CN104808219A - Novel space-time joint anti-interference method - Google Patents

Abstract

The invention relates to a space-time array antenna anti-interference method in satellite navigation equipment, in particular to a novel space-time joint anti-interference method. The space-time joint anti-interference method comprises the following steps of restraining narrow-band interference in a navigation signal direction through the improved frequency domain narrow-band interference filter-out technology and achieving iteration solution of space-time array weight in combination with minimum variable step size mean square error to reduce the algorithm complexity of the method. The space-time array antenna anti-interference method in the satellite navigation equipment can restrain interference in the navigation signal direction, is low in algorithm complexity, simple in achievement and particularly suitable for application of future mini-sized high-dynamic anti-interference terminals.

Description

A kind of novel space-time joint anti-interference method

Technical field

The present invention relates to a kind of space-time array Antenna Anti-jamming method on satellite navigation, especially the arrowband interference based on first utilizing on frequency domain filtering method filtering navigation signal direction is related to, recycling Variable Step Algorithm asks for the satellite navigation signals anti-interference method of space-time two-dimensional weight vector, is specially a kind of novel space-time joint anti-interference method.

Background technology

Along with the complexity day by day of satellite navigation application electromagnetic space environment, the difficulty day by day that navigation terminal precision positioning also becomes, various navigation terminal suppresses perturbation technique to produce thereupon.Traditional receiver Anti-Jamming Technique has spatial domain filtering technique, temporal filtering technique and frequency domain filtering technology, and these Anti-Jamming Techniques all exist certain defect in performance.Space-time adaptive treatment technology is not when increasing array element, and suppressible interference number increases greatly, and has the ability processing interference in broadband, thus makes the antijamming capability of receiver have the raising of matter.

At present, the implementation method that when key point of Anti-Jamming Technique is sky time empty, weight vector solves.The general flow figure that weight vector solves is as shown in figure (1).Each array element of array antenna increases the delay tap of identical number, thus forms space-time two-dimensional Combined Treatment structure, from the angle of single array element, form time domain FIR structure; From node at the same time, form airspace filter structure.Single-frequency, arrowband, wideband interferer signal can be suppressed by space-time joint processing.

Space-time two-dimensional adaptive anti-jamming processing technology uses linear constraint minimal variance (LCMV) criterion usually, principle is well below Noise and Interference power according to satellite-signal power, thus retrain to adjust weights to the spatial domain in space-time joint processing structure and time domain, make output signal variance minimum, greatly weaken interfering energy.Response that minimum variance is undistorted (MVDR) criterion is a kind of special case of LCMV criterion.When satellite-signal is empty steering vector S oneself when knowing, response g=1 when constraint satellite-signal is empty on direction, namely output signal power minimum: $W_{\mathrm{opt}}=\arg \min W^H{R}_x^{-1}\mathrm{W\; s}.t.W^{H}S=1---\left(1\right),$ Best initial weights closed solution based on linearly constrained minimum variance can be expressed as: in formula, W=[ω ₁₁, ω ₁₂..., ω _mP] ^tfor MP × 1 dimensional vector, M is the array number of array antenna, and P is delay cell number; R _x=E (XX ^h) for MP × MP dimension Received signal strength covariance matrix; X=(x ₁₁, x ₁₂..., x _mP) ^tfor MP × l signal vector; S is the constrained vector of MP × l dimension.When direct utilization (2) formula carries out weight computing, matrix inversion formula can be introduced, work as M, when P dimension is larger, good anti-jamming effectiveness can be obtained, and increase anti-interference number.But its operand increases, and large operand can cause system complexity to increase, and causes the real-time of system to be deteriorated.Given this, the general recursive algorithm that adopts solves weights.

If the signal x that array element 1 receives ₁n () is wanted signal, i.e. d (n)=x ₁(n).Then the output of wave filter is: y (n)=W ' ^hx ', e (n)=W ₁ ^hd (n)-y (n) (3), wherein W '=(ω ₂₁, ω ₂₂..., ω _mP) ^t, X '=(x ₂₁, x ₂₂..., x _mP) ^t, W ₁=[1,1 ..., 1] ^t, the recursion formula utilizing steepest descent method can derive W is: W _n+1'=W _n'+2 μ X _ne (n) ^*(4), in formula, μ is a constant, is called converging factor.Upgrade according to the continuous iteration of formula (4) and just can obtain best initial weights W _optestimated value.Weights after stable export: W=[W ₁ ^t, W _opt ^t] ^t(5), from the above, the iterative algorithm that step-length is fixed can avoid the inversion process of matrix, thus calculated amount is significantly reduced.But the iterative algorithm of fixed step size can run into speed of convergence and the incompatible problem of steady-state error solving array weight value, and this can cause algorithm to decline to some extent in antijamming capability.

When the direction of undesired signal is identical with navigation signal time, very dark zero can be produced its arrival bearing and fall into, cause navigation signal seriously to be weakened.The conventional STAP algorithm in conjunction with FIR filter when IIR notch filter and sky, can solve the problem of this respect at present.But due to the restriction of FIR filter tap, the frequency resolution of adaptive processor is very low, simultaneously because the frequency spectrum of arrowband interference and statistical property are unknown, therefore the method is not suitable for catching fast-changing signal.

Summary of the invention

The present invention disturb with the equidirectional arrowband of navigation signal to effectively suppress, and Anti-interference algorithm computation process complexity when avoiding sky, is difficult to the shortcoming of Project Realization, provides a kind of novel space-time joint anti-interference method.

The present invention adopts following technical scheme to realize: a kind of novel space-time joint anti-interference method, comprises the following steps:

Sampling in each array element each road first in pair array Received signal strength, then carries out radio frequency down-conversion conversion and digitizing, obtain digitized intermediate frequency signal X _i, X _i=(x _i1, x _i2..., x _iP), i=1,2,3 ... M;

By digitized intermediate frequency signal X _ibe input in LMS frequency domain filter, by the arrowband interference filtering of the Frequency Domain LMS iterative algorithm in wave filter by navigation signal direction, obtain trap signal Z _i(k)=(z _i1, z _i2..., z _iP) (i=1,2, M), the frequency domain weights coefficient iterative formula of Frequency Domain LMS iterative algorithm is: U (k+1)=α U (k)+2uE (k) * F (k), wherein, U (k+1), U (k) is respectively the filter coefficient vector in k+1 and k moment; α is weight; μ is converging factor; F (k) is digitized intermediate frequency signal X _ifourier transform; E (k)=F (k)-U (k) * F (k) is error signal; * be vector dot, then Z _i(k)=f ^-1(U (K) * F (K)), wherein f ^-1for inverse Fourier transform;

By trap signal Z _ik () is sent in space-time array signal processing filter, eliminate arrowband broadband interference further, and the step-length of this algorithm filter more new formula is: $\left\{\begin{array}{c}\mathrm{\μ}\left(k\right)=\mathrm{\α}\tanh \left|\mathrm{\βe}\left(k\right)e(k-1)\right|\\ \mathrm{\μ}\left(k\right)=\mathrm{\α}(1-\frac{1}{(100\×e\left(k\right)e(k-1))}+\mathrm{\μ}(k-1)\end{array}\right.,$ The step that change step length least mean square ERROR ALGORITHM iteration goes out best initial weights is: the signal that the 2nd to M array element of Array Signal Processing wave filter receives can be expressed as: Z (k)=(Z ₂, Z ₃... Z _m) ^h, can be expressed as after space-time array signal transacting: Y (k)=W ' ^hz (k), the signal receive first array element and Received signal strength Y (k) make the error signal that difference obtains array element: e (k)=W ₁ ^hz ₁(k)-Y (k); The recursion formula utilizing steepest descent method can derive W is: W _k+1'=W _k'+2 μ (k) Z (k) e (k) ^*, then the optimal weighting value after stablizing is W _opt, now space-time array signal processing filter exports pure navigation signal $y^{\′\left(k\right)}=e\left(k\right)={W_1}^H{Z}_1\left(k\right)-$ $W_{\mathrm{opt}}^{H}Z\left(k\right).$

The pre-service of improved LMS frequency domain interference can under ensureing that navigation signal does not have shown prerequisite as far as possible, and filtering and the equidirectional arrowband of navigation signal disturb as much as possible.The signal of pre-filtering is sent into space-time array signal processing unit, under the prerequisite ensureing interference free performance, the complexity of algorithm can be reduced by the least mean square algorithm improved.This technology is more suitable for the application of following miniaturization high dynamic receiver.

Accompanying drawing explanation

Fig. 1 is Frequency Domain LMS Suppression of narrow band interference schematic diagram.

Fig. 2 is traditional anti-interference schematic diagram of space-time adaptive.

Fig. 3 is the anti-interference schematic diagram of space-time adaptive of the present invention.

Embodiment

Sampling in each array element each road first in pair array Received signal strength, then carries out radio frequency down-conversion conversion and digitizing, obtain digitized intermediate frequency signal X _i, X _i=(x _i1, x _i2..., x _iP), i=1,2,3 ... M;

By digitized intermediate frequency signal X _ibe input in LMS frequency domain filter, by the arrowband interference filtering of the Frequency Domain LMS iterative algorithm in wave filter by navigation signal direction, obtain trap signal Z _i(k), the frequency domain weights coefficient iterative formula of Frequency Domain LMS iterative algorithm is: U (k+1)=α U (k)+2 μ E (k) * F (k), wherein, U (k+1), U (k) is respectively the filter coefficient vector in k+1 and k moment; α is weight; μ is converging factor; F (k) is the Fourier transform of digitized intermediate frequency signal Xi; E (k)=F (k)-U (k) * F (k) is error signal; * be vector dot, then Z _i(k)=f ^-1(U (K) * F (K)), wherein f ^-1for inverse Fourier transform; Weights coefficient vector U (k+1) can be expressed as: U (k+1)=aU (k)+2 μ [1-U (k)] * F (k) * F (k), and the mathematical expectation of weights coefficient vector U (k+1) can be expressed as: E{U (k+1) }=aE{U (k) }+2 μ [E{ ‖ F (k) ‖ ²-E{U (k) ‖ F (k) ‖ ²], make E{ ‖ F (k) ‖ here ²}=σ ², wherein σ ²for digitized intermediate frequency signal X _ipower, then E{U (k+1) }=[a-2 μ σ ²] E{U (k)+2 μ σ ², therefore, U (∞) can be expressed as: $U(\∞)={\lim }_{k\→\∞}U\left(k\right)=\frac{2\mathrm{\μ}{\mathrm{\σ}}^2}{1-a+2\mathrm{\μ}{\mathrm{\σ}}^2},$ Here make $B=\frac{1-a}{2\mathrm{\μ}},$ Then $U(\∞)=\frac{{\mathrm{\σ}}^2}{{\mathrm{\σ}}^2+B},$ Therefore σ is worked as ²during > > B, U (∞) ≈ 1, the effect of weights leakage factor a can be found out: under enough large prerequisite is suppressed to high-power signal, the suppression of low-power level signal is reduced greatly, when namely LMS frequency domain filter can filter out interference, too large suppression can not be produced to navigation signal;

By trap signal Z _ik () is sent in space-time array signal processing filter, eliminate arrowband broadband interference further, and the step-length of this algorithm filter more new formula is: $\left\{\begin{array}{c}\mathrm{\μ}\left(k\right)=\mathrm{\α}\tanh \left|\mathrm{\βe}\left(k\right)e(k-1)\right|\\ \mathrm{\μ}\left(k\right)=\mathrm{\α}(1-\frac{1}{(100\×e\left(k\right)e(k-1))}+\mathrm{\μ}(k-1)\end{array}\right.,$ The step that change step length least mean square ERROR ALGORITHM iteration goes out best initial weights is: the signal that the 2nd to M array element of Array Signal Processing wave filter receives can be expressed as: Z (k)=(Z ₂, Z ₃... Z _m) ^h, can be expressed as after space-time array signal transacting: Y (k)=W ' ^hz (k), the signal receive first array element and Received signal strength Y (k) make the error signal that difference obtains array element: e (k)=W ₁ ^hz ₁(k)-Y (k); The recursion formula utilizing steepest descent method can derive W is: W _k+1'=W _k'+2 μ (k) Z (k) e (k) ^*, then the optimal weighting value after stablizing is W _opt, now space-time array signal processing filter exports pure navigation signal this algorithm utilizes the step factor of the autocorrelation estimation of error signal and previous step jointly to determine the iterative formula of step factor, and in the iteration starting stage, because error function autocorrelation value is comparatively large, the value of hyperbolic sine function is also comparatively large, causes speed of convergence faster; After iteration convergence, because the change of signal errors value is less, the change of step factor reduces thereupon, and convergency value tends towards stability, and final wave filter exports pure navigation signal.

Claims (1)

1. a novel space-time joint anti-interference method, is characterized in that comprising the following steps:

Sampling in each array element each road first in pair array Received signal strength, then carries out radio frequency down-conversion conversion and digitizing, obtain digitized intermediate frequency signal X _i,

X _i＝(x _i1，x _i2，...，x _iP)，i＝1，2，3，...M；

By digitized intermediate frequency signal X _ibe input in LMS frequency domain filter, by the arrowband interference filtering of the Frequency Domain LMS iterative algorithm in wave filter by navigation signal direction, obtain trap signal Z _i(k)=(z _i1, z _i2..., z _iP) (i=1,2, M), the frequency domain weights coefficient iterative formula of Frequency Domain LMS iterative algorithm is: U (k+1)=α U (k)+2 μ E (k) * F (k), wherein, U (k+1), U (k) is respectively the filter coefficient vector in k+1 and k moment; α is weights leakage factor; μ is converging factor; F (k) is digitized intermediate frequency signal X _ifourier transform be; E (k)=F (k)-U (k) F (k) is error signal; * be matrix dot product, then Z _i(k)=f ^-1(U (K) * F (K)), wherein f ^-1for inverse Fourier transform;

By trap signal Z _ik () is sent in space-time array signal processing filter, eliminate arrowband broadband interference further, and the step-length of this algorithm filter more new formula is: $\left\{\begin{array}{c}\mathrm{\μ}\left(k\right)=\mathrm{\α}\tanh \left|\mathrm{\βe}\left(k\right)e(k-1)\right|\\ \mathrm{\μ}\left(k\right)=\mathrm{\α}(1-\frac{1}{(100\×e\left(k\right)e(k-1))})+\mathrm{\μ}(k-1)\end{array}\right.,$ The step that change step length least mean square ERROR ALGORITHM iteration goes out best initial weights is: the signal that the 2nd to M array element of Array Signal Processing wave filter receives can be expressed as: Z (k)=(Z ₂, Z ₃... Z _m) ^h, can be expressed as after space-time array signal transacting: Y (k)=W ' ^hz (k), the signal receive first array element and Received signal strength Y (k) make the error signal that difference obtains array element: e (k)=W ₁ ^hz ₁(k)-Y (k); The recursion formula utilizing steepest descent method can derive W is: W _k+1'=W _k'+2 μ (k) Z (k) e (k) ^*, then the optimal weighting value after stablizing is W _opt, now space-time array signal processing filter exports pure navigation signal $y^{\′\left(k\right)}=e\left(k\right)={W_1}^H{Z}_1\left(k\right)-$ $W_{\mathrm{opt}}^{H}Z\left(k\right).$