CN1847877A - Passive channel correcting method based on non-linear antenna array - Google Patents

Abstract

The passive channel correcting method based on non-linear antenna array includes: configuring the antenna array as non-linear form containing translation invariant array element pairs; detecting single-bearing sea echo with the translation invariant array element pairs; estimating the channel amplitude mismatch ratio with the sea echo to realize amplitude correction; and estimating the channel phase mismatch ratio with the single-bearing sea echo with corrected amplitude and known array position information to realize phase correction. The present invention is one true passive channel correcting method needing no any auxiliary signal source, and has the advantages of no ship echo interference and multipath effect, high precision and robustness, less operation amount, capacity of uninterrupted operation, greatly improved radar application flexibility, raised detection performance, and low cost.

Description

A kind of method for correcting passive channels based on non-linear antenna array

Technical field

The present invention relates to a kind of method of high-frequency ground wave radar being carried out the passive channel correction based on non-linear antenna array.

Background technology

High-frequency ground wave radar utilizes frequency electromagnetic waves in the little characteristics of conduction ocean surface diffraction propagation decay, adopts vertical polarized antenna radiation electric wave, can the following moving targets such as naval vessel, aircraft and guided missile that occur of over-the-horizon detection sea level sight line.In addition, high-frequency ground wave radar utilizes single order scattering and the second order dispersion mechanism of ocean surface to frequency electromagnetic waves, extracts extra large attitude information such as wind field, Lang Chang, flow field from radar return, can realize to marine environment on a large scale, high precision and round-the-clock real-time monitoring.

Because the effect of multiple factors such as the difference of hardware itself, the non-ideal characteristic of receiving cable, surrounding enviroment influence, the magnitude-phase characteristics of each passage of aerial array is discrepant in the reality, cause echoed signal inconsistent, generally be referred to as the passage mismatch through amplitude and phase change (complex gain) behind the different passages.The passage mismatch causes that the error of beam scanning and DOA estimation increases, even complete failure, is one of key issue that influences the high-frequency ground wave radar detection performance.In order to guarantee that radar can effectively work, must take measures to make the passage mismatch to be limited in certain scope: on the one hand, to make each passage when making, guarantee its consistance by adequate measures (as the components and parts screening) as far as possible; On the other hand, can measure or estimate, then by proofreading and correct the difference of further dwindling channel characteristic the passage mismatching.

Existing channel correcting method can be divided into active and passive two classes.In the active correction method, auxiliary source is placed antenna array the place ahead open area emission correction signal enough far away, measure the output of each receiving cable again, the phase differential according to deduction array manifold position, known signal source orientation causes can obtain passage mismatch information.In passive bearing calibration, need not accurately known auxiliary source of orientation, directly utilize the measured data and some prioris (as array format) estimating channel mismatching that receive, compensate correction then.Some passive bearing calibration can also realize the estimation of uniting of aspect and passage mismatch.In " Estimation of Spatial Spectrum and application thereof " (publishing house of Chinese University of Science and Technology 1997) book that Liu Deshu, Luo Jingqing etc. writes, general passive bearing calibration is elaborated.

Be subjected to influence of various factors such as topographic condition, operation wavelength, radio wave propagation, radar system, aerial array, (firmly) target echo, ocean clutter, noise, the channel correcting of high-frequency ground wave radar implements the comparison difficulty, existing method all can only solve subproblem, and waste time and energy, with high costs, cost is very big.Radar antenna battle array the place ahead is the sea, as adopt the active correction method then auxiliary source can only be placed on naval vessel or the island, it safeguards very trouble and expensive, is difficult to long-term stable operation.Existing passive bearing calibration generally needs repeatedly complicated interative computation, and calculated amount is very big, the requirement of real time surely that differs, and might converge on local optimum, rather than global optimum, so that full of prunes estimated value appears.The difference of actual radar system and ideal model also makes the applicable elements of general passive bearing calibration not to be met, and is difficult to practicality.Channel correcting has become restriction high-frequency ground wave radar detection performance and has hindered the great technical barrier of its practical application, must be properly settled.

Wuhan University's wave propagation laboratory was once considered the known natural or artificial object in sea the reflected signal of radar wave as correction signal.As long as distance, the speed of reflection sources are known, just can from echo, detect correction signal, go out the mismatching of each passage then according to known reflection sources DOA estimation.Its concrete implementation detail can be with reference to No. 03128238.5 Chinese invention patent application " a kind of method of utilizing marine echo to carry out array channel calibration ".This invention can utilize surveys known fixation reflex things such as island, beacon and drilling platform in the marine site, does not have the placement and the maintenance issues of auxiliary source, also need not extra hardware spending, has realized online in real time from normal moveout correction, has certain practical value.But this invention is actually a kind of special active correction method, and range of application and actual effect are limited, are not suitable for the marine site that does not have the known fixed reverberation, and still is subjected to the influence of unfavorable factors such as noise, naval vessel echo, multipath effect.Because this invention has proposed a kind of separation and detection technique to the nonoverlapping folk prescription of frequency spectrum position marine echo, satisfied the basic demand of the method for correcting passive channels of the present invention's proposition, therefore will introduce it below.

High-frequency ground wave radar generally adopts FMICW (frequency modulated interrupted continuouswave, abbreviation FMICW) system.People such as Rafaat Khan deliver is entitled as " high-frequency ground wave radar target detection and tracking " (Target Detection and Tracking With a High Frequency Ground Wave Radar, IEEE Journal ofOceanic Engineering, 1994,19 (4): in the paper 540～548) this is had a detailed description.Under this waveform system, after marine echo (comprising extra large clutter and hard goal echo) enters receiver, can be through mixing, low-pass filtering, A/D conversion and Two-dimensional FFT (as shown in Figure 1) apart from-two-dimentional echo spectrum of Doppler's (speed) (as shown in Figure 2).In two-dimentional echo spectrum, a large amount of marine echos that radar receives separate with speed by distance, are dispersed on a lot of spectrum points.When the FFT second time (Doppler-shift) the coherent accumulation time is grown (about 10 minutes), radar can obtain very high speed resoluting force, the spectrum point corresponding with marine echo can reach more than 1000 in the two dimension echo spectrum, is well suited for statistical method the nonoverlapping folk prescription of its intermediate frequency spectrum position echo being detected.

The detection of folk prescription position echo is to realize by the statistical study to the two-dimentional echo spectrum output of particular form array (as shown in Figure 3).The particular form array is made of array element 1～4, and its position coordinates is (x _i, y _i), certain corresponding two-dimentional echo spectrum point is output as Y _i, i=1,2,3,4. Array element 1 and 2 is formed array element idol A ₁, 3 and 4 form array element idol A ₂, A ₁With A ₂Between have translation invariance, then have

$\left\{\begin{array}{c}(x_2,y_2)=(x_1+d,y_1)\\ (x_4,y_4)=(x_3+d,y_3)\end{array}\right.$

Order ${\mathrm{\&eta;}}_1=\frac{Y_2{Y}_3}{Y_1{Y}_4},$ Proof is not having noise ideally easily, the corresponding η of single azimuth spectrum point in the two-dimentional echo spectrum ₁Be an only invariant relevant, might as well be designated as η with the passage mismatch ₁'.Noise is inevitably in the real system, the η that single azimuth spectrum point is corresponding ₁Concentrate and be distributed in η ₁' near.On the other hand, by simple analysis and numerical simulation as can be known, the η that multi-faceted spectrum point is corresponding ₁Be one with target range, (radially) speed, orientation and all relevant variable of echoed signal amplitude because the randomness of these target components, η ₁To be the random dispersion state.Comprehensive above the analysis is with all surpass the corresponding η of spectrum point of certain signal-noise ratio threshold in the two-dimentional echo spectrum ₁Be marked on the complex plane, then have and only have a zone (η ₁' near) appearance height clustering phenomena, wherein most of η ₁Value is corresponding to single azimuth spectrum point.Order ${\mathrm{\&eta;}}_2=\frac{Y_2{\mathrm{<figure-callout\; id="1"\; label="Y"\; filenames="A20061007136000142.png,A20061007136000152.png"\; state="\{\{state\}\}">Y</figure-callout>}}_1^{*}}{Y_4{\mathrm{<figure-callout\; id="3"\; label="Y"\; filenames="A20061007136000153.png"\; state="\{\{state\}\}">Y</figure-callout>}}_3^{*}},$ By similarly analyzing as can be known η with the front ₂Also accumulation area can appear, wherein most of η on complex plane ₂Value is corresponding to single azimuth spectrum point.Order ${\mathrm{\&eta;}}_3=\frac{Y_2{Y}_4^{*}}{Y_1{\mathrm{<figure-callout\; id="3"\; label="Y"\; filenames="A20061007136000153.png"\; state="\{\{state\}\}">Y</figure-callout>}}_3^{*}},$ η ₃Equally accumulation area can appear, wherein most of η on complex plane ₃Value is corresponding to single azimuth spectrum point.

Find the η that multi-faceted spectrum point is corresponding through theoretical analysis and numerical simulation ₁, η ₂And η ₃Fall into simultaneously that the accumulation area probability is minimum separately, therefore can use η ₁, η ₂And η ₃Can fall into accumulation area separately simultaneously as the criterion that detects single azimuth spectrum point.A ₁And A ₂As a translation invariant array element idol group, constituted the particular form array of detection folk prescription position echo (spectrum point).If translation invariant array element idol group surpasses one in the array, then will whether be detected as criterion by a plurality of array element idol groups simultaneously, can further filter out single azimuth spectrum point.The array format that contains translation invariant array element idol group is very common, as even linear array (or uniform planar battle array).

Summary of the invention

At the limitation of existing method, the objective of the invention is to utilize the folk prescription position marine echo of high-frequency ground wave radar reception, a kind of method for correcting passive channels based on non-linear antenna array is provided, to reduce passage width of cloth phase mismatch, improve the radar system performance.The non-linear antenna array here is meant not to be the array that all bays all are located along the same line.

To achieve these goals, technical scheme provided by the invention is: a kind of method for correcting passive channels based on non-linear antenna array, aerial array are set to contain the non-rectilinear form of translation invariant array element idol group; Detect folk prescription position marine echo by translation invariant array element idol group; Utilize folk prescription position marine echo estimating channel amplitude mismatch coefficient, realize amplitude correction; Utilize and passed through the folk prescription position marine echo of amplitude correction and known array position information estimating channel phase mismatch coefficient, realize phase correction.

In said method, can utilize folk prescription position marine echo to pass through ${\hat{g}}_i=\sqrt{\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|Y_i\left(l\right)\right|}^2/\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|Y_1\left(l\right)\right|}^2}$ Or ${\hat{g}}_i=\sqrt{\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\frac{{\left|Y_i\left(l\right)\right|}^2}{{\left|Y_1\left(l\right)\right|}^2}}$ Or ${\hat{g}}_i=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\left|Y_i\left(l\right)\right|/\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\left|Y_1\left(l\right)\right|$ Or ${\hat{g}}_i=\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\frac{\left|Y_i\left(l\right)\right|}{\left|Y_1\left(l\right)\right|}$ Estimating channel amplitude mismatch coefficient, wherein Be the channel amplitude mismatching estimated value of array element i, i=1,2, Λ, M, M are element number of array, Y _i(l) l the folk prescription position echo output that receives for array element i, l=1,2, Λ, L, L are folk prescription position echo number.

The present invention can also utilize and pass through through the folk prescription position marine echo of amplitude correction and known array position information $\widehat{\mathrm{\&Psi;}}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-f^{\&prime;}\left(\mathrm{\&Psi;}\right)\left|\right|}^2$ Or $\widehat{\mathrm{\&Psi;}}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-\tilde{f}\left(\mathrm{\&Psi;}\right)\left|\right|}^2$ Estimating channel phase mismatch coefficient, wherein

Ψ＝[θ ₁，θ ₂，Λ，θ _L，φ ₂，φ ₃，Λ，φM] ^T

$Y=\left[\begin{array}{c}{Y}_2\\ Y_3\\ M\\ Y_M\end{array}\right]$

Y _i＝[Y _i(1)，Y _i(2)，Λ，Y _i(L)] ^T

$f^{\&prime;}\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="A20061007136000142.png,A20061007136000152.png"\; state="\{\{state\}\}">f</figure-callout>}}_2^{\&prime;}\left(\mathrm{\&Psi;}\right)\\ {\mathrm{<figure-callout\; id="3"\; label="f"\; filenames="A20061007136000153.png"\; state="\{\{state\}\}">f</figure-callout>}}_3^{\&prime;}\left(\mathrm{\&Psi;}\right)\\ M\\ f_M^{\&prime;}\left(\mathrm{\&Psi;}\right)\end{array}\right]$

$f_i^{\&prime;}\left(\mathrm{\&Psi;}\right)={[Y_1\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},Y_1\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},Y_1\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_i]}]}^T$

$\tilde{f}\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}{\tilde{f}}_2\left(\mathrm{\&Psi;}\right)\\ {\tilde{f}}_3\left(\mathrm{\&Psi;}\right)\\ M\\ {\tilde{f}}_M\left(\mathrm{\&Psi;}\right)\end{array}\right]$

${\tilde{f}}_i\left(\mathrm{\&Psi;}\right)={\left[\hat{A}\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},\hat{A}\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},\hat{A}\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_i]}\right]}^T$

$\hat{A}\left(l\right)=\frac{1}{M}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{Y}_i\left(l\right)e^{-j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)+{\mathrm{\&phi;}}_i]}=\frac{1}{M}\{\underset{i=2}{\overset{M}{\mathrm{\&Sigma;}}}{Y}_i\left(l\right)e^{-j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)+{\mathrm{\&phi;}}_i]}+Y_1\left(l\right)\}$

θ _lBe the angle of arrival of l folk prescription position echo, φ _iBe the channel phases mismatching of array element i, (x _i, y _i) be the element position coordinate, array element 1 is true origin, i.e. (x ₁, y ₁)=(0,0), λ is the echoed signal wavelength,

Estimated value for Ψ.

The present invention selects 3 array elements and constitutes a triangle battle array from all array elements, this triangle battle array is reduced the processing of global optimization's dimension, and estimate and local optimization method estimating channel phase mismatch coefficient by initial value, to reduce the operand of multidimensional parameter estimation.

1) from all array elements, selects earlier 3 array elements and constitute a triangle battle array, select 3 folk prescription position echoes again and be used for parameter estimation; 2) be the benchmark passage with the arbitrary array element in this triangle battle array, adopt global optimization's method to estimate the phase mismatch coefficient of other two passages and the angle of arrival of 3 folk prescription position echoes; 3) increase the parameter estimation that a folk prescription position echo is used for the triangle battle array, draw the angle of arrival of this folk prescription position echo; 4) set by step 3) draw the angle of arrival of other folk prescription position echo; 5) triangle battle array and another array element are combined into 4 yuan of battle arrays, again all folk prescription position echoes are used for the parameter estimation of these 4 yuan of battle arrays, newly added the channel phases mismatching of array element; 6) set by step 5) draw the channel phases mismatching of other array element.

The present invention also can be after the process step 4), the parameter estimation that all folk prescription position echoes is used for the triangle battle array, with the folk prescription position echo angle of arrival of having tried to achieve and the estimated value of channel phases mismatching is initial value, adopt the suboptimization method to try to achieve the more accurate estimated value of these parameters, carry out the step 5) and 6 of back then).

Through after the step 6), all folk prescription position echoes are used for the parameter estimation of whole array, be initial value with the folk prescription position echo angle of arrival of having tried to achieve and the estimated value of channel phases mismatching, adopt the suboptimization method to try to achieve the more accurate estimated value of these parameters.

For the L shaped battle array of M unit, array element 1,2, M are used for parameter estimation as selected triangle battle array in the step 1), realize the channel phases correction.

For 4 yuan of T shape battle arrays, array element 1,2,4 or array element 2,3,4 are used for parameter estimation as selected triangle battle array in the step 1), realize the channel phases correction.

For 4 yuan of rectangle battle arrays, wherein any 3 array elements are used for parameter estimation as selected triangle battle array in the step 1), realize the channel phases correction.

Advantage of the present invention is the Practical Performance that it is outstanding: fully without any need for auxiliary source, be a kind of real method for correcting passive channels.The present invention has only utilized folk prescription position echo, does not have thorny problems such as ship echo interference that the source channels bearing calibration faces, multipath effect; Utilized a large amount of (can reach more than 100) high strength folk prescription positions marine echo, information utilization is higher, has good precision and robustness; Operand is less when adopting some special processings, the energy requirement of real time; Because marine echo always exists in a large number, this method can uninterrupted for a long time steady operation; Improved the application flexibility of radar greatly, antenna system can arbitrarily be changed, increases and decreases, move, and this is former to be unthinkable; When improving detection performance, significantly reduced the development cost and the maintenance cost of radar.

Description of drawings

Fig. 1 is the high-frequency ground wave radar fundamental diagram;

Fig. 2 is the two-dimentional echo spectrogram of high-frequency ground wave radar distance-Doppler (speed);

Fig. 3 is the particular form array synoptic diagram that is used to detect folk prescription position echo;

Fig. 4 is any non-rectilinear array synoptic diagram of M unit of the present invention;

Fig. 5 is a triangle battle array synoptic diagram;

Fig. 6 is a M unit L shaped battle array synoptic diagram;

Fig. 7 is 4 yuan of rectangle battle array synoptic diagram;

Fig. 8 is 4 yuan of T shape battle array synoptic diagram;

Embodiment

Key of the present invention is by setting up the folk prescription position echo signal model that non-linear antenna array receives the channel correcting problem have been changed into the parameter estimation problem, and has obtained comparatively accurate passage mismatch estimation thus.

At first consider M shown in Figure 4 (M 〉=3) the unit situation of non-linear antenna array arbitrarily, below step by step the present invention's embodiment in this case is described.

(1) signal model

If the element position coordinate of non-linear antenna array shown in Figure 4 is respectively (x _i, y _i) (i=1,2, Λ, M), wherein array element 1 is true origin, i.e. (x ₁, y ₁)=(0,0).Marine echo can be regarded plane wave as, supposes that detected folk prescription position echo has L (L 〉=3) individual from the two-dimentional echo spectrum of distance-Doppler's (speed), then the l that receives of array element i (l=1,2, Λ, L) individual folk prescription position echo is output as

$Y_i\left(l\right)=g_i{e}^{j{\mathrm{\&phi;}}_i}[A\left(l\right)e^{\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20061007136000142.png,A20061007136000152.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)}+W_i\left(l\right)]---\left(1\right)$

Wherein, θ _lAnd A (l) is respectively the angle of arrival and the complex amplitude of l folk prescription position echo, g _iAnd φ _iBe respectively channel amplitude mismatching and the phase mismatch coefficient of array element i, λ is the echoed signal wavelength, W _i(l) be additive noise.With array element 1 is the benchmark passage, promptly $g_1{e}^{j{\mathrm{\&phi;}}_1}=1$ Then have by (1) formula

Y ₁(l)＝A(l)+W ₁(l) (2)

To additive noise W _i(l) adopt following hypothesis:

1) corresponding to the W of different i or l _i(l) separate;

2) W _i(l) for having identical variances sigma ²White Gaussian noise.

Then (1) and (2) formula has constituted passage mismatch estimated signals model.

(2) channel amplitude is proofreaied and correct

The channel amplitude mismatch of array element i is estimated and can directly be passed through

${\hat{g}}_i=\sqrt{\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|Y_i\left(l\right)\right|}^2/\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|Y_1\left(l\right)\right|}^2}---\left(3\right)$

Calculating is tried to achieve, according to

Can proofread and correct the channel amplitude mismatch.(3) formula can also have other form, as ${\hat{g}}_i=\sqrt{\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\frac{{\left|Y_i\left(l\right)\right|}^2}{{\left|Y_1\left(l\right)\right|}^2}},$ ${\hat{g}}_i=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\left|Y_i\left(l\right)\right|/\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\left|Y_1\left(l\right)\right|,$ ${\hat{g}}_i=\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\frac{\left|Y_i\left(l\right)\right|}{\left|Y_1\left(l\right)\right|},$ Or the like.

(3) channel phases is proofreaied and correct

The array received passage is through behind the amplitude correction, and l the folk prescription position echo that array element i receives is output as

$Y_i\left(l\right)=g_i{e}^{j{\mathrm{\&phi;}}_i}[A\left(l\right)e^{\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20061007136000142.png,A20061007136000152.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)}+W_i\left(l\right)]---\left(4\right)$

Definition

Y _i＝[Y _i(1)，Y _i(2)，Λ，Y _i(L)] ^T i＝2，3，Λ，M

$Y=\left[\begin{array}{c}{Y}_2\\ Y_3\\ M\\ Y_M\end{array}\right]$

Ψ＝[θ ₁，θ ₂，Λ，θ _L，φ ₂，φ ₃，Λ，φ _M] ^T

$f_i\left(\mathrm{\&Psi;}\right)={\left[A\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},A\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},A\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_i]}\right]}^T$

$f\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}{f}_2\left(\mathrm{\&Psi;}\right)\\ f_3\left(\mathrm{\&Psi;}\right)\\ M\\ f_M\left(\mathrm{\&Psi;}\right)\end{array}\right]$

Be parameter vector to be estimated with Ψ, according to (4) formula and noise model hypothesis, the estimated value that adopts maximum likelihood method (please refer to a prominent personage and show " modern signal processing ", publishing house of Tsing-Hua University, 1994) can get Ψ is

$\widehat{\mathrm{\&Psi;}}\arg \underset{\mathrm{\&Psi;}}{\min }\left\{{[Y-f]\mathrm{\&Psi;}}^H\right[Y-f\left(\mathrm{\&Psi;}\right)\left]\right\}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-f\left(\mathrm{\&Psi;}\right)\left|\right|}^2---\left(5\right)$

Wherein,

The value of Ψ when representing back expression formula minimization, ‖ X ‖ ²The 2-norm of expression vector X.Obviously, the channel phases mismatching and the folk prescription position echo angle of arrival have realized uniting estimation.

A (l) is included in noisy Y _i(l) in, can not directly obtain f _iAlso be impossible directly construct (Ψ), what be used for the estimation of channel phases mismatch in the reality can not be (5) formula, need improve.General signal to noise ratio (S/N ratio) condition (〉=20dB) under, can get Y by (2) formula ₁(l) ≈ A (l) uses Y ₁(l) A (l) in replacement (5) formula, the estimated value practical expression that obtains Ψ is

$\widehat{\mathrm{\&Psi;}}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-f^{\&prime;}\left(\mathrm{\&Psi;}\right)\left|\right|}^2---\left(6\right)$

Wherein

$f^{\&prime;}\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="A20061007136000142.png,A20061007136000152.png"\; state="\{\{state\}\}">f</figure-callout>}}_2^{\&prime;}\left(\mathrm{\&Psi;}\right)\\ {\mathrm{<figure-callout\; id="3"\; label="f"\; filenames="A20061007136000153.png"\; state="\{\{state\}\}">f</figure-callout>}}_3^{\&prime;}\left(\mathrm{\&Psi;}\right)\\ M\\ f_M^{\&prime;}\left(\mathrm{\&Psi;}\right)\end{array}\right]$

$f_i^{\&prime;}\left(\mathrm{\&Psi;}\right)={[Y_1\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},Y_1\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},Y_1\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_i]}]}^T$

According to

Can proofread and correct the channel phases mismatch.As replace the A (l) in (5) formula with other amount, (6) formula can also have other form, as $\widehat{\mathrm{\&Psi;}}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-\tilde{f}\left(\mathrm{\&Psi;}\right)\left|\right|}^2,$ Wherein Y is identical with the front with the Ψ definition,

Be defined as

$\tilde{f}\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}{\tilde{f}}_2\left(\mathrm{\&Psi;}\right)\\ {\tilde{f}}_3\left(\mathrm{\&Psi;}\right)\\ M\\ {\tilde{f}}_M\left(\mathrm{\&Psi;}\right)\end{array}\right]$

${\tilde{f}}_i\left(\mathrm{\&Psi;}\right)={\left[\hat{A}\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},\hat{A}\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},\hat{A}\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_i]}\right]}^T$

$\hat{A}\left(l\right)=\frac{1}{M}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{Y}_i\left(l\right)e^{-j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)+{\mathrm{\&phi;}}_i]}=\frac{1}{M}\{\underset{i=2}{\overset{M}{\mathrm{\&Sigma;}}}{Y}_i\left(l\right)e^{-j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)+{\mathrm{\&phi;}}_i]}+Y_1\left(l\right)\}$

Theoretical analysis and emulation experiment show, the applicable elements of above-mentioned channel phases bearing calibration is to adopt the non-rectilinear array and exist at least two angles to differ the folk prescription position echo that is not 0 ° or 180 °.The channel phases mismatch is estimated to be actually a multidimensional parameter estimation problem owing to relate to all array elements, try to achieve by multi-dimensional search, and this just relates to the selection of optimization method.Owing to there is local minimum, must adopt global optimization's method (to please refer to the paper " from the local minimum to the global optimum " that Tang Fang, Wang Ling deliver, computer engineering and application, 2002.6:56～58) the channel phases mismatch is estimated, operand is very surprising under the situation of parameter more (can reach more than 100), does not reach the real-time requirement.The present invention has adopted special processing to reduce operand, will elaborate to this below.

Figure 5 shows that the simplest non-rectilinear array: by the triangle battle array that constitutes of 3 array elements of straight line altogether not.If the channel phases mismatching and the folk prescription position echo angle of arrival that only utilize 3 folk prescription position echoes to unite to estimate this triangle battle array then can be judged as 5 dimensions according to (6) formula and search for (array element being arranged as the benchmark passage).Because dimension is less, even adopt global optimization's methods such as simulated annealing, evolutionary computation, Chaos Search, random sampling, the operand of this 5 dimension search is also little, but requirement of real time.

For any non-rectilinear array of M unit shown in Figure 4, in order to reduce the operand of multidimensional parameter estimation, can utilize certain the triangle battle array that wherein comprises to treat earlier to estimate parameter to carry out initial value and estimate, adopt suboptimization method (as method of steepest descent) to try to achieve more accurate estimation again, concrete steps are as follows:

1) from all array elements, selects earlier 3 array elements and constitute a triangle battle array, from the echo of a large amount of folk prescriptions position, select 3 again and be used for parameter estimation.

2) be the benchmark passage with certain array element in this triangle battle array, adopt global optimization's method to estimate the phase mismatch coefficient of other two passages and the angle of arrival of 3 folk prescription position echoes, this is 5 dimension search.

3) increase the parameter estimation that a folk prescription position echo is used for the triangle battle array, then waiting to estimate parameter in (6) formula has also increased by one (angle of arrival of promptly newly-increased folk prescription position echo), has become 6 dimension search by 5 dimension search; With step 2) 5 estimates of parameters substitutions of trying to achieve contain 6 (6) formulas of waiting to estimate parameter, adopt only surplus (6) formula of waiting to estimate parameter to estimate the angle of arrival of newly-increased folk prescription position echo again, and this is 1 dimension search.

4) 3 folk prescription position echoes except selecting in the step 1), the angle of arrival estimated value of other L-3 folk prescription position echo all can adopt the method in the step 3) to try to achieve.

5) all L folk prescription position echoes are used for the parameter estimation of triangle battle array, with L+2 parameter (L the folk prescription position echo angle of arrival and 2 the channel phases mismatchings) estimated value of having tried to achieve is initial value, adopts the suboptimization method to try to achieve the more accurate estimated value of these parameters.

6) triangle battle array and another array element are combined into 4 yuan of battle arrays, all L folk prescription position echoes are used for the parameter estimation of these 4 yuan of battle arrays again, then parameter to be estimated is L+3 in (6) formula; With L+2 estimates of parameters substitution (6) formula that step 5) has been tried to achieve, then (6) formula only remains a parameter to be estimated, and promptly newly adds the channel phases mismatching of array element, can adopt 1 dimension search to try to achieve its estimated value.

7) the triangle battle array except selecting in the step 1), the channel phases mismatch estimated value of other M-3 array element all can adopt the method in the step 6) to try to achieve.

8) all L folk prescription position echoes are used for the parameter estimation of whole M element array, with L+M-1 parameter (L the folk prescription position echo angle of arrival and M-1 the channel phases mismatching) estimated value of having tried to achieve is initial value, adopts the suboptimization method to try to achieve the more accurate estimated value of these parameters.

Under the little situation of parameter initial value predictor error, suboptimization gained result also is a global optimum, and metered amount is more much smaller than global optimization.Initial value in the above special processing is estimated and has in fact been taken most of computing time, but wherein only relates to the global optimization of 5 dimension search at most, thereby can requirement of real time.More than 8 typical way that step is special processing, can simplify as the case may be in actual applications, enrich, adjust or improve, its basic thought is to utilize the dimension that the processing of triangle battle array is reduced global optimization, and estimate by initial value and to adopt the suboptimization method as far as possible, to reach the purpose that reduces multidimensional parameter estimation operand.

Figure 6 shows that the L shaped battle array of M unit, is a more common embodiment of the present invention.Array element 1～M-1 constitutes an even linear array, wherein can mark off a plurality of translation invariant array element idol groups that are used to detect folk prescription position echo.Channel amplitude is proofreaied and correct and can be realized by (3) formula, and key is that channel phases is proofreaied and correct.The triangle battle array that array element 1,2, M are constituted is used for the parameter initial value to be estimated, and can realize that by (6) formula and special processing channel phases proofreaies and correct.

Figure 7 shows that 4 yuan of rectangle battle arrays.In this embodiment, 4 array elements have only constituted a translation invariant array element idol group that can be used for detecting folk prescription position echo, and the triangle battle array that wherein any 3 array elements constitute all can be used for the parameter initial value and estimates.

Figure 8 shows that 4 yuan of T shape battle arrays.In this embodiment, array element 1～3 constitutes 3 yuan of even linear arrays, can only mark off a translation invariant array element idol group that is used to detect folk prescription position echo.Two triangle battle arrays that array element 1,2,4 and array element 2,3,4 constitute all can be used for the parameter initial value and estimate.

Though the channel correcting method that the present invention describes has at first obtained success on high-frequency ground wave radar, in essence, this method also might be applied to other detection system or communication system of having received a large amount of folk prescriptions position signal.

Claims (14)

1. method for correcting passive channels based on non-linear antenna array is characterized in that: aerial array is set to contain the non-rectilinear form of translation invariant array element idol group; Detect folk prescription position marine echo by translation invariant array element idol group; Utilize folk prescription position marine echo estimating channel amplitude mismatch coefficient, realize amplitude correction; Utilize and passed through the folk prescription position marine echo of amplitude correction and known array position information estimating channel phase mismatch coefficient, realize phase correction.

2. a kind of method for correcting passive channels based on non-linear antenna array according to claim 1 is characterized in that: utilize folk prescription position marine echo to pass through ${\hat{g}}_i=\sqrt{\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|Y_i\left(l\right)\right|}^2/\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|Y_1\left(l\right)\right|}^2}$ Estimating channel amplitude mismatch coefficient is realized amplitude correction, wherein

Be the channel amplitude mismatching estimated value of array element i, i=1,2, Λ, M, M are element number of array, Y _i(l) l the folk prescription position echo output that receives for array element i, l=1,2, Λ, L, L are folk prescription position echo number.

3. a kind of method for correcting passive channels based on non-linear antenna array according to claim 1 is characterized in that: utilize folk prescription position marine echo to pass through ${\hat{g}}_i=\sqrt{\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\frac{{\left|Y_i\left(l\right)\right|}^2}{{\left|Y_1\left(l\right)\right|}^2}}$ Estimating channel amplitude mismatch coefficient is realized amplitude correction, wherein Be the channel amplitude mismatching estimated value of array element i, i=1,2, Λ, M, M are element number of array, Y _i(l) l the folk prescription position echo output that receives for array element i, l=1,2, Λ, L, L are folk prescription position echo number.

4. a kind of method for correcting passive channels based on non-linear antenna array according to claim 1 is characterized in that: utilize folk prescription position marine echo to pass through ${\hat{g}}_i=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\left|Y_i\left(l\right)\right|/\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\left|Y_1\left(l\right)\right|$ Estimating channel amplitude mismatch coefficient is realized amplitude correction, wherein Be the channel amplitude mismatching estimated value of array element i, i=1,2, Λ, M, M are element number of array, Y _i(l) l the folk prescription position echo output that receives for array element i, l=1,2, Λ, L, L are folk prescription position echo number.

5. a kind of method for correcting passive channels based on non-linear antenna array according to claim 1 is characterized in that: utilize folk prescription position marine echo to pass through ${\hat{g}}_i=\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\frac{\left|Y_i\left(l\right)\right|}{{|Y}_1\left(l\right)|}$ Estimating channel amplitude mismatch coefficient is realized amplitude correction, wherein Be the channel amplitude mismatching estimated value of array element i, i=1,2, Λ, M, M are element number of array, Y _i(l) l the folk prescription position echo output that receives for array element i, l=1,2, Λ, L, L are folk prescription position echo number.

6. according to described any one the method for correcting passive channels of claim 1 to 5, it is characterized in that: utilize and pass through through the folk prescription position marine echo of amplitude correction and known array position information based on non-linear antenna array $\widehat{\mathrm{\&Psi;}}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-f^{\&prime;}\left(\mathrm{\&Psi;}\right)\left|\right|}^2$ Estimating channel phase mismatch coefficient is realized phase correction, wherein

Ψ＝[θ ₁，θ ₂，Λ，θ _L，φ ₂，φ ₃，Λ，φ _M] ^T

$Y=\left[\begin{array}{c}{Y}_2\\ Y_3\\ M\\ Y_M\end{array}\right]$

Y _i＝[Y _i(1)，Y _i(2)，Λ，Y _i(L)] ^T

$f^{\&prime;}\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}{f}_2^{\&prime;}\left(\mathrm{\&Psi;}\right)\\ f_3^{\&prime;}\left(\mathrm{\&Psi;}\right)\\ M\\ f_M^{\&prime;}\left(\mathrm{\&Psi;}\right)\end{array}\right]$

$f_i^{\&prime;}\left(\mathrm{\&Psi;}\right)={[Y_1\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},Y_1\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},Y_1\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_I]}]}^T$

θ _lBe the angle of arrival of l folk prescription position echo, φ _iBe the channel phases mismatching of array element i, (x _i, y _i) be the element position coordinate, array element 1 is true origin, i.e. (x ₁, y ₁)=(0,0), λ is the echoed signal wavelength,

Estimated value for Ψ.

7. according to described any one the method for correcting passive channels of claim 1 to 5, it is characterized in that: utilize and pass through through the folk prescription position marine echo of amplitude correction and known array position information based on non-linear antenna array $\widehat{\mathrm{\&Psi;}}=\arg \underset{\mathrm{\&Psi;}}{\min }{\left|\right|Y-\tilde{f}\left(\mathrm{\&Psi;}\right)\left|\right|}^2$ Estimating channel phase mismatch coefficient is realized phase correction, wherein

Ψ＝[θ ₁，θ ₂，Λ，θ _L，φ ₂，φ ₃，Λ，φ _M] ^T

$Y=\left[\begin{array}{c}{Y}_2\\ Y_3\\ M\\ Y_M\end{array}\right]$

Y _i＝[Y _i(1)，Y _i(2)，Λ，Y _i(L)] ^T

$\tilde{f}\left(\mathrm{\&Psi;}\right)=\left[\begin{array}{c}\widetilde{f_2}\left(\mathrm{\&Psi;}\right)\\ \widetilde{f_3}\left(\mathrm{\&Psi;}\right)\\ M\\ \widetilde{f_M}\left(\mathrm{\&Psi;}\right)\end{array}\right]$

$\widetilde{f_i}\left(\mathrm{\&Psi;}\right)={[\hat{A}\left(1\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)+{\mathrm{\&phi;}}_i]},\hat{A}\left(2\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_2+y_i\cos {\mathrm{\&theta;}}_2)+{\mathrm{\&phi;}}_i]},\mathrm{\&Lambda;},\hat{A}\left(L\right)e^{j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_L+y_i\cos {\mathrm{\&theta;}}_L)+{\mathrm{\&phi;}}_I]}]}^T$

$\hat{A}\left(l\right)=\frac{1}{M}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{Y}_i\left(l\right)e^{-j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)+{\mathrm{\&phi;}}_i]}=\frac{1}{M}\{\underset{i=2}{\overset{M}{\mathrm{\&Sigma;}}}{Y}_i\left(l\right)e^{-j[\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_i\sin {\mathrm{\&theta;}}_l+y_i\cos {\mathrm{\&theta;}}_l)+{\mathrm{\&phi;}}_i]}+Y_1\left(l\right)\}$

θ _lBe the angle of arrival of l folk prescription position echo, φ _iBe the channel phases mismatching of array element i, (x _i, y _i) be the element position coordinate, array element 1 is true origin, i.e. (x ₁, y ₁)=(0,0), λ is the echoed signal wavelength, Estimated value for Ψ.

8. according to claim 6 or 7 described a kind of method for correcting passive channels based on non-linear antenna array, it is characterized in that: from all array elements, select 3 array elements and constitute a triangle battle array, this triangle battle array is reduced the processing of global optimization's dimension, and estimate and local optimization method estimating channel phase mismatch coefficient by initial value.

9. a kind of method for correcting passive channels based on non-linear antenna array according to claim 8 is characterized in that: 1) select earlier 3 array elements and constitute a triangle battle array from all array elements, select 3 folk prescription position echoes again and be used for parameter estimation; 2) be the benchmark passage with the arbitrary array element in this triangle battle array, adopt global optimization's method to estimate the phase mismatch coefficient of other two passages and the angle of arrival of 3 folk prescription position echoes; 3) increase the parameter estimation that a folk prescription position echo is used for the triangle battle array, draw the angle of arrival of this folk prescription position echo; 4) set by step 3) draw the angle of arrival of other folk prescription position echo; 5) triangle battle array and another array element are combined into 4 yuan of battle arrays, again all folk prescription position echoes are used for the parameter estimation of these 4 yuan of battle arrays, newly added the channel phases mismatching of array element; 6) set by step 5) draw the channel phases mismatching of other array element.

10. a kind of method for correcting passive channels according to claim 9 based on non-linear antenna array, it is characterized in that: after the process step 4), the parameter estimation that all folk prescription position echoes is used for the triangle battle array, with the folk prescription position echo angle of arrival of having tried to achieve and the estimated value of channel phases mismatching is initial value, adopt the suboptimization method to try to achieve the more accurate estimated value of these parameters, carry out later step then.

11. a kind of method for correcting passive channels according to claim 9 based on non-linear antenna array, it is characterized in that: after the process step 6), the parameter estimation that all folk prescription position echoes is used for whole array, with the folk prescription position echo angle of arrival of having tried to achieve and the estimated value of channel phases mismatching is initial value, adopts the suboptimization method to try to achieve the more accurate estimated value of these parameters.

12. a kind of method for correcting passive channels based on non-linear antenna array according to claim 9 is characterized in that: for the L shaped battle array of M unit, array element 1,2, M are used for parameter estimation as selected triangle battle array in the step 1), realize the channel phases correction.

13. a kind of method for correcting passive channels according to claim 9 based on non-linear antenna array, it is characterized in that: for 4 yuan of T shape battle arrays, array element 1,2,4 or array element 2,3,4 are used for parameter estimation as selected triangle battle array in the step 1), realize the channel phases correction.

14. a kind of method for correcting passive channels according to claim 9 based on non-linear antenna array, it is characterized in that: for 4 yuan of rectangle battle arrays, wherein any 3 array elements are used for parameter estimation as selected triangle battle array in the step 1), realize the channel phases correction.

Images