CN101977169B - Time domain parameter blind evaluation method of OFDM (Orthogonal Frequency Division Multiplexing) signals - Google Patents

Abstract

The invention discloses a time domain parameter blind evaluation method of OFDM (Orthogonal Frequency Division Multiplexing) signals, which mainly solves the problems of poor evaluation performance and high calculation load of the traditional evaluation method under the conditions of low signal to noise ratio and multipath channels. The evaluation method comprise the steps of: designing a power spectrum of an approaching signal of an optimum cosine roll-off filter, and evaluating an oversampling rate; a calculating correlation coefficient function sequence of an OFDM baseband sampling signal, and evaluating an effective data length; calculating a cycle autocorrelation function of a signal moving autocorrelation function sequence, searching a cycle frequency corresponding to a peak value position of the cycle autocorrelation function, and evaluating a total symbol length; and evaluating a cycle prefix length by utilizing the evaluated total symbol length and the evaluated effective data length. The invention can ensure that the evaluation error of the oversampling rate and the total symbol length is small, the accuracy rate of the evaluation of the effective data length is high, and the performance of the evaluation of the total symbol length is free of the influence of frequency shift and phase shift, and the method can be used for the time domain parameter blind evaluation of the OFDM baseband sampling signals in the communication technical field.

Description

The time domain parameter blind estimating method of ofdm signal

Technical field

The invention belongs to communication technical field, be specifically related under a kind of multipath channel, the Low SNR time domain parameter blind estimating method of ofdm signal be can be used for the estimation of over-sampling rate, valid data length, symbol total length and the circulating prefix-length of ofdm signal.

Background technology

OFDM (OFDM) technology is used widely in various fields because of its good anti-frequency selective fading characteristic, has become the critical transmissions technology of following the 4th Generation Mobile Communication System.Meanwhile, estimation and the test problems of ofdm signal modulation parameter become increasingly conspicuous, and in non-cooperation communication system, realize timing sequence reconstruct and the demodulation of total blindness's ofdm signal, at first need estimating OFDM time domain and frequency domain parameter.Wherein time domain parameter mainly comprises the over-sampling rate of ofdm signal, circulating prefix-length, and symbol total length and valid data length, frequency domain parameter mainly contains counting of IFFT conversion, the bandwidth of ofdm signal etc.

At present, the research of OFDM time domain parameter blind estimating method has obtained a large amount of achievements, but less for the method for over-sampling rate estimation.The method that cycle frequency detects utilizes the cyclostationarity of the reception signal that over-sampling brings to estimate over-sampling rate, and utilizing the cyclostationarity estimate symbol total length of ofdm signal itself, peak-value detection method utilizes the characteristics of the variable correlation delay function of ofdm signal to estimate effective data length.Referring to M.Shi, Y.Bar Ness, and W.Su. " Blind OFDM systems parameters estimation for software definedradio; " Proc.IEEE Int.Symposium on New Frontiers in Dynamic Spectrum AccessNetworks, Dublin, Ireland, 2007, pp.119-122.This method amount of calculation is large, the poor-performing under low signal-to-noise ratio, multipath channel condition.The method that the method that frequency spectrum approaches utilizes ideal low-pass filter to approach the frequency spectrum that receives signal is estimated over-sampling rate.Referring to Vincent Le Nir, Toon van Waterschoot, Marc Moonen and Jonathan Duplicy. " Blind CP-OFDM and ZP-OFDM Parameter Estimaion in Frequency Selective Channels; " EURASIP Journal on Wireless Communications and Networking, 2009.This method is not considered the characteristics of the unloaded ripple of existence of ofdm signal, so that the over-sampling rate estimated performance is relatively poor.Utilize the cyclophysis of the time domain auto-correlation function of ofdm signal based on the time domain correlation technique of Cyclic Prefix characteristic, seek the distance estimations symbol total length between peak value.Referring to Liu Peng, Li Bingbing, Lu Zhaoyang and Gong Fengkui. " A Blind Time-parameters Estimation Scheme for OFDM in Multi-path Channle; " Wireless Communications, Networking and Mobile Computing, 2005, vol.1, pp.242-247.This method is based on the detection of distance between the peak value, and estimated performance is relatively poor under low signal-to-noise ratio, multipath channel condition.Therefore, above method is not suitable for using in the wireless channel of reality.

Summary of the invention

The objective of the invention is to overcome the deficiency of above-mentioned prior art, a kind of time domain parameter method of estimation of ofdm signal is provided, the estimated performance of estimating to improve the ofdm signal time domain parameter, and reduce computation complexity.

Realize the technical scheme of the object of the invention, comprise the steps:

(1) utilize the Welch method to estimate the power spectrum of OFDM base band oversampled signals r (i), i=1 wherein, 2, Λ, N, N is the data length of intercepting;

(2) power spectrum that obtains is carried out wavelet noise and process, obtain new power spectrum Y;

(3) according to new ideal low-pass filter of power spectrum Y design, select best rolloff-factor α _Opt=0.2, on the basis of ideal low-pass filter, design again a best cosine roll off filter, with the cut-off frequency ω of best cosine roll off filter _sAs the cut-off frequency of power spectrum signal Y, estimate the over-sampling rate of OFDM base band oversampled signals

$\hat{q}=\frac{2\mathrm{\π}}{2{\mathrm{\ω}}_s};$

(4) the correlation coefficient function sequence ρ (k) of calculating OFDM baseband sampling signal r ' (i '), the peak of detection ρ (k)

Select with

2 power number formulary d with minimum Eustachian distance _Opt, estimate the valid data length of OFDM baseband sampling signal

$\left\{\begin{array}{c}{d}_{\mathrm{opt}}=\underset{d}{\arg \min }\left(\mathrm{abs}({\hat{N}}_D-2^d)\right),d=\mathrm{1,2},\mathrm{\Λ},13\\ {\hat{N}}_{\mathrm{Tu}}=2^{d_{\mathrm{opt}}}\end{array}\right.;$

I '=1,2 wherein, Λ, N ', N ' is the data length of intercepting, d is variable power number formulary, d _OptFor with

Power number formulary with minimum Eustachian distance;

(5) will estimate valid data length

As correlation delay length, calculate the mobile auto-correlation function sequence R (m) of r ' (i '), R (m) is asked absolute value, then carry out median filter smoothness of image and remove average value processing, go to become 0 less than 0 functional value after the average, obtain pretreated mobile auto-correlation function R ' (m);

(6) calculate pretreated mobile auto-correlation function R ' Cyclic Autocorrelation Function (m), the cycle frequency α ' corresponding to peak of search Cyclic Autocorrelation Function in the hunting zone of setting _Opt, estimate the symbol total length and be

${\hat{N}}_S=\frac{1}{{\mathrm{\α}}_{\mathrm{opt}}^{\′}};$

(7) utilize the symbol total length that estimates

With valid data length

Estimate the circulating prefix-length of ofdm signal

${\hat{N}}_G={\hat{N}}_S-{\hat{N}}_{\mathrm{Tu}}.$

The present invention compared with prior art has following advantage:

1) the present invention is when the over-sampling rate of estimating OFDM base band oversampled signals, consider that there is unloaded ripple in ofdm signal, behind the best ideal low-pass filter of design, select again best cosine roll off coefficient, designed the best cosine roll off filter that approaches, with the cut-off frequency of the best cosine roll off filter cut-off frequency as power spectrum signal, improved the estimated performance of over-sampling rate;

2) the present invention is when the valid data length of estimating OFDM baseband sampling signal, after searching out the peak of correlation coefficient function, estimated performance has improved as the valid data length of estimating in 2 the power side of selecting to have minimum Eustachian distance with peak;

3) the present invention is when the symbol total length of estimating OFDM baseband sampling signal, calculate first the mobile auto-correlation function of OFDM baseband sampling signal, then utilize the cycle frequency detection method to estimate the cycle period of mobile auto-correlation function, it is the symbol total length of ofdm signal, and according to the selection characteristics of ofdm signal circulating prefix-length, define the hunting zone of cycle frequency, not only improved estimated performance, and reduced the computation complexity of method.

Simulation result shows that best cosine roll off coefficient is 0.2; In multipath channel, signal to noise ratio be-during 2dB, the mean square error of over-sampling rate method of estimation proposed by the invention is no more than 0.2; In multipath channel, signal to noise ratio be-during 5dB, the accuracy rate of valid data length method of estimation proposed by the invention can reach 97%; In multipath channel, when signal to noise ratio is 0dB, the evaluated error of symbol total length method of estimation proposed by the invention is 0.21%, and estimated performance is not subject to the impact of frequency shift (FS) and phase deviation, and estimated performance is better than existing time domain autocorrelation method based on Cyclic Prefix and existing cycle frequency detection method.

Description of drawings

It is the flow chart of the blind estimation of ofdm signal time domain parameter of the present invention among Fig. 1;

That cosine roll off coefficient and the wavelet decomposition number of plies were on the analogous diagram that affects of estimated performance during over-sampling rate of the present invention was estimated among Fig. 2;

That over-sampling rate evaluated error of the present invention is with signal to noise ratio situation of change analogous diagram among Fig. 3;

That the accuracy of valid data length estimation of the present invention is with signal to noise ratio situation of change analogous diagram among Fig. 4;

That symbol total length evaluated error of the present invention is with signal to noise ratio situation of change analogous diagram among Fig. 5;

Fig. 6 is that the evaluated error estimated of symbol total length of the present invention is with phase shift change situation analogous diagram;

Fig. 7 is that the evaluated error estimated of symbol total length of the present invention is with frequency shift (FS) situation of change analogous diagram.

Embodiment

The ofdm signal that uses among the present invention is DVB-T ofdm signal 2K FFT pattern, and multipath channel is GSM TU 6 footpath channel models.

With reference to Fig. 1, specific implementation step of the present invention is as follows:

Step 1 utilizes the Welch method to estimate the power spectrum of OFDM base band oversampled signals r (i), i=1 wherein, and 2, Λ, N, N is the data length of intercepting:

1.1) the OFDM base band oversampled signals r (i) of intercepting is divided into the L section, every segment data length M _L, the OFDM base band oversampled signals data of l section can be expressed as:

r _l(k′)＝r(k′+(l-1)·M _L)，k′＝1，2，ΛM _L，l＝1，2，Λ，L 1)

Wherein

L represents the data segment variable, k ' expression data symbol variable;

1.2) window function w (k ') is added on every one piece of data, obtain the modified periodogram of each section, the modified periodogram of l section is expressed as:

$I_l\left(\mathrm{\ω}\right)=\frac{1}{U}{\left|\underset{k^{\′}=1}{\overset{M_L}{\mathrm{\Σ}}}{r}_l\left(k^{\′}\right)w\left(k^{\′}\right)e^{-\mathrm{j\ω}{k}^{\′}}\right|}^2---2)$

ω is the frequency domain variable in the formula, and j is the imaginary part vector of plural number, and U is normalization factor, and its expression formula is:

$U=\frac{1}{M_L}\underset{k^{\′}=1}{\overset{M_L}{\mathrm{\Σ}}}{w}^2\left(k^{\′}\right);---3)$

1.3) with the modified periodogram I of each section _lBe similar to (ω) and regard incoherent mutually as, the power spectrum of then estimating with the Welch method

Be expressed as:

$\hat{P}\left(e^{\mathrm{j\ω}}\right)=\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{I}_l\left(\mathrm{\ω}\right)---4)$

Among the present invention, the ofdm signal data length N=2048*9 of intercepting is divided into the L=9 section with these data, every segment data length M _L=2048, window function w (k ') elects the Hamming window as.

Step 2 is carried out wavelet noise to power spectrum and is processed, and obtains new power spectrum Y.

Because there are a lot of high frequency details in the ofdm signal power spectrum that uses the Welch method to estimate, can extract high frequency details in the signal by wavelet noise, make the profile of signal spectrum more clear, so that the best ideal low-pass filter of subsequent design.

The wavelet noise process is directly realized by the function wden () that carries among the Matlab.This function wden () comprises selectable parameter small echo type, the wavelet decomposition number of plies and threshold values processing mode etc.The small echo type of using among the present invention is 3 Daubechies small echo as wavelet-order, and the wavelet decomposition number of plies is defined as 4 by emulation, and the threshold values processing mode is that soft threshold values is processed.After utilizing the Daubechies small echo that power spectrum is carried out the wavelet noise processing, obtain new power spectrum Y.

Step 3 according to new best ideal low-pass filter of power spectrum Y design, is selected best rolloff-factor α _Opt=0.2, on the basis of best ideal low-pass filter, design again a best cosine roll off filter, with the cut-off frequency ω of best cosine roll off filter _sAs the cut-off frequency of power spectrum signal Y, estimate the over-sampling rate of OFDM base band oversampled signals

3.1) because the baseband OFDM signal is low-pass signal, therefore by formula 5) best ideal low-pass filter H of design ^Target:

$H^{t\arg \mathrm{et}}=Y_{i_{\mathrm{opt}}}^{t\arg \mathrm{et}}---5)$

$\left\{\begin{array}{c}{Y}_{i^{\′\′}}^{t\arg \mathrm{et}}={[\frac{A}{2i^{\′\′}}\mathrm{ones}\left(i^{\′\′}\right),\mathrm{zeros}(N-2i^{\′\′}),\frac{A}{2i}\mathrm{ones}\left(i^{\′\′}\right)]}^T\\ i_{\mathrm{opt}}=\underset{i^{\′\′}}{\arg \min }{(Y-Y_{i^{\′\′}}^{t\arg \mathrm{et}})}^2\end{array}\right.;$

Wherein [] ^TThe computing of expression transposition, Be the ideal low-pass filter of design, A=sum (Y) is the gross energy of power spectrum signal Y, N _FFTThe counting of Fourier transform during estimated power spectrum, ones (i ") produces i continuously " individual 1 value, zeros (N _FFT-2i ") produces (N continuously _FFTIndividual 0 value of-2i "), " value is that " hunting zone of value is 1～N for the passband dead length of ideal low-pass filter, i to i _FFT/ 2, i _OptPassband dead length for best ideal low-pass filter;

3.2) be 36MHz in sample frequency, signal to noise ratio is under the multipath channel condition of 0dB, chooses best cosine roll off factor alpha by emulation _OptBe 0.2;

3.3) best cosine roll off factor alpha that utilize to select _Opt, by formula 6) and the best cosine roll off filters H (ω ') of approaching of design:

$H\left({\mathrm{\&omega;}}^{\&prime;}\right)=\left\{\begin{array}{cc}\frac{A}{2i_{\mathrm{opt}}},& 0\&le;{\mathrm{\&omega;}}^{\&prime;}<(1-{\mathrm{\&alpha;}}_{\mathrm{opt}})i_{\mathrm{opt}}\\ \frac{A}{4i_{\mathrm{opt}}}(1+\sin \left(\mathrm{\&pi;}\frac{i_{\mathrm{opt}}-{\mathrm{\&omega;}}^{\&prime;}}{2{\mathrm{\&alpha;}}_{\mathrm{opt}}{i}_{\mathrm{opt}}}\right)),& (1-{\mathrm{\&alpha;}}_{\mathrm{opt}})i_{\mathrm{opt}}\&le;{\mathrm{\&omega;}}^{\&prime;}<(1+{\mathrm{\&alpha;}}_{\mathrm{opt}})i_{\mathrm{<figure-callout\; id="0"\; label="opt"\; filenames="HDA0000031494530000022.png,HDA0000031494530000032.png"\; state="\{\{state\}\}">opt</figure-callout>}}\\ 0,& {\mathrm{\&omega;}}^{\&prime;}\&GreaterEqual;(1+{\mathrm{\&alpha;}}_{\mathrm{opt}})i_{\mathrm{opt}}\end{array}\right.---6)$

Wherein ω ' is the frequency domain variable of cosine roll off filter;

3.4) with the cut-off frequency ω of best cosine roll off filter _sCut-off frequency as power spectrum signal Y:

${\mathrm{\&omega;}}_s=\frac{(1+{\mathrm{\&alpha;}}_{\mathrm{opt}})i_{\mathrm{opt}}}{N_{\mathrm{FFT}}}\&CenterDot;2\mathrm{\&pi;}---\left(7\right)$

By formula 8) over-sampling rate of estimating OFDM base band oversampled signals

For:

$\hat{q}=\frac{2\mathrm{\&pi;}}{2{\mathrm{\&omega;}}_s}=\frac{N_{\mathrm{FFT}}}{2(1+{\mathrm{\&alpha;}}_{\mathrm{opt}})i_{\mathrm{opt}}}.---8)$

Step 4, the correlation coefficient function sequence ρ (k) of calculating OFDM baseband sampling signal r ' (i '), the peak of detection ρ (k)

Select with 2 power number formulary d with minimum Eustachian distance _Opt, estimate the valid data length of OFDM baseband sampling signal

4.1) OFDM baseband sampling signal r ' (i ') is expressed as:

r′(i′)＝s(i′)+n(i′)，i′＝1，2，Λ，N′ 9)

Wherein s (i ') represents the useful signal of transmission, and n (i ') expression average is 0, and variance is 1 additive white Gaussian noise, i ' expression data variable, and N ' is the data length of intercepting.

The auto-correlation function of OFDM baseband sampling signal is expressed as:

E[wherein] the expression auto-correlation computation, the conjugation of r ' * () expression r ' (), k represents variable correlation delay length,

With The energy that represents respectively useful signal and additive white Gaussian noise, ε represent the frequency deviation because of the channel generation, N _DBe the real valid data length of ofdm signal;

4.2) utilize the degree of correlation between correlation coefficient function sequence ρ (k) expression OFDM baseband sampling signal r ' (i ') data, correlation coefficient function sequence ρ (k) and formula 10) in auto-correlation function have identical characteristic, calculate correlation coefficient function sequence ρ (k):

$\mathrm{\&rho;}\left(k\right)=\frac{\left|\underset{i^{\&prime;}=1}{\overset{N^{\&prime;}-k}{\mathrm{\&Sigma;}}}(r^{\&prime;}\left(i^{\&prime;}\right)\&CenterDot;r^{\&prime;*}(k+i^{\&prime;}))\right|}{\underset{i^{\&prime;}=1}{\overset{N^{\&prime;}}{\mathrm{\&Sigma;}}}(r^{\&prime;}\left(i^{\&prime;}\right)\&CenterDot;r^{\&prime;*}\left(i^{\&prime;}\right))}---11)$

Wherein N ' is the data length of intercepting, and k is variable correlation delay length, and span is that 1～8000, r ' (i ') is the individual data of i ', i '=1,2, Λ, N ';

By formula 10) can find out that when the variable delay correlation length of auto-correlation function equaled the valid data length of ofdm signal, peak value appearred in auto-correlation function, so ofdm signal auto-correlation function sequence ρ (k) also can peak value occur in the relevant position;

4.3) search correlation coefficient function sequence ρ (k) peak

${\hat{N}}_D=\underset{k}{\arg \max }\left(\mathrm{\&rho;}\left(k\right)\right);---12)$

4.4) be numerically equal to counting of ofdm signal IFFT conversion according to ofdm signal valid data length, and counting of ofdm signal IFFT conversion be 2 power side, select with Has 2 power side of minimum Eustachian distance as the valid data length of estimating

$\left\{\begin{array}{c}{d}_{\mathrm{opt}}=\arg \min \left(\right|{\hat{N}}_D-2^d\left|\right),d=\mathrm{1,2},\mathrm{\&Lambda;},13\\ {\hat{N}}_{\mathrm{Tu}}=2^{d_{\mathrm{opt}}}\end{array}\right.,---13)$

Among the present invention, consider the problem of disturbing between subcarrier, the IFFT conversion is counted can be very not large, and the IFFT conversion of DVB-TOFDM signal maximum to count be 2 ¹³, therefore the span of 2 power number formulary d is 1～13.

Step 5 is with the valid data length that estimates

As correlation delay length, calculate the mobile auto-correlation function sequence R (m) of r ' (i '), R (m) is asked absolute value, then carry out median filter smoothness of image and remove average value processing, go to become 0 less than 0 functional value after the average, obtain pretreated mobile auto-correlation function R ' (m).

5.1) set correlation delay length and be

Calculate the mobile auto-correlation function sequence R (m) of OFDM baseband sampling signal:

$R\left(m\right)=\left|\underset{j^{\&prime;}=1}{\overset{L^{\&prime;}}{\mathrm{\&Sigma;}}}(r^{\&prime;}(j^{\&prime;}+m)\&CenterDot;r^{\&prime;*}(j^{\&prime;}+m+{\hat{N}}_{\mathrm{Tu}}))\right|,$ $m=\mathrm{1,2},\mathrm{\&Lambda;},N^{\&prime;}-{\hat{N}}_{\mathrm{Tu}}-L^{\&prime;}---14)$

Wherein N ' is the data length of intercepting, and m is the position of Moving Window, and L ' is Moving Window length, and r ' (j ') is the individual data of j ', j '=1,2, Λ, L ';

By formula 14) can draw, the mobile auto-correlation function sequence R (m) of ofdm signal is the one-period sequence, the size in its cycle equals the symbol total length of ofdm signal;

5.2) mobile auto-correlation function sequence R (m) is carried out median smoothing filtering obtain intermediate sequence value R " (m);

5.3) middle sequential value R " (m) is removed average value processing, and will become 0 less than 0 functional value, obtain pretreated mobile auto-correlation function R ' (m).

Step 6 is calculated pretreated mobile auto-correlation function R ' Cyclic Autocorrelation Function (m), the cycle frequency α ' corresponding to peak of search Cyclic Autocorrelation Function in the hunting zone of setting _Opt, the estimate symbol total length

6.1) the pretreated mobile auto-correlation function R ' of calculating Cyclic Autocorrelation Function (m)

$C({\mathrm{\&alpha;}}^{\&prime;},{\hat{N}}_{\mathrm{Tu}})=\left|\frac{1}{M^{\&prime;}}\underset{m=1}{\overset{M^{\&prime;}}{\mathrm{\&Sigma;}}}{R}^{\&prime;}\left(m\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000031494530000021.png,HDA0000031494530000022.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}{\mathrm{\&alpha;}}^{\&prime;}m}\right|---15)$

Wherein M ' is (m) length of sequence of pretreated mobile auto-correlation function R ',

α ' is variable cycle frequency;

6.2) extract and to make C

Corresponding value α ' when obtaining maximum _Opt:

According to the definition of cycle frequency, the cycle frequency α ' that the peak of Cyclic Autocorrelation Function is corresponding _OptBe the inverse in mobile auto-correlation function cycle, the symbol lengths of estimating OFDM signal

${\hat{N}}_S=\frac{1}{{\mathrm{\&alpha;}}_{\mathrm{opt}}^{\&prime;}}---17)$

Necessarily greater than valid data length, and consider loss and the implementation complexity of system and the factors such as peakedness ratio of system of the information transfer efficiency that Cyclic Prefix brings according to the total length of OFDM symbol, in real system, circulating prefix-length N _GCan not surpass 1/4 of OFDM symbol valid data length, i.e. N _G≤ N _D/ 4, therefore the hunting zone with cycle frequency α ' is set as

Can reduce like this computation complexity of method of estimation.

Step 7 is utilized the symbol total length of estimating

With the valid data length of estimating Estimate the circulating prefix-length of ofdm signal

${\hat{N}}_G={\hat{N}}_S-{\hat{N}}_{\mathrm{Tu}}.$

Effect of the present invention can further specify by analogous diagram:

Simulated environment sees Table 1

Table 1: simulated environment

Emulation content and result:

The emulation signal to noise ratio is 0dB, sample frequency f _sUnder=36MHz the condition, the wavelet decomposition number of plies and cosine roll off coefficient are on the impact of over-sampling rate estimated performance, and simulation result is Fig. 2.The rolloff-factor of cosine roll off filter is very large on the impact of over-sampling rate estimated performance as can be seen from Figure 2,0.1 and the better performances that obtained in 0.2 o'clock, when its greater than 0.2 the time, performance progressively worsens.In addition, find out that easily best rolloff-factor is 0.2 when the decomposition number of plies is 3～5; Decomposing the number of plies is 6～8 o'clock, and best rolloff-factor is 0.1.Among the present invention, selecting the wavelet decomposition number of plies is 4, and the cosine roll off coefficient is 0.2.

The emulation wavelet decomposition number of plies is 4, the cosine roll off coefficient is 0.2, when sample frequency is 28MHz, for the DVB-TOFDM signal, cosine roll off filter approximating method proposed by the invention and ideal filter approach method over-sampling rate estimated performance comparison diagram, simulation result is Fig. 3.As can be seen from Figure 3, the estimated performance of ideal filter approach method is along with the increase of signal to noise ratio variation on the contrary.Its reason is to have unloaded ripple in the ofdm signal, and best low pass filter only can approach the shared spectrum width of useful subcarrier.Under Low SNR, because the impact of noise, so that the power spectrum of signal produces " extending out " phenomenon, thereby the cut-off frequency of best low pass filter is closer to real signal cut-off frequency; And along with the increase of signal to noise ratio, the power spectrum of signal is more and more close to desirable power spectrum, and the shared part of unloaded ripple is left in the basket.Therefore under the high s/n ratio condition, the performance of estimating based on the over-sampling rate of best low pass filter also just worse and worse.And adopted best cosine roll off filter further to approach among the present invention, and overcome this problem that exists, be-2dB under the multipath channel condition, to estimate that mean square error is no more than 0.2 in signal to noise ratio.

The estimation accuracy of emulation valid data length is with the situation of change of signal to noise ratio, and simulation result is Fig. 4.As can be seen from Figure 4, the present invention is at SNR=-5dB, and the accuracy rate that valid data length is estimated under the multipath channel condition can reach 97%.

The evaluated error of emulation symbol total length is with the situation of change of signal to noise ratio, and simulation result is Fig. 5.As can be seen from Figure 5, the symbol total length mean deviation that the method that the present invention proposes is estimated under signal to noise ratio 0dB, multipath channel condition is approximately 0.21%, and it is better than traditional based on the time domain autocorrelation method of Cyclic Prefix and traditional circulation autocorrelation method.

The evaluated error of emulation symbol total length is with the situation of change of phase deviation, and simulation result is Fig. 6.Therefore the evaluated error of symbol total length can illustrate that method of the present invention is not subjected to the impact of phase deviation along with the variation of phase deviation is fluctuateed in a very little scope as can be seen from Figure 6.

The evaluated error of emulation symbol total length is with the situation of change of frequency shift (FS), and simulation result is Fig. 7.Therefore the evaluated error of symbol total length can illustrate that method of the present invention is not subjected to the impact of frequency shift (FS) along with the variation of frequency shift (FS) is fluctuateed in a very little scope as can be seen from Figure 7.

Claims (5)

1. an ofdm signal time domain parameter blind estimating method comprises the steps:

(1) utilize the Welch method to estimate the power spectrum of OFDM base band oversampled signals r (i), i=1 wherein, 2 ..., N, N is the data length of intercepting;

(2) power spectrum that obtains is carried out wavelet noise and process, obtain new power spectrum Y;

(3) according to new best ideal low-pass filter of power spectrum Y design, select best rolloff-factor α _Opt=0.2, on the basis of ideal low-pass filter, design again a best cosine roll off filter, with the cut-off frequency ω of best cosine roll off filter _sAs the cut-off frequency of power spectrum signal Y, estimate the over-sampling rate of OFDM base band oversampled signals

(4) the correlation coefficient function sequence ρ (k) of calculating OFDM baseband sampling signal r ' (i '), the peak of detection ρ (k)

Select with

2 power number formulary d with minimum Eustachian distance _Opt, estimate the valid data length of OFDM baseband sampling signal

I '=1,2 wherein ..., N ', N ' is the data length of intercepting, d is variable power number formulary, d _OptFor with

Power number formulary with minimum Eustachian distance;

(5) with the valid data length that estimates As correlation delay length, calculate the mobile auto-correlation function sequence R (m) of r ' (i '), R (m) is asked absolute value, then carry out median filter smoothness of image and remove average value processing, go to become 0 less than 0 functional value after the average, obtain pretreated mobile auto-correlation function R ' (m);

(6) calculate pretreated mobile auto-correlation function R ' Cyclic Autocorrelation Function (m), the cycle frequency α ' corresponding to peak of search Cyclic Autocorrelation Function in the hunting zone of setting _Opt, estimate the symbol total length and be

(7) utilize the symbol total length that estimates

With valid data length Estimate the circulating prefix-length of ofdm signal

2. according to the ofdm signal time domain parameter blind estimating method described in claims 1, the described over-sampling rate that estimates OFDM base band oversampled signals of step (3) wherein Carry out as follows:

2.1) by best ideal low-pass filter H of following formula design ^Target:

Wherein [] ^TThe computing of expression transposition,

Be the ideal low-pass filter of design, A=sum (Y) is the gross energy of power spectrum signal Y, N _FFTThe counting of Fourier transform during estimated power spectrum, ones (i ") produces i continuously " individual 1 value, zeros (N _FFT-2i ") produces (N continuously _FFTIndividual 0 value of-2i "), " value is that " hunting zone of value is 1 ~ N for the passband dead length of ideal low-pass filter, i to i _FFT/ 2, i _OptPassband dead length for best ideal low-pass filter;

2.2) be 36MHz in sample frequency, signal to noise ratio is under the multipath channel condition of 0dB, chooses best cosine roll off factor alpha by simulation result _OptBe 0.2;

2.3) utilize the best cosine roll off factor alpha choose _Opt=0.2, roll filters H (ω ') by the best cosine that approaches of following formula design:

Wherein ω ' is the frequency domain variable of cosine roll off filter;

2.4) cosine is rolled cut-off frequency ω with filter _sAs the cut-off frequency of power spectrum signal Y, by the over-sampling rate of following formula estimating OFDM base band oversampled signals For:

3. according to the ofdm signal time domain parameter blind estimating method described in claims 1, the valid data length of the described estimating OFDM baseband sampling of step (4) signal wherein

Carry out as follows:

3.1) be calculated as follows the correlation coefficient function sequence ρ (k) of OFDM baseband sampling signal r ' (i '):

Wherein N ' is the data length of intercepting, and k is variable correlation delay length, and span is that 1 ~ 8000, r ' (i ') is the individual data of i ', i '=1,2 ..., N ', r ' ^*The conjugation of () expression r ' ();

3.2) search correlation coefficient function sequence ρ (k) peak

3.3) select and peak

2 power number formulary d with minimum Eustachian distance _Opt, by the valid data length of following formula estimating OFDM baseband sampling signal

4. according to the ofdm signal time domain parameter blind estimating method described in claims 1, wherein the pretreated mobile auto-correlation function R ' of the described calculating of step (5) (m) carries out as follows:

4.1) set correlation delay length and be

Be calculated as follows the mobile auto-correlation function sequence R (m) of signal:

Wherein N ' is the data length of intercepting, and m is the position of Moving Window, and L ' is Moving Window length, and r ' (j ') is the individual data of j ', j '=1,2 ..., L ';

4.2) mobile auto-correlation function sequence R (m) is carried out median smoothing filtering obtain intermediate sequence value R " (m);

4.3) middle sequential value R " (m) is removed average value processing, and will become 0 less than 0 functional value, obtain pretreated mobile auto-correlation function R ' (m).

5. according to the ofdm signal time domain parameter blind estimating method described in claims 1, the symbol total length of the described estimating OFDM baseband sampling of step (6) signal wherein

Carry out as follows:

5.1) be calculated as follows pretreated mobile auto-correlation function R ' Cyclic Autocorrelation Function (m)

Wherein α ' is variable cycle frequency, and the hunting zone is M ' is pretreated mobile auto-correlation function R ' data length (m),

5.2) the search Cyclic Autocorrelation Function Cycle frequency α ' corresponding to peak _Opt:

5.3) by the symbol total length of following formula estimating OFDM baseband sampling signal

Images