenames="435696DEST\_PATH\_H200710047406520080202E000012.png,496846DEST\_PATH\_H200710047406520080202E000011.png"\; state="\{\{state\}\}">M</figure-callout>}}_1}\underset{k=0}{\overset{M_1-1}{\mathrm{\&Sigma;}}}w\left(t\right)W_s\left(t\right)$

Wherein, $x\left(t\right)=\left\{\begin{array}{cc}{g}_s(t-{\mathrm{\&tau;}}_0)& t\&Element;[{\mathrm{\&tau;}}_0,T_s)\\ g_s(t+T_s-{\mathrm{\&tau;}}_0)& t\&Element;[0,{\mathrm{\&tau;}}_0)\end{array}\right.$

Be called as noise template (Noise Template), reason is

Each the bar branch, the time-delay τ that have comprised multipath ₀And noise component(s).

After average, the variance of noise component(s) is reduced to σ ²/ M ₁, along with M ₁Increase, NT then levels off to desirable template.

Yet the waveform (like Fig. 1 (a)) when the observation signal to noise ratio is relatively low is even find M ₁=100, all can't suppress noise effectively; Attempt making average (like Fig. 1 (b)) and just partly suppressed noise with 800 symbols.And in the middle of the practical communication, it is quite big to accomplish the once time synchronization costs with 800 symbols, largely limit baud rate.Consider that average calculating operation is to be based upon on the periodicity that receives signal, thus the cycle that shortens average calculating operation will increase the number of times of average calculating operation greatly, make template receive noise effect littler.For this reason, the present invention proposes frame level noise suppressed (Frame-level Noise Suppression, FNS) notion.Frame level noise suppressed is that the cycle of average calculating operation is reduced to the frame level, and this moment, training sequence bit can not be introduced the TH sign indicating number.Analyze the three big functions of TH sign indicating number in UWB communication: 1) insert for information bit provides multiple access (Multi-access, MA); 2) the albefaction frequency spectrum is to reduce the interference to other communication modes; 3) support data low intercepting and capturing rate (Low Probability ofInterception, LPI).Because training sequence only accounts for the very little part of whole bit stream, therefore this scheme will can not exert an influence to above-mentioned 3.

Constructing NT through frame level noise suppressed mode is shown below:

${\tilde{p}}_{\mathrm{rf}}\left(t\right)=\frac{1}{M_1{N}_f}\underset{k=0}{\overset{M_1{N}_f-1}{\mathrm{\&Sigma;}}}r(t+k{T}_f)W_f\left(t\right),t\&Element;[0,T_f)---\left(9\right)$

Wherein

Subsequently will

With T _fFor the cycle carry out continuation as shown in the formula, being extended to length is T _sNT, can carry out the collection of energy.The NT of neotectonics as shown in the formula

${\tilde{P}}_{\mathrm{NT}}\left(t\right)={\tilde{p}}_{\mathrm{rf}}\left(t\mathrm{mod}{T}_f\right),t\&Element;[0,T_s)---\left(10\right)$

If adopt M ₁Individual symbol averages, and each symbol comprises N _fFrame, under the then symbol level average mode, noise variance is σ ²/ M ₁And the frame level on average can make noise variance drop to σ ²/ (M ₁N _f),, can see in environment that the noise inhibiting ability of frame level average calculating operation is more a lot of by force than symbol level average calculating operation by Fig. 2 than low signal-to-noise ratio.

After the NT structure is accomplished, just can begin to utilize preamble sequence and NT to carry out parameter Estimation.Except before the M that mentions ₁Individually be used for making up outside complete "+1 " sequence of NT, for the needs of parameter Estimation, also disposed one group in addition after synchronizing sequence constitute by the subclass of two mutual exclusions: S ₊:={ s [n]=s [n-1] } and S _-:={ s [n]=-s [n-1] }, M wherein ₂And M ₃Be respectively S set ₊And S _-Gesture.

The dominance expression formula that obtains y [n] according to the derivation in [1] appendix A is:

y[n]＝A _ξ(s[n](N _f-n _f-λ _ξ)+s[n-1](n _f+λ _ξ))+w[n] (11)

Wherein $A_{\mathrm{\&xi;}}:=\sqrt{{\mathrm{\&epsiv;}}_s}{\&Integral;}_0^{T_f}{g}_s(t+T_f-\mathrm{\&xi;})p_{\mathrm{Rf}}\left(t\right)\mathrm{Dt}$ The energy of representing a frame.(11) show that the energy that y [n] collects is the energy of former and later two continuous symbols owing to the existence of time-delay, and the CML algorithm utilizes just front and back two parts energy to estimate time-delay.

[5] the method derivation through CML obtains parameter A _ξThe analytical expression of estimating is:

${\hat{A}}_{\mathrm{\&xi;}}=\frac{1}{N_f{M}_2}\underset{n\&Element;S_{+}}{\mathrm{\&Sigma;}}y\left[n\right]s\left[n\right]$

The observation following formula is found, for A _ξEstimation, be to utilize S set in essence ₊Interior complete "+1 " sequencal estimation obtains the energy of a frame.Estimate A _ξAccurately whether directly have influence on n again _fEstimated accuracy.The present invention proposes noise template cross-correlation and estimates that (NT Cross-correlation Estimation, mode NTCE) is estimated A thus _ξ, further suppress The noise.

Noise template cross-correlation method of estimation utilizes preamble sequence to produce the template NT that two noises do not have correlation shown in Fig. 3 (b) ₁And NT ₂, and adopt the cross-correlation of NT to estimate A _ξ, promptly

${\hat{A}}_{\mathrm{\&xi;}}^{*}=\frac{1}{N_f}\underset{0}{\overset{T_s}{\&Integral;}}{\tilde{p}}_{{\mathrm{NT}}_1}\left(t\right)\widetilde{p_{N{T}_2}}\left(t\right)\mathrm{dt}---\left(13\right)$

Can find out, change after the design of estimator that two components of associative operation all pass through the processing of frame level noise suppressed, and the A of former CML algorithm _ξEstimate, see formula (12), wherein y [n] only handles through a spot of symbol level noise suppressed, so A _ξEstimated performance obviously estimate not as noise template cross-correlation.

A _ξBe updated to n after estimating to accomplish _fCan obtain thick synchronous time-delay in the expression formula of estimating and estimate, promptly

Former CML algorithm is after signal to noise ratio is higher than certain level, and there is the performance lower limit in mean square error, and tracing it to its cause is because the final time-delay that obtains is estimated ${\widehat{\mathrm{\&tau;}}}_0=T_f{\hat{n}}_f$ Be T _fIntegral multiple, thereby ξ becomes the systematic error of estimator inevitably.In order further to improve the estimated accuracy under the high s/n ratio environment, (SlideCorrelation Search, smart synchronized algorithm SCS) are estimated ξ through catching the corresponding τ value of relevant peaks, promptly based on the slip relevant search to increase one-level

$\widehat{\mathrm{\&xi;}}=\arg \underset{\mathrm{\&tau;}\&Element;[-T_f,T_f)}{\max }\underset{0}{\overset{T_s}{\&Integral;}}r(t+({\mathrm{<figure-callout\; id="1"\; label="M"\; filenames="435696DEST\_PATH\_H200710047406520080202E000012.png,496846DEST\_PATH\_H200710047406520080202E000011.png"\; state="\{\{state\}\}">M</figure-callout>}}_1+M_2+M_3)T_s+{\widetilde{\mathrm{\&tau;}}}_0+\mathrm{\&tau;}){\tilde{p}}_{\mathrm{NT}}(t+\widetilde{{\mathrm{\&tau;}}_0}+\mathrm{\&tau;})\mathrm{dt}---\left(15\right)$

Wherein τ exists

Front and back 2T _fWhat kind of stepping step-searching in the scope adopts search for this zone and can adjust according to the requirement of precision; And for NT, adopt step cycle to be shifted and cooperate the slip of correlation window on r (t), to realize the slip related operation.

The estimation of finally always being delayed time ${\widetilde{\mathrm{\&tau;}}}_{\mathrm{Total}}=T_f{\hat{n}}_f+\widehat{\mathrm{\&xi;}}.$

After accomplishing synchronously; Can confirm the original position of symbol through

; Simultaneously; is written into frame level noise template and it is carried out the inner time domain of frame proofreaies and correct, and introduce the demodulation templates that continuation when jumping can construct band TH sign indicating number.(Inter-frame Interference, IFI), all need the add up influence of former frame or even front cross frame (is looked channel length and is decided, here hypothesis for the structure of each frame template to consider the inter-frame-interference that the TH sign indicating number introduced ${\mathrm{\&tau;}}_L+T_c{\mathrm{Max}}_{k\&Element;[1,N_f)}\left(c_k\right)\&le;2T_f).$ So the method for structure demodulation NT is as follows.

At first introduce

carried out cyclic shift, then

${\tilde{p}}_{\mathrm{rf}}(t;\widehat{\mathrm{\&xi;}})=\left\{\begin{array}{cc}{\tilde{p}}_{\mathrm{rf}}(t-\widehat{\mathrm{\&xi;}})& t\&Element;[\widehat{\mathrm{\&xi;}},T_f)\\ {\tilde{p}}_{\mathrm{rf}}(t+T_f-\widehat{\mathrm{\&xi;}})& t\&Element;[0,\widehat{\mathrm{\&xi;}})\end{array}\right.---\left(16\right)$

Again

handled when jumping; Since need to consider current with the influence of former frame, so

${\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{\left(k\right)}(t;\widehat{\mathrm{\&xi;}})=\left\{\begin{array}{cc}{\tilde{p}}_{\mathrm{rf}}(t-c_k{T}_c;\widehat{\mathrm{\&xi;}})& t\&Element;[c_k{T}_c,c_k{T}_c+T_f)\\ 0& t\&Element;[0,c_k{T}_c)\&cup;t\&Element;[c_k{T}_c+T_f,2T_f)\end{array}k\&Element;[1,N_f)\right.---\left(17\right)$

Carry out time domain at last and launch, obtain

$p_{\mathrm{NT}-\mathrm{De}\mathrm{mod}}\left(t\right)\text{=}\left\{\begin{array}{cc}{\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{\left(1\right)}(t;\widehat{\mathrm{\&xi;}})& t\&Element;[0,T_f)\\ {\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{\left(2\right)}(t-T_f;\widehat{\mathrm{\&xi;}})+{\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{\left(1\right)}(t;\widehat{\mathrm{\&xi;}})& t\&Element;[T_f,2T_f)\\ .& \\ .& \\ .& \\ {\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{\left(k\right)}(t-(k-1)T_f;\widehat{\mathrm{\&xi;}})+{\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{(k-1)}(t-(k-2)T_f;\widehat{\mathrm{\&xi;}})& t\&Element;[(k-1)T_f,k{T}_f)\\ .& \\ .& \\ .& \\ {\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{\left(N_f\right)}(t-(N_f-1)T_f;\widehat{\mathrm{\&xi;}})+{\tilde{p}}_{\mathrm{rf}-\mathrm{TH}}^{(N_f-1)}(t-(N_f-2)T_f;\widehat{\mathrm{\&xi;}})& t\&Element;[(N_f-1)T_f,T_s)\end{array}\right.---\left(18\right)$

Carry out MATLAB emulation to the BPSK-UWB system.Adopt CM in the modified model S-V model ₁Channel (LOS 0-4m), channel parameter are (1/ Λ, 1/ λ, Γ, γ)=(43,0.4,7.1,4.3) ns.Used impulse waveform is Gauss's second order pulse of unit energy, and pulse duration is 1ns.Basic parameter is following: N _f=20, T _f=50ns.Configuration training sequence M ₁=M ₂=M ₃=20.When adding the SCS algorithm, make stepping length Δ=T _f/ 20.Three kinds of improvement projects of this paper be loaded on the CML algorithm one by one be optimized, adopt the normalization mean square error (Normalized MSE, $\mathrm{NMSE}=E\left\{{|({\widehat{\mathrm{\&tau;}}}_0-{\mathrm{\&tau;}}_0)/T_s|}^2\right\})$ Analyze its net synchronization capability, simulation result is as shown in Figure 4.

Simulation result shows; FNS and NTCE have improved the synchronization accuracy under the low signal-to-noise ratio situation really to a certain extent; Yet under high s/n ratio the target of algorithm be thick synchronously, precision is about a frame scope, so FNS does not have ability with NTCE and breaks through the lower limit of existence in theory; When appending the smart synchronization and estimation of one-level SCS, all square performance lower limit is broken through about an one magnitude.Yet, can find that from Fig. 4 SCS can't optimize all square performance under the low signal-to-noise ratio situation, reason is total time-delay estimation ${\widehat{\mathrm{\&tau;}}}_0=T_f{\hat{n}}_f+\widehat{\mathrm{\&xi;}},$ Wherein

Accounted for very big proportion, in case

Bigger deviation appears,

Whether accurate estimation contribution to total time-delay little, be that the frame level has obtained correct time-delay estimation so launch the prerequisite of SCS.

The above is a preferable embodiment of the present invention; But protection scope of the present invention is not limited thereto; Any those of ordinary skill in the art are in the technical scope of the present invention's explanation, and the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection range of claims.

List of references

[1]Zhi?Tian?and?G?B.Giannakis，“A?GLRT?Approach?to?Data-Aided?Timing?Acquisitionin?UWB?Radios-Part?I：Algorithms”，IEEE?Transactions?on?wireless?communications，Vol.4，No.6，pp.2956-2967，Nov.2005

Claims (3)

1. ultra-wideband communication system synchronization method of estimating based on condition maximum likelihood, it is characterized in that: at first will construct one group of training sequence, this training sequence is made up of preamble sequence and back synchronizing sequence two parts, wherein preamble sequence M ₁+ M ₂Be used for the generation of noise template NT, back synchronizing sequence, M ₃Be used for estimated frame level timing offset amount n _f

The concrete steps of synchronizing process are following:

(1) utilizes preceding M in the preamble sequence ₁Individual symbol sebolic addressing structure noise template NT ₁, suppress the influence that noise produces, noise template NT through the wave-average filtering computing of frame level ₁Structural formula is following:

Wherein

N _fBe the frame number in the single symbol, T _fBe frame period, r (t) is for being written into the reception waveform of timing offset, and t is a time variable;

Then will

With T _fFor the cycle is carried out continuation, being extended to length is T _sNT ₁, the NT of neotectonics ₁As shown in the formula:

T wherein _s=N _f* T _f, represent single mark space;

(2) utilize back M in the preamble sequence again ₂Individual symbol sebolic addressing structure noise template NT ₂, suppress the influence that noise produces, template NT through the wave-average filtering computing of frame level ₂Structural formula is following:

Wherein

Then will

With T _fFor the cycle is carried out continuation, being extended to length is T _sNT ₂, the NT of structure ₂As shown in the formula:

(3) NT to generating ₁And NT ₂Thereby, average and obtain noise template NT, be shown below:

(4) NT to generating ₁And NT ₂Carry out noise template cross-correlation and estimate, obtain to characterize the parameter Estimation of a frame energy Be shown below:

(5) according to the CML algorithm, the estimated value

of utilizing noise template NT, back synchronizing sequence M3 and other known parameters of being constructed in the estimation gained

(3) in (4) to calculate frame level timing offset amount is shown below:

Wherein { ± 1} is the transmission information symbol to

s [n] ∈; S-is the subclass of transmission information symbol, and satisfy the condition S-:={s [n] of the opposite in sign of neighbor information symbol=-s [n-1];

(6) on the basis of the estimated value

that obtains frame level timing offset amount; Define

and increase regional slip relevant search; Estimate ξ through catching the corresponding ξ value of relevant peaks, that is:

Wherein ξ exists

Front and back span be 2T _fStep-searching in the zone, step size Δ τ, Δ τ ∈ (0, T _f] adjust according to the requirement of precision; And for noise template NT, adopt step cycle to be shifted and cooperate the slip of correlation window on r (t), to realize the slip related operation;

(7) estimation

of finally always being delayed time

2. the ultra-wideband communication system synchronization method of estimating based on condition maximum likelihood as claimed in claim 1 is characterized in that in the described step (1), and preamble sequence is made up of one group of complete "+1 " sequence, and modulation system is BPSK.

3. the ultra-wideband communication system synchronization method of estimating based on condition maximum likelihood as claimed in claim 1 is characterized in that in the described step (6), and back synchronizing sequence replaces the sequence of arranging by one group+1 and-1 and constitutes: "+1 ,-1 ,+1 ,-1 ... "

Images