CN101527698A - Non-stationary interference suppression method based on Hilbert-Huang transformation and adaptive notch - Google Patents

Abstract

The invention provides a non-stationary interference suppression method based on Hilbert-huang transformation and adaptive notch and relates to the interference suppression technical application field of a direct-sequence spread-spectrum communication system. Firstly, sampling results of accepted data are normalized; the Hilbert-Huang transformation is carried out on the normalized data to obtain a Hilbert spectrum; a threshold level for inhibiting noises and a direct-sequence spread spectrum is determined according to the Hilbert spectrum to obtain the non-stationary interference Hilbert spectrum; non-stationary interference instantaneous frequency is worked out according to the non-stationary interference Hilbert spectrum; parameters of an adaptive IIR lattice notch filter are determined by the non-stationary interference instantaneous frequency; the convolution operation between the sampling results and the impulse responses of the adaptive notch filter is carried out to obtain data after the inhibition of the non-stationary interference. The method fully utilizes the good adaptability and the energy accumulation of the Hilbert-Huang transformation and inhibits the non-stationary interference more effectively.

Description

Non-stationary disturbance restraining method based on Hilbert Huang and adaptive notch

Technical field

The present invention relates to direct-sequence communications system Anti-Jamming Technique application, specifically, be related to the antijamming capability of further raising direct-sequence communications system and a kind of non-stationary disturbance restraining method of taking based on Hilbert Huang (HHT) and adaptive notch.

Background technology

Direct-sequence communications system is with the pseudo random sequence with good correlation data message to be carried out spread spectrum, carries out despreading at receiving terminal to received signal with the local pseudo random sequence of precise synchronization, realizes the communication system of communication task thus.Because of it has that the signal power spectrum density is low, good confidentiality, antijamming capability are strong, the anti-multipath decline, be easy to advantage such as networking, become a kind of communication core technology that can solve many technical barriers such as communication spectrum utilance, the stupid sustainability of communication system and communication task confidentiality, become the main developing direction and the system of contemporary Communication Jamming inhibition technology and communication countermeasures technology, at military communication, test the speed, military field such as navigator fix and commercial communication field all obtained using widely.

Though spread spectrum system utilizes spreading gain that the interference signal energy is expanded on the frequency band more much bigger than data bandwidth, and most of interference signal energy can be suppressed by data filter, has stronger antijamming capability, but along with direct-sequence communications system electromagnetic environment of living in becomes increasingly complex, except steady narrow band interference, the non-stationary broad-band interference also appears.All be steadily to disturb at the arrowband to carry out the research of inhibition method, because of the difference of interference characteristic, these methods are disturbed non-stationary and are suppressed to be difficult to prove effective in the past.At present, non-stationary disturbance restraining method commonly used, will disturb with useful signal to make a distinction by analyzing the correlated characteristic that non-stationary disturbs for adopting Time-Frequency Analysis Method (distributing as Wigner-Ville), disturb thereby suppress non-stationary.But there is cross term as the Wigner-Ville distribution, influenced accurate detection and effectively inhibition that non-stationary disturbs, and in the two-dimensional space processing, the algorithm more complicated is time-consuming.

Hilbert Huang (HHT) is by adaptive decomposition of signal itself and Hilbert transform, has the excellent energy aggregation, be particularly suitable for the analysis and the processing of non-stationary nonlinear properties, can provide non-stationary signal more effective more accurate estimation, disturb thereby can more effectively suppress non-stationary.

For trapper, when the trap frequency of the frequency of signal and trapper is consistent, just can effectively suppress this signal.So when the trap frequency of notch filter just can have inhibition to disturb synchronously with disturbing instantaneous frequency.The frequency change of following the non-stationary interference when the trap frequency variation of trapper just can realize effective inhibition that non-stationary disturbs, and disturbs so can adopt adaptive notch filter to suppress non-stationary.

To combine with the adaptive notch disturbance restraining method based on the non-stationary Interference Detection of HHT, just can suppress non-stationary effectively and disturb.

Summary of the invention

The objective of the invention is for improving the antijamming capability of direct-sequence communications system, provides a kind of non-stationary disturbance restraining method based on Hilbert Huang and adaptive notch in order to overcome the deficiency of prior art.

The objective of the invention is to be achieved through the following technical solutions.

A kind of non-stationary disturbance restraining method based on Hilbert Huang and adaptive notch, concrete steps are as follows:

Step 1, for the ease of the processing of data, the sampled result that base band is received data is carried out normalization.Performing step is:

At first, ask for the absolute value maximum x of sampled result r (n) _Max

Afterwards, with the maximum x of sampled result r (n) divided by absolute value among the r (n) _Max:

x(n)＝r(n)/x _max

X (n) is the data after sampled result r (n) normalization.

Step 2, to the data after the normalization, decompose and Hilbert transform according to the empirical mode decomposition in the Hilbert Huang (EMD) algorithm, obtain hilbert spectrum.Performing step is:

A, according to the empirical mode decomposition of Hilbert Huang, will be decomposed in M through the data x after the normalization that step 1 obtains (n) and accumulate mode function (IMF) c _i(n) component sum, promptly

$\stackrel{\‾}{x}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{c}_i\left(n\right)+r_M\left(n\right)$

Wherein, r _M(n) for decomposing the back residual components, M for decompose obtain in accumulate mode function number, the value of M is according to r _M(n) size requirements is determined.

B, carry out Hilbert transform, obtain corresponding analytic signal, a resulting M analytic signal is brought into accumulateing the mode function component in each $\stackrel{\‾}{x}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{c}_i\left(n\right)+r_M\left(n\right)$ In, x (n) is expressed as

$\stackrel{\&OverBar;}{x}\left(n\right)\&ap;\mathrm{Re}\left(\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{a}_i\left(n\right)\exp \left(\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20091008141900202.png,A20091008141900211.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}\underset{b=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i\left(n\right)T_s\right)\right)$

$a_i\left(n\right)=\sqrt{c_i^2\left(n\right)+H^2\left[c_i\left(n\right)\right]}$

${\mathrm{\&theta;}}_i\left(n\right)=\arctan \frac{H\left[c_i\left(n\right)\right]}{c_i\left(n\right)}=2\mathrm{\&pi;}\underset{b=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i\left(n\right)T_s$

Wherein, H[c _i(n)] be c _i(n) boolean's Bert conversion, a _i(n) be c _i(n) instantaneous amplitude of analytic signal, f _i(n) be c _i(n) instantaneous frequency, θ _i(n) be c _i(n) instantaneous phase.T _sBe the sampling interval, the sequence number that i accumulates the mode function component in being, i.e. i=1～M; B is the sampled point sequence number, b=1～n; N is a n sampled point.

C, x (n) is expressed as time n, instantaneous frequency ω, amplitude a _i(n) three-dimensional spectrum, promptly hilbert spectrum H (ω, n):

$H(\mathrm{\&omega;},n)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{b}_i{a}_i\left(n\right)$

Wherein, b _iFor the switch factor, as ω=ω _iThe time, b _i=1, otherwise b _i=0.The number of accumulateing the mode function component in the M representative, the sequence number that i accumulates the mode function component in being, i.e. i=1～M.

The hilbert spectrum that step 3, basis obtain determines to suppress the threshold level of noise and direct sequence signal spectrum, and the spectrum that will be higher than threshold level keeps, and will be lower than the spectrum zero clearing of threshold level, obtains the hilbert spectrum of non-stationary interference.Performing step is:

(1) suppresses (CME) algorithm by consecutive mean, determine to suppress the threshold level of noise and direct sequence signal spectrum, promptly

1. with all Hilbert spectral lines as initial spectral line, and they are put into set I, obtain all Hilbert spectral line values add up and, i.e. A _mWith A _mDivided by Hilbert spectral line sum, obtain the Hilbert spectral line average E[|H (ω, n) |]:

$A_m=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|H(\mathrm{\&omega;},n)\right|$

E[|H(ω，n)|]＝A _m/(N×M)

Wherein, the number that M accumulates the mode function component in being, N is a sampling number.

2. according to the average E[|H of Hilbert spectral line (ω, n) |], adopt following formula estimate the standard deviation sigma of Hilbert spectral line [| H (ω, n) |]:

$\mathrm{\&sigma;}\left[\right|H(\mathrm{\&omega;},n)\left|\right]=\sqrt{\frac{1}{N\&times;M-1}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{|H(\mathrm{\&omega;},j)-E[\left|H(\mathrm{\&omega;},n)\right|\left]\right|}^2}$

According to the standard deviation sigma that obtains [| H (ω, n) |], determine thresholding η

η＝E[|H(ω，n)|]+3σ[|H(ω，n)|]

The mould and the thresholding η that 3. will gather the Hilbert spectral line among the I compare, and are considered to disturbed Hilbert spectral line greater than the Hilbert spectral line of this thresholding, remove from set I.Set I is updated.

4. after upgrading set I, upgrade A _mAnd E[|H (ω, n) |], repeat afterwards 2. and 3., till the mould that does not have the Hilbert spectral line is greater than thresholding η.Finally obtain not having the set I ' of the mould of Hilbert spectral line greater than thresholding η.

5. with the Hilbert spectral line of set among the I ', estimate final thresholding η=E[|H (ω, n) |]+3 σ [| H (ω, n) |].

(2) hilbert spectrum that will be higher than final thresholding η level keeps, and will be lower than the hilbert spectrum zero clearing of thresholding η level, thereby obtain the hilbert spectrum that non-stationary disturbs.

Step 4, according to the hilbert spectrum that non-stationary disturbs, seek out the instantaneous frequency that non-stationary disturbs, the parameter by this instantaneous frequency decision self adaptation infinite impulse response (IIR) lattice type trapper obtains the adaptive notch filter impulse response thus.

Step 5, sampled result in the step 1 and adaptive notch filter impulse response are carried out convolution algorithm, the non-stationary data after interference that is inhibited is carried out related operation with these data and synchronous PN sign indicating number sequence, finishes the despreading that receives data.

Beneficial effect

The inventive method utilizes the Hilbert Huang to estimate the instantaneous frequency that non-stationary disturbs, and than more accurate with adopting the Wigner-Ville distribution to estimate, efficient is higher.For correlator output Signal to Interference plus Noise Ratio improvement factor, employing is based on the lattice type iir filter of the inventive method, can obtain to many improvement of 7dB, have the performance that better inhibition non-stationary disturbs than five coefficient FIR filters that adopt the Wigner-Ville location mode.

Description of drawings

Fig. 1 is the flow chart that the present invention is based on the non-stationary disturbance restraining method of Hilbert Huang and adaptive notch;

Fig. 2 is the frequency spectrum of sampled result r (n) used in the embodiment of the invention;

Fig. 3 estimates in the embodiment of the invention that for adopting non-stationary disturbs instantaneous frequency and the comparison schematic diagram of estimating with the Wigner-Ville location mode;

Fig. 4 is the structure chart of self adaptation infinite impulse response (IIR) the lattice type trapper that adopted in the embodiment of the invention;

Fig. 5 is based on the non-stationary disturbance restraining method block diagram of Hilbert Huang and self adaptation infinite impulse response (IIR) lattice type trapper in the embodiment of the invention;

Fig. 6 is that sampled result r (n) suppresses non-stationary interference back gained data through algorithm process in the embodiment of the invention

Frequency spectrum.

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer,, the present invention is described in further detail below by in conjunction with specific embodiments and accompanying drawing.

The present invention disturbs process of inhibition to realize on the Desktop Computing machine platform to the non-stationary based on Hilbert Huang and self adaptation infinite impulse response (IIR) lattice type trapper.In the specific implementation, non-stationary disturbs and suppresses to be undertaken by flow process as shown in Figure 1, finally provides the inhibition data after interference.The sample frequency that base band receives data is f _s=2000Hz, number N=1500, information code length is 100bits, the spread spectrum code sequence cycle is L=15.Jamming-to-signal ratio JSR=10dB, signal to noise ratio snr=5dB, the frequency spectrum of gained data as shown in Figure 2, middle amplitude is higher to be exactly the frequency spectrum that non-stationary disturbs.Sampled result r (n) is as shown in table 1:

Table 1 sampled result r (n)

r(n)＝

[6.6009 6.4157 4.6043 2.8640 3.7252 1.5890 -2.9121 -5.7211 -8.5248 -6.2168

-6.4400 -7.3346 -3.5176 -4.6545 1.2088 1.1827 4.8304 2.7025 9.7289 6.3441

8.3165 5.7965 7.7253 3.4604 1.6281 -3.6675 -5.1425 -9.8763 -8.6077 -12.0674

-7.5318 -5.7143 -4.3168 1.8203 -1.1740 6.6820 6.5412 4.8397 8.9164 5.9087

4.6300 6.5819 3.6268 -3.4207 -2.3384 -7.2667 -8.7919 -10.6744 -9.1430 -5.2549

-4.3033 -1.7824 -0.4419 4.4467 7.1761 10.0178 7.1495 8.8643 9.3208 6.1655

0.6243 -1.7543 -3.7080 -5.1465 -10.2583 -9.1297 -7.0078 -4.8218 -4.8722 -2.0538

-1.8952 5.3987 4.3015 8.1956 7.1643 6.6400 5.5613 2.0502 -1.9698 -0.9228

-8.0027 -6.4848 -11.5578 -7.7461 -5.1579 -2.5042 1.3633 2.9861 8.1960 7.9960

5.5935 6.5649 6.0398 0.0819 -1.0573 -5.0753 -4.8577 -8.4346 -11.8858 -3.9772

-5.7614 -4.0611 4.4534 5.0245 4.7529 7.2238 9.0395 8.2649 5.1373

6.4575 -0.7380 -3.1079 -9.3065 -8.6466 -9.4891 -6.1520 -4.9375 -1.7703 2.6316

3.9739 9.2903 10.4682 10.5251 11.2707 3.3415 -0.7237 -3.7562 -3.3919 -6.4092

-10.1090 -6.7797 -5.9170 -2.8065 -0.9832 4.5098 8.7730 10.2356 12.8403

9.5770 1.6068 4.5510 -5.1991 -6.6846 -8.2162 -9.6765 -8.5111 -8.2203 0.0767

0.5425 4.1779 9.8166 10.1141 7.3952 4.2950 5.2113 -1.8675 -3.3073 -8.9271

-6.5659 -6.7637 -4.7095 -8.8926 3.5180 5.9699 11.1229 7.1151 10.0625 4.1278

5.0031 3.0212 -2.7540 -5.0330 -7.3944 -6.0212 -9.5542 -2.9895 -0.6222 5.7836

10.7380 10.4172 10.4273 6.3742 0.2914 -1.5333 -6.0433 -4.3380 -10.7215

-5.9719 -1.4251 -4.6546 1.7758 7.1056 9.1880 6.2547 5.5959 4.0286 1.5658

-5.2645 -5.4942 -7.8717 -6.3188 -4.6386 1.7282 2.0907 8.8793 7.0380 11.8542

6.3846 3.2944 -0.7553 -2.6393 -6.5275 -9.9035 -8.1462 -6.4446 1.3699 3.6649

8.1790 11.2138 8.9098 5.8219 3.4841 -1.7887 -3.3481 -8.9178 -7.9481 -9.9880

-2.5317 2.4138 1.1043 6.3454 8.0134 5.2855 5.6490 -2.9447 -3.9557 -7.4354

-10.012 8-8.5548 -7.6379 -2.3598 2.2579 9.1545 6.8847 7.0208 7.1997

2.0386 -3.7627 -8.9163 -5.7024 -8.3477 -6.2449 1.6488 4.9357 8.5782 11.7298

5.9339 4.6543 -0.8858 2.2030 -9.7983 -10.4144 -6.2934 -3.4240 2.1432 5.1969

7.1147 8.7379 7.4178 3.4328 1.4693 -2.3332 -8.2826 -5.6414 -7.1288 -0.6256

0.9756 7.2370 9.6904 6.5243 5.0929 1.0647 -3.8405 -5.8616 -12.0485

-4.7500 -2.3041 2.4294 5.3687 6.3457 8.9607 5.1692 4.2898 -1.3436 -5.8057

-7.2621 -7.4369 -8.5079 1.2516 3.6825 6.7507 10.3251 7.7195 0.7555 -5.4989

-6.4191 -9.3589 -8.0066 -2.1720 0.1930 3.5088 10.5289 5.3844 5.8331

-2.5168 -1.6378 -8.3253 -7.2901 -4.4708 -2.6581 6.9683 4.10631 1.5825 8.5334

4.0315 -2.7905 -6.9092 -7.9034 -7.5785 -7.6748 -1.3949 4.3577 6.1734 8.4131

6.9708 -0.3514 -2.0063 -9.5227 -9.6831 -5.6262 0.3182 2.3239 7.6233 8.0760

6.9545 4.6972 -3.3728 -6.9300 -9.2574 -7.4896 -2.6600 2.0981 9.0604 8.7850

6.9629 1.2520 -3.9129 -8.0803 -7.0192 -8.5187 -1.8546 4.9001 6.5947 8.5020

6.6045 2.4524 -3.5972 -7.0790 -8.9916 -5.5452 -0.5218 5.3088 4.6952 9.1877

4.7557 1.4555 -6.8688 -10.4818 -10.3330 -4.2539 -0.2833 7.3249 7.9638 6.3072

3.1718 -1.7004 -3.8602 -7.1633 -7.0914 -0.0312 6.4985 7.0974 7.2969 3.8636

1.0048 -6.6882 -6.5733 -9.3537 -10.0101 2.8863 3.9409 8.2108 9.3031 5.0825

-4.6011 -7.1993 -11.5756 -7.5542 1.0655 6.5479 7.9143 6.6765 2.9669 -4.2672

-5.2383 -8.0020 -5.8258 1.0467 3.0204 10.3949 9.9025 4.1305 -0.8964

-4.0080 -9.7394 -3.8233 1.1346 1.2823 8.0977 4.2110 1.7157 2.2144

-8.9776 -7.9130 -6.3474 0.7942 2.2380 4.5348 6.4828 1.3790 -1.4391

-5.5592 -10.6058 -5.5292 2.5193 7.8486 5.6808 8.1400 -0.2581 -5.5454 -9.3906

-10.4034 -2.7961 7.5973 7.8439 7.8506 3.4674 -0.9236 -4.9140 -10.4520 -6.9093

-0.5279 6.4205 8.7371 2.8396 1.7873 -6.8891 -9.0885 -6.2817 -4.2765 4.5519

4.85241 0.3096 3.5551 -8.3180 -9.7853 -5.9745 -4.5397 1.5914 5.2384 9.2640

5.1817 -1.6547 -7.3448 -10.0564 -2.9141 -0.8364 8.5488 8.8169 4.2857 -2.1598

-6.7699 -11.8449 -7.9049 1.0097 2.8150 4.6003 3.7094 -1.0252 -3.1296 -9.4310

-5.2544 -1.2150 3.0524 7.4932 6.0985 -1.0395 -5.9841 -6.6706 -4.4646

-0.6649 8.8108 9.2616 4.7570 -1.3845 -6.6133 -6.9596 -6.9258 1.3972 10.0900

7.1986 3.1197 -3.1708 -3.5544 -5.7965 -6.0369 0.3356 8.3734 8.3105 4.9496

-4.4925 -5.8210 -7.9077 -3.8454 5.3546 11.5328 7.1869 2.0145 -7.4961 -8.3377

-4.9389 0.7035 5.0355 8.7419 3.0576 -6.5208 -5.8380 -9.0395 -2.5450 6.9739

9.1303 6.6423 3.3602 -4.0170 -3.2392 -3.5952 2.3119 9.1836 8.0126 2.1627

-3.0334 -10.4857 -5.1733 1.7875 5.1572 5.8349 5.9452 -4.9841 -8.4454 -6.7499

-2.3486 2.6952 7.4788 4.7916 1.5616 -8.9376 -5.8773 -6.3500 1.0116 7.6836

7.3921 -0.2978 -5.2948 -9.3564 -7.0815 -0.3122 9.4130 6.8940 0.3662 -3.9699

-11.7623 -3.9804 -0.1300 6.6202 9.3914 -0.9245 -5.2505 -8.1013 -3.7462 3.2331

7.2525 10.7827 0.8945 -5.4982 -8.4081 -3.1671 2.5417 8.7430 10.0121 2.1941

-5.9130 -8.9807 -5.5631 1.8192 6.3348 7.1250 0.5075 -7.5702 -8.7659 -2.1648

3.5286 6.6901 5.9326 -2.1797 -5.9437 -5.1172 -3.4036 6.5319 10.9282 5.5363

0.8209 -7.1083 -7.8309 2.4661 8.3111 10.1216 -1.1651 -7.4043 -9.8504 -4.2062

0.9756 12.4741 5.2639 -3.8850 -9.4607 -5.7729 0.6656 4.7358 10.5522 1.5883

-6.7604 -8.7692 -3.4466 0.6349 6.2860 4.7427 0.5565 -6.8530 -10.0146 -1.9473

6.9941 10.6770 2.7370 -5.2922 -8.9367 -3.6250 4.2313 11.0028 8.4412 -1.7043

-9.0612 -7.8359 1.5205 8.7840 5.5569 1.3308 -6.5384 -7.1819 -3.4083 5.6529

10.5822 0.0522 -4.2590 -10.1350 -5.2854 2.7692 9.1226 5.5266 -5.4354 -9.6939

-5.8733 3.4592 9.2241 4.2325 -3.1077 -6.1330 -6.3566 2.1655 10.5872 6.8722

-2.1399 -6.0318 -8.6209 -2.0131 7.4085 6.2506 0.3558 -8.5256 -8.2889 -1.8253

4.6667 8.9501 -1.7030 -5.0722 -6.6795 -0.5846 6.4287 5.7078 -0.2842 -6.2491

-7.6067 -1.3123 5.6383 5.8970 2.7263 -9.0828 -6.8981 0.0168 5.1851 11.2426

-2.1136 -10.8206 -4.9045 -0.8434 9.5390 7.2578 -3.5920 -8.4098 -4.7288 -1.3991

10.9235 5.9464 -1.9180 -6.7085 -6.6277 3.07231 0.3264 3.3078 -5.6364 -9.3671

-5.3363 5.3470 8.2025 0.3092 -4.9505 -8.6505 0.1625 2.2958 8.1701 3.0291

-7.7139 -7.3689 3.1425 10.3639 10.7804 -1.1927 -6.6937 -4.6271 5.3012 10.6891

3.6113 -8.0459 -8.3748 -1.4583 6.6453 5.4941 0.0523 -7.8357 -4.7460 1.9324

9.9685 3.7600 -4.6794 -3.7821 -2.9866 8.4182 6.1975 0.2506 -8.1656 -3.5612

5.2344 9.9963 4.9838 -4.2224 -9.3746 -1.5908 11.0522 8.1827 -0.3214 -11.3551

-5.7090 3.9121 10.9898 4.9617 -5.5324 -8.8929 -0.5175 3.6509 6.0837 -4.1507

-9.6138 -7.0414 1.9005 7.8731 2.8001 -9.9436 -3.8635 0.7394 7.1618 4.4641

-5.5749 -5.0750 -0.2420 8.9311 7.8683 -3.1123 -8.2853 -3.9400 7.0389

10.5480 -1.8533 -6.4040 -4.8988 2.2594 10.1544 3.4588 -7.0473 -4.5673 2.2442

7.0643 0.4916 -7.9303 -9.0185 -3.6419 10.7374 2.8766 -3.4085 -7.9242 -4.1593

12.1280 8.8361 -4.5713 -7.2804 -0.6050 9.6925 7.0833 -4.3910 -6.3754 -4.5601

7.4297 5.4252 2.0188 -9.0284 -4.3360 7.6200 7.3229 -4.9436 -12.5728 -2.1145

5.5637 8.5401 1.7136 -6.6631 -1.2765 4.0440 7.4639 -2.6912 -6.4793 -0.9116

4.9166 5.3861 0.0391 -8.9916 -1.4048 9.3942 8.6392 -3.3612 -7.6315 1.5476

6.5204 7.7742 -3.2388 -8.1871 -0.8108 8.1526 2.0246 -6.0797 -10.5661 1.4871

9.2186 5.5257 -4.5861 -8.6051 3.1716 7.0179 1.8792 -11.9659 -5.1960 4.6238

5.7283 -0.7051 -7.5885 -2.3197 5.0188 3.8126 -7.7472 -10.5954 -1.0988 9.0560

9.4282 -7.6366 -6.7530 -0.1441 6.2347 6.3293 -7.4898 -8.3773 7.4215 10.9196

-0.6563 -10.0258 -7.0634 5.2063 4.1700 -4.5047 -9.7464 3.0110 8.9456 -1.0764

-7.6255 -6.1064 4.6809 9.7986 0.2727 -7.1591 -2.0645 8.5966 5.4061 -6.6128

-5.1519 2.3441 10.1513 -1.4664 -9.9296 -5.3496 7.2690 4.9862 -3.7977 -6.7507

2.6090 10.2573 2.7362 -7.6729 -4.8024 6.0514 8.0016 -4.3617 -6.6826 0.6436

7.8355 -1.0709 -5.1317 -8.0022 6.8244 7.8394 -4.5072 -10.2087 3.1987 6.8365

0.7438 -9.2962 -6.5043 5.7884 6.0699 -3.7352 -9.0128 3.2349 9.1212 -1.8822

-8.3938 0.9678 8.9096 4.4859 -6.2075 -8.3266 5.0453 7.0283 -2.2446 -5.8717

0.2013 9.0345 5.1134 -8.0785 -3.6290 5.9578 3.8150 -7.2473 -7.9194 3.0299

8.6715 -1.0583 -8.2097 1.7821 4.7751 -0.0638 -4.8038 -2.5960 11.3067 3.9246

-6.8596 -2.0046 6.1195 3.8881 -3.5047 -6.5200 3.4206 7.2456 -6.6328 -10.4343

3.3927 10.2978 -1.9740 -6.1521 0.5990 11.0394 1.5473 -8.5500 -7.0950 4.4865

4.6623 -6.3566 -4.0253 0.4709 4.4924 -1.6990 -2.6410 4.4095 9.1636 -4.4050

-6.9789 6.8835 8.5786 -1.5593 -8.2582 0.8854 8.1078 0.8912 -6.2454

2.8480 7.9728 0.8564 -9.7591 -0.6407 6.9144 1.1585 -9.1114 -1.7280 9.7898

0.5248 -8.5238 -3.2486 9.8546 2.1512 -6.6858 -2.4338 10.7942 2.5531 -6.9806

-2.9785 8.0943 2.6007 -6.8424 -6.1788 6.3062 3.6731 -8.9769 -3.8224 9.4119

1.9842 -9.2684 -1.2204 9.0432 2.1834 -7.3082 0.0313 7.6846 4.2083 -6.8879

0.2857 7.1488 3.0421 -10.5980 3.5601 6.4524 -1.0719 -10.328 10.3100 4.2367

-2.9364 -7.7387 -1.5193 8.0523 -5.4848 -8.3136 8.0048 7.7579 -2.2467 -7.4223

2.3898 5.5040 -5.8503 -8.7941 7.1427 6.3938 -8.0504 -4.2286 8.4934 4.2524

-9.8446 1.5934 6.2448 0.7170 -9.1071 -1.1677 9.6621 -1.2212 -9.4567 4.4592

9.7930 -2.5714 -7.9682 7.1492 6.3094 -9.7990 -3.8944 6.6386 1.5734 -9.1559

-1.9693 10.3395 -2.4379 -11.270 11.1178 6.9584 -6.3599 -7.4362 6.0417

7.0178 -7.7777 -1.4494 8.8202 2.5661 -8.1632 1.4482 10.5969 -1.5509 -7.4751

8.1284 8.2974 -7.8512 -2.6835 5.4007 -0.4628 -7.7867 -1.0579 5.9117 -0.9166

-8.7383 6.4730 7.3480 -7.4227 -4.8497 8.1005 2.2575 -7.1387 4.9974

10.0538 -3.4447 -5.2959 4.6296 3.6622 -7.6437 1.9839 13.2039 -2.7565 -10.6727

4.7283 4.8029 -9.4267 -5.1429 8.0628 0.0683 -10.3406 2.3855 7.1269 -8.8245

-3.8839 4.3120 0.5540 -6.5506 1.9608 8.0530 -5.8713 -6.2690 8.0448 4.1430

-12.0511 -4.2331 5.5976 -5.5837 -6.8598 9.1456 0.8963 -8.8348 1.7866 10.1223

-6.0130 -6.5518 9.2246 -2.9344 -10.1999 3.6522 8.7256 -5.8725 -4.9227 7.6548

0.6535 -7.9786 4.4334 7.8801 -8.6174 -4.5211 7.9545 -4.3546 -9.2172 5.5691

7.1747 -8.0327 0.3984 4.7100 -0.9968 -7.7870 10.9118 2.5527 -11.1381 2.7633

7.9501 -8.9759 -2.7860 8.7441 -0.2753 -8.0732 7.0108 7.1376 -10.0321 1.0147

8.2495 -3.5758 -4.6487 6.4147 0.4380 -7.8627 4.9370 4.5576 -9.1727 1.4993

4.2623 -6.4757 -4.2793 7.1267 -3.3641 -6.6913 3.8110 4.7204 -5.0493 0.6936

8.8144 -6.3843 -2.9382 11.1899 -2.5501 -7.4117 5.7313 2.6582 -7.5274 2.5724

5.6066 -7.6553 -4.7531 9.7853 -5.1237 -5.4970 9.8261 0.3647 -6.2756 5.0601

2.9082 -10.4344 1.7828 5.7229 -12.3336 1.4717 8.4369 -5.8995 -3.3740 11.1035

-1.2472 -5.2159 11.2790 4.0008 -9.5382 4.6330 5.0753 -10.7664 2.7382 3.3307

-7.7816 -2.4886 6.8166 -8.1045 -3.8757 8.5790 -1.7775 -6.4950 9.1349 1.8527

-6.4025 4.4336 0.8807 -9.4585 4.8291 5.1396 -7.6327 2.9445 4.9589 -8.5094

2.4424 8.7051 -8.0850 -1.0509 10.0909 -5.5604 -1.9577 6.6252 -10.7162 -2.2256

8.3228 -0.8360 -6.5001 7.2694 -2.7292 -5.1417 10.1172 0.0891 -6.8816 11.0712

1.8128 -7.0769 6.0054 0.9219 -6.5422 3.0692 3.7247 -5.8588 3.8229 1.4760

-7.4896 3.7401 3.7724 -13.9271 4.6634 6.4029 -7.8570 3.1437 3.2542 -7.3002

3.9021 6.8324 -10.6548 1.1327 2.1733 -7.9796 1.9941 4.9268 -9.1913 -0.3485

5.4754 -9.9199 2.6348 7.1139 -10.7765 5.5345 5.8524 -8.3726 2.6696 4.9874

-11.7234 5.2617 4.8669 -6.4090 4.8517 0.8279 -11.1032 4.3336 2.9443

-10.5809 12.6243 1.2923 -9.1553 7.6740 -0.0864 -10.3077 7.1322 1.6754 -4.0189

12.0049 -2.5810 -6.9641 6.7100 -4.4771 -3.0482 6.9879 -0.4643 -1.4104 7.2220

-11.0736 -0.3027 3.1477 -13.4273 2.6889 1.7676 -9.0924 3.0181 3.1744 -6.1674

6.8160 4.0940 -7.0509 6.2824 0.9006 -3.3878 8.2637 -4.9580 -3.2481 10.5308

-8.2105 -3.3148 8.1923 -8.6149 1.2212 3.7588 -5.6479 4.3956 -2.9892 -8.4072

8.6557 -1.1503 -4.9821 12.1615 -3.1649 -1.2539 10.5550 -8.3452 -0.4177 6.6400

-10.1547 3.2020 0.8712 -9.2281 8.4370 -1.0420 -4.76921 3.2513 -3.0937 -2.0347

10.0263 -8.3902 0.8310 5.0096 -8.8817 6.0063 2.4374 -10.3827 8.4896 -0.7046

-1.9322 9.9599 -5.5981 1.3213 6.4105]

It is as follows to invent concrete implementation step:

Step 1, to the sampling gained sampled result r (n) carry out normalization.Performing step is:

At first, ask for the absolute value maximum x of sampled result r (n) _MaxSecondly, press as shown in the formula r (n) is carried out normalization, promptly

x(n)＝r(n)/x _max

X (n) is the data after r (n) normalization.

Step 2, to the data x after the normalization (n), decompose and Hilbert transform according to the empirical mode decomposition algorithm in the Hilbert Huang, obtain hilbert spectrum H (ω, n).Performing step is:

A, according to the empirical mode decomposition of Hilbert Huang, data x (n) is decomposed in M accumulates mode function (IMF) c _i(n) component sum, promptly

$\stackrel{\&OverBar;}{x}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{c}_i\left(n\right)+r_M\left(n\right)$

Wherein, r _M(n) be the residual components after decomposing, M for decompose obtain in accumulate mode function number, the value of M is according to r _M(n) size requirements is determined.

B, carry out Hilbert transform, obtain corresponding analytic signal, a resulting M analytic signal is brought into accumulateing the mode function component in each $\stackrel{\&OverBar;}{x}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{c}_i\left(n\right)+r_M\left(n\right)$ In, obtain:

$\stackrel{\&OverBar;}{x}\left(n\right)\&ap;\mathrm{Re}\left(\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{a}_i\left(n\right)\exp \left(\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20091008141900202.png,A20091008141900211.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}\underset{b=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i\left(b\right)T_s\right)\right)$

$a_i\left(n\right)=\sqrt{c_i^2\left(n\right)+H^2\left[c_i\left(n\right)\right]}$

${\mathrm{\&theta;}}_i\left(n\right)=\arctan \frac{H\left[c_i\left(n\right)\right]}{c_i\left(n\right)}=2\mathrm{\&pi;}\underset{b=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i\left(b\right)T_s$

Wherein, H[c _i(n)] be c _i(n) Hilbert transform, a _i(n) be c _i(n) instantaneous amplitude of analytic signal, f _i(n) be c _i(n) instantaneous frequency, θ _i(n) be c _i(n) instantaneous phase.T _sBe the sampling interval, the sequence number that i accumulates the mode function component in being, i.e. i=1～M; B is the sampled point sequence number, b=1～n; N is a n sampled point.

C, each analytic signal x (n) all is expressed as time n, instantaneous frequency ω, amplitude a _i(n) three-dimensional spectrum, promptly hilbert spectrum H (ω, n):

$H(\mathrm{\&omega;},n)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{b}_i{a}_i\left(n\right)$

Wherein, b _iFor the switch factor, as ω=ω _iThe time, b _i=1, otherwise b _i=0.

Step 3, in the Hilbert three-dimensional spectrum, the spectral amplitude of interference is obviously greater than the spectral amplitude of noise and direct sequence signal.At this moment, (ω n) determines the threshold level that inhibition noise and direct sequence signal are composed, and the spectrum that will be higher than threshold level keeps, and will be lower than the spectrum zero clearing of threshold level, obtains the hilbert spectrum that non-stationary disturbs according to the hilbert spectrum H that obtains.Performing step is:

(1) suppresses (CME) algorithm by consecutive mean, determine to suppress the threshold level of noise and direct sequence signal spectrum, promptly

1. with all Hilbert spectral lines as initial spectral line, and they are put into set I, obtain all Hilbert spectral line values add up and, i.e. A _mWith A _mDivided by Hilbert spectral line sum, obtain the Hilbert spectral line average E[|H (ω, n) |]:

$A_m=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|H(\mathrm{\&omega;},n)\right|$

E[|H(ω，n)|]＝A _m/(N×M)

Wherein, the number that M accumulates the mode function component in being, N is a sampling number.

2. according to the average E[|H of Hilbert spectral line (ω, n) |], adopt following formula estimate the standard deviation sigma of Hilbert spectral line [| H (ω, n) |]:

$\mathrm{\&sigma;}\left[\right|H(\mathrm{\&omega;},n)\left|\right]=\sqrt{\frac{1}{N\&times;M-1}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{|H(\mathrm{\&omega;},j)-E[\left|H(\mathrm{\&omega;},n)\right|\left]\right|}^2}$

According to the standard deviation sigma that obtains [| H (ω, n) |], determine thresholding η:

η＝E[|H(ω，n)|]+3σ[|H(ω，n)|]

The mould and the thresholding η that 3. will gather the Hilbert spectral line among the I compare, and are considered to disturbed Hilbert spectral line greater than the Hilbert spectral line of this thresholding, remove from set I.Set I is updated.

4. after upgrading set I, upgrade A _mAnd E[|H (ω, n) |], repeat afterwards 2. and 3., till the mould that does not have the Hilbert spectral line is greater than thresholding η.Finally obtain not having the set I ' of the mould of Hilbert spectral line greater than thresholding η.

5. with the Hilbert spectral line of set among the I ', estimate final thresholding η=E[|H (ω, n) |]+3 σ [| H (ω, n) |]

(2) hilbert spectrum that will be higher than thresholding η level keeps, and will be lower than the hilbert spectrum zero clearing of thresholding η level, thereby obtain the hilbert spectrum that non-stationary disturbs.

Step 4, the hilbert spectrum that disturbs according to non-stationary and the corresponding relation between the frequency seek out the instantaneous frequency that non-stationary disturbs, by the parameter of this instantaneous frequency decision self adaptation infinite impulse response (IIR) lattice type trapper.

Suppose with sample frequency f _s/ 2 is normalized frequency, and quantization unit is 400, and it is f that non-stationary disturbs the instantaneous frequency of hilbert spectrum correspondence _H, the instantaneous frequency f that then actual non-stationary disturbs is:

f＝(f _H/400)×(f _s/2)

Wherein, f _sBe sample frequency, f _HFor disturbing the normalization instantaneous frequency.By curve fit, obtain interference signal instantaneous frequency curve, as shown in Figure 3, this is higher than the instantaneous frequency precision that adopts the Wigner-Ville distribution to estimate.

Step 5, the gained base band of will sampling receive data and the adaptive notch filter impulse response carries out convolution algorithm, and the non-stationary data after interference that is inhibited is carried out related operation with these data and synchronous PN sign indicating number sequence, can finish the despreading that receives data.

The infinite impulse response that is adopted (IIR) lattice type trapper structure is determined the relevant parameter k of adaptive lattice type infinite impulse response (IIR) trapper as shown in Figure 4 by the instantaneous frequency of non-stationary interference ₀, a ₀, k ₁And a ₁, promptly

${\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="A20091008141900203.png,A20091008141900221.png"\; state="\{\{state\}\}">k</figure-callout>}}_0=-\frac{2r\cos {\mathrm{\&omega;}}_0}{1+{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A20091008141900202.png,A20091008141900211.png"\; state="\{\{state\}\}">r</figure-callout>}}^2}$

$a_0=-\frac{2\mathrm{\&alpha;r}\cos {\mathrm{\&omega;}}_0}{1+{\mathrm{\&alpha;}}^2{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A20091008141900202.png,A20091008141900211.png"\; state="\{\{state\}\}">r</figure-callout>}}^2}$

k ₁＝r ²

a ₁＝α ²r ²

Wherein, r is a polar radius, desirable r=1; α is the parameter of a decision trapper width and the degree of depth, desirable α=0.95; ω ₀Be trap frequency, ω ₀=2 π f/f _sThereby can get adaptive lattice type trapper transfer function be

$H\left(z\right)=\frac{1-2r\cos {\mathrm{\&omega;}}_0{z}^{-1}+r^2{z}^{-2}}{1-2\mathrm{\&alpha;}r\cos {\mathrm{\&omega;}}_0{z}^{-1}+{\mathrm{\&alpha;}}^2{r}^2{z}^{-2}}$

According to non-stationary interference inhibition block diagram as shown in Figure 5, the impulse response of the gained data of will sampling and self adaptation infinite impulse response (IIR) lattice type trapper carries out convolution, the non-stationary data after interference that is inhibited sequence

, as shown in table 2, promptly

$\hat{s}\left(n\right)=r\left(n\right)*h\left(n\right)$

Table 2 suppresses non-stationary data after interference sequence

Spectrogram as shown in Figure 6, the visible frequency spectrum that non-stationary has been disturbed suppresses.With data

Carry out related operation with synchronous PN sign indicating number sequence, just can finish the despreading that the base band after the inhibition interference is received data.

Claims (2)

1,, it is characterized in that concrete steps are as follows based on the non-stationary disturbance restraining method of Hilbert Huang and adaptive notch:

Step 1, the sampled result that base band is received data are carried out normalization;

Step 2, to the data after the normalization, decompose and Hilbert transform according to the empirical mode decomposition algorithm in the Hilbert Huang, obtain hilbert spectrum;

The hilbert spectrum that step 3, basis obtain determines to suppress the threshold level of noise and direct sequence signal spectrum, and the spectrum that will be higher than threshold level keeps, and will be lower than the spectrum zero clearing of threshold level, obtains the hilbert spectrum of non-stationary interference, and the specific implementation step is:

(1) suppresses algorithm by consecutive mean, determine to suppress the threshold level of noise and direct sequence signal spectrum, promptly

1. with all Hilbert spectral lines as initial spectral line, and they are put into set I, obtain adding up and A of all Hilbert spectral line values _m, with A _mDivided by Hilbert spectral line sum, obtain the Hilbert spectral line average E[|H (ω, n) |]:

$A_m=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|H(\mathrm{\&omega;},n)\right|$

E[|H(ω，n)|]＝A _m/(N×M)

Wherein, the number that M accumulates the mode function component in being, N is a sampling number;

2. according to the average E[|H of Hilbert spectral line (ω, n) |], adopt following formula estimate the standard deviation sigma of Hilbert spectral line [| H (ω, n) |]:

$\mathrm{\&sigma;}\left[\right|H(\mathrm{\&omega;},n)\left|\right]=\sqrt{\frac{1}{N\&times;M-1}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}|H(\mathrm{\&omega;},j)-E[\left|H(\mathrm{\&omega;},n)\right|]{|}^2}$

According to the standard deviation sigma that obtains [| H (ω, n) |], determine thresholding η

η＝E[|H(ω，n)|]+3σ[|H(ω，n)|]

The mould and the thresholding η that 3. will gather the Hilbert spectral line among the I compare, and are considered to disturbed Hilbert spectral line greater than the Hilbert spectral line of this thresholding, remove from set I, and set I is updated;

4. after upgrading set I, upgrade A _mAnd E[|H (ω, n) |], repeat afterwards 2. and 3., till the mould that does not have the Hilbert spectral line is greater than thresholding η, finally obtain not having the set I ' of the mould of Hilbert spectral line greater than thresholding η;

5. with the Hilbert spectral line of set among the I ', estimate final thresholding η=E[|H (ω, n) |] 3 σ [| H (ω, n) |];

(2) hilbert spectrum that will be higher than final thresholding η level keeps, and will be lower than the hilbert spectrum zero clearing of thresholding η level, thereby obtain the hilbert spectrum that non-stationary disturbs;

Step 4, according to the hilbert spectrum that non-stationary disturbs, seek out the instantaneous frequency that non-stationary disturbs, the parameter by this instantaneous frequency decision self adaptation infinite impulse response lattice type trapper obtains the adaptive notch filter impulse response thus;

Step 5, sampled result in the step 1 and adaptive notch filter impulse response are carried out convolution algorithm, the non-stationary data after interference that is inhibited is carried out related operation with these data and synchronous PN sign indicating number sequence, finishes the despreading that receives data.

2, the non-stationary disturbance restraining method based on Hilbert Huang and adaptive notch according to claim 1 is characterized in that, the specific implementation step of step 2 is as follows:

A, according to the empirical mode decomposition of Hilbert Huang, will be decomposed in M through the data x after the normalization that step 1 obtains (n) and accumulate mode function (IMF) c _i(n) component sum, promptly

$\stackrel{\&OverBar;}{x}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{c}_i\left(n\right)+r_M\left(n\right)$

Wherein, r _M(n) for decomposing the back residual components, M for decompose obtain in accumulate mode function number, the value of M is according to r _M(n) size requirements is determined;

B, carry out Hilbert transform, obtain corresponding analytic signal, a resulting M analytic signal is brought into accumulateing the mode function component in each $\stackrel{\&OverBar;}{x}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{c}_i\left(n\right)+r_M\left(n\right)$ In, x (n) is expressed as

$\stackrel{\&OverBar;}{x}\left(n\right)\&ap;\mathrm{Re}\left(\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{a}_i\left(n\right)\exp \left(j2\mathrm{\&pi;}\underset{b=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i\left(n\right)T_s\right)\right)$

$a_i\left(n\right)=\sqrt{c_i^2\left(n\right)+H^2\left[c_i\left(n\right)\right]}$

${\mathrm{\&theta;}}_i\left(n\right)=\arctan \frac{H\left[c_i\left(n\right)\right]}{c_i\left(n\right)}=2\mathrm{\&pi;}\underset{b=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i\left(n\right)T_s$

Wherein, H[c _i(n)] be c _i(n) Hilbert transform, a _i(n) be c _i(n) instantaneous amplitude of analytic signal, f _i(n) be c _i(n) instantaneous frequency, θ _i(n) be c _i(n) instantaneous phase; T _sBe the sampling interval, the sequence number that i accumulates the mode function component in being, i.e. i=1～M; B is the sampled point sequence number, b=1～n; N is a n sampled point;

C, x (n) is expressed as time n, instantaneous frequency ω, amplitude a _i(n) three-dimensional spectrum, promptly hilbert spectrum H (ω, n):

$H(\mathrm{\&omega;},n)=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{b}_i{a}_i\left(n\right)$

Wherein, b _iFor the switch factor, as ω=ω _iThe time, b _i=1, otherwise b _i=0; The number of accumulateing the mode function component in the M representative, the sequence number that i accumulates the mode function component in being, i.e. i=1～M.

Images