CN103217670A - Outer radiation source weak signal detection method based on PCA (principal component analysis) - Google Patents

Abstract

The invention provides an outer radiation source weak signal detection method based on PCA (principal component analysis). The method comprises the following steps of solving an autocorrelation matrix R=E{x(t)x<H>(t)} of a sampling sequence x(t); carrying out SVD (singular value decomposition) on the autocorrelation matrix R, to obtain R=Q Lambda Q<H>, utilizing the decomposed diagonal matrix Lambda and unitary matrix Q to carry out linear conversion on the sampling sequence x(t) to obtain a strength-decreased sequence y(t) of received signals; carrying out the two-dimensional correlation of time frequency distribution on the third path to the Mth path of signals in the received signal y(t) to obtain the corresponding correlation value, namely the absolute value of psi i (tao, f); when the signal-to-noise ratio of one certain path is greater than or equal to a threshold value, determining the time delay and Doppler frequency corresponding to a peak point as the time delay and Doppler frequency displacement of one target; and when the signal-to-noise ratio of one path is smaller than the threshold value, judging that the signal of the path has no target. The method has the advantage that a weak target signal can be extracted under the conditions of strong same-frequency interference and low signal-to-noise ratio.

Description

A kind of external radiation source method for detecting weak signals based on PCA

Technical field

The invention belongs to signal processing technology, be specifically related to the Detection of Weak Signals technology.

Background technology

Two multistatic radar system based on the external radiation source can utilize geographic position known commercial television signal, broadcast singal, signal of communication, satellite-signal, and cellular base station etc. is as irradiation source, by detection, realize the detection of target is followed the tracks of with the location to the echoed signal of aerial sports target reflection.Because external radiation source radar self do not transmit, whole radar has disguised strong, advantage such as antijamming capability is strong and viability is strong, thereby is subjected to the extensive concern of various countries.

The principle signal of system as shown in Figure 1, with the broadcast singal is example, radar comes from the direct wave of broadcast emission tower itself, the echo of moving target, the direct wave of other common-frequency broadcasting launching tower transmission and the echo of other object reflection to including of received signal, such as the still life echo of various buildings reflections.Therefore, the target echo signal in the radar received signal is extremely faint, and is submerged in the various interference.

It is faced with many challenges in practical application, wherein, how the external radiation source radar system that the most direct problem just is based on civil signal more effectively realizes detection and the tracking to weak target, mainly be presented as in the received signal of radiation source radar outside, target echo signal is submerged in clutters such as direct wave and multipath signal and the receiver thermonoise, has increased the difficulty of target detection.

The tradition externally detection method of radiant source target signal is to receive the direct wave signal with reference channel, and main antenna receives multiple signals, comprising direct wave signal, target echo signal and noise and interference.Processing procedure as shown in Figure 2, at first adopt the NLMS algorithm that direct wave is carried out adaptive cancellation with main channel signal and reference channel signal, by airspace filter co-channel interference is suppressed again, adopt time-frequency two-dimensional associated extraction target time delay, Doppler frequency at last, thus estimating target DOA (target arrival angle) parameter.Yet this kind method will lose efficacy under the very faint situation of echo signal, and simultaneously, because co-channel interference also may be present in the same wave beam, this will cause extracting target time delay, Doppler frequency parameter, thereby be unable to estimate the DOA of target.

Summary of the invention

Technical matters to be solved by this invention is to provide a kind of the existence effectively to realize the method that the weak target echoed signal detects under the co-channel interference situation.

The present invention solves the problems of the technologies described above the technical scheme that is adopted to be, a kind of external radiation source method for detecting weak signals based on main component analysis PCA may further comprise the steps:

Step 1: the take over party carries out bandpass sampling to the pending main channel array digital modulation signals that receives and obtains sample sequence x (t), and t represents sampling instant;

Step 2: autocorrelation matrix R=E{x (t) x that asks sample sequence x (t) ^H(t) }, E{} represents to average;

Step 3: autocorrelation matrix R is carried out svd SVD obtain R=Q Λ Q ^H, utilize to decompose the diagonal matrix Λ that obtains and unitary matrix Q to sample sequence x (t) carry out linear transformation obtain received signal from strong to weak arrangement y (t),

Y (t)=[y wherein ₁(t) y ₂(t) ... y _M(t)] ^T, M is a receiving antenna array element number;

Step 4: with among the received signal y (t) the 3 the road to M road signal respectively with the 1 road signal y ₁(t) do the time-frequency distributions two-dimensional correlation and obtain corresponding correlation | Ψ _i(τ, f) |,

I=3 ..., M, L are bandpass sampling length, τ is time delay, f Doppler shift;

Step 5: respectively at the correlation of 3 to M each road signal correspondence | Ψ _i(τ, f) | in search plain peak point, with the time delay of each road signal peak value point correspondence and Doppler shift as a useful signal, calculate each road signal to noise ratio (S/N ratio), as a certain road signal to noise ratio (S/N ratio) during more than or equal to threshold value, time delay and the Doppler shift that the time delay and the Doppler of its peak point correspondence is defined as a target frequently then, otherwise judge in this road signal and do not have target.

The invention has the beneficial effects as follows that the present invention can be implemented under strong co-channel interference and the low signal-to-noise ratio condition and extract weak target signal.The present invention is when the relevant coupling of time-frequency two-dimensional is extracted time delay, Doppler shift in addition, do not need to adopt the reference antenna received signal more as a reference, thereby played the effect of economize on hardware cost, the present invention need not adopt steps such as adaptive cancellation in addition, and the integral operation amount also has bigger reducing than classic method.

Description of drawings

Fig. 1: based on the system principle synoptic diagram of broadcasting external radiation source location.

Fig. 2: classic method signal processing flow figure.

Fig. 3: signal processing flow figure of the present invention.

Fig. 4: target decay-60dB, under the situation of interference attenuation-10dB, the time-frequency two-dimensional correlogram of classic method echo signal.

Fig. 5: target decay-60dB, under the situation of interference attenuation-10dB, the time-frequency two-dimensional correlogram of echo signal of the present invention.

Fig. 6: target decay-60dB, under the situation of interference attenuation-10dB, the time-frequency two-dimensional correlogram time delay of classic method echo signal dimension.

Fig. 7: target decay-60dB, under the situation of interference attenuation-10dB, the time-frequency two-dimensional correlogram time delay of echo signal of the present invention dimension.

Fig. 8: target decay-60dB, under the situation of interference attenuation-10dB, the time-frequency two-dimensional correlogram frequency dimension of classic method echo signal.

Fig. 9: target decay-60dB, under the situation of interference attenuation-10dB, the time-frequency two-dimensional correlogram frequency dimension of echo signal of the present invention.

Embodiment

Processing signals model of the present invention is as follows:

The digital modulation signals y (t) that array antenna is received carries out modeling, is primarily aimed at bandpass signal and handles, and establishes the continuous signal after ovennodulation of frequency modulation broadcasting of s (t) expression some broadcasting television towers emission, then

Be the echoed signal of target, wherein M is a receiving antenna array element number, a ₁..., a _MThe decay of expression target echo, τ ₁... τ _MThe time delay of expression target, f ₁..., f _MThe expression target Doppler shift, then by array antenna received to signal be $y\left(t\right)=s\left(t\right)+a_{1}s(t-{\mathrm{\&tau;}}_1)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA00002987011200014.png,HDA00002987011200021.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}{f}_{1}t}+\&CenterDot;\&CenterDot;\&CenterDot;+a_{M}s(t-{\mathrm{\&tau;}}_M)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA00002987011200014.png,HDA00002987011200021.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}{f}_{M}t}+z\left(t\right)+n\left(t\right),$ Z (t) expression co-channel interference signal, the thermonoise of n (t) expression receiver, then the work that will do of the present invention is time delay, the Doppler frequency parameter of extracting target from the signal that receives, this is for can estimating target DOA in subsequent treatment, thereby final the realization prepared to the location and the tracking of target.

Implementation step is as shown in Figure 3:

Step 1: the take over party carries out bandpass sampling to the pending main channel array digital modulation signals that receives, and obtains sample sequence x (t)=[x ₁(t) x ₂(t) ... x _M(t)] ^T, wherein M represents array number, and t represents sampling instant, and sampling length is L, T representing matrix transposition.X (t) includes the information of target time delay, Doppler frequency and DOA.

Step 2: autocorrelation matrix R=E{x (t) x that asks sample sequence x (t) ^H(t) }, E{} represents to average.Can suppose that sample sequence is a stochastic process stably, asks statistical average can adopt seeking time on average to replace, then like this in sampling length L $R=\frac{1}{L}\underset{t}{\mathrm{\&Sigma;}}x\left(t\right)x^H\left(t\right),$ H represents conjugate transpose.

Step 3: autocorrelation matrix R is carried out svd SVD obtain R=Q Λ Q ^H, utilize and decompose diagonal matrix Λ and the unitary matrix Q that obtains, x (t) is carried out linear transformation obtain

Y (t)=[y wherein ₁(t) y ₂(t) ... y _M(t)] ^T, y (t) be with received signal from strong to weak arrangement, y ₁(t) be the strongest direct wave signal, i.e. reference signal.

Step 4: adopt the time-frequency two-dimensional coherent matched filter to extract the time delay and the Doppler shift of target: because the direct wave signal is more intense, so after carrying out principal component analysis resulting first via signal y ₁(t) can be used as direct wave signal, y ₁(t)=Aexp{j[2 π f _c(t-D ₁)+θ (t-D ₁)]+n (t), wherein n (t) is a distracter, f _cBe the carrier frequency of broadcast singal, D ₁For broadcast singal receives institute's elapsed time, θ (t-D from being transmitted into ₁) be the sound signal modulation item.And because jamming power is only second to direct wave power, so the second road signal y ₂(t) can regard the co-channel interference signal as.y _i(t), i ≠ 1,2 is for comprising the signal of target echo parameter, because through principal component analysis, wherein contained interference and direct wave component are smaller, so its expression formula can be write y _i(t)=Bexp{j[2 π (f _c+ f _Dm) (t-D _m)+θ (t-D _m)]+m (t), wherein f _DmBe target Doppler frequency, D _mBe the scattering arrival radar receiving antenna institute elapsed time of broadcast singal from emission, process target, m (t) can regard as and estimate required (τ, f _d) the irrelevant distracter of parameter, to y ₁(t) doing time delay can get:

y ₁(t-n△t)=Aexp{j[2πf _c(t-D ₁-n△t)+θ(t-D ₁-n△t)]}+n((t-n△t))

Time-delay length △ t uses the sampling period, perhaps the integral multiple in sampling period.Use y ₁The conjugation-type and the y of (t-n △ t) _i(t), multiply each other and can get in i ≠ 1,2:

$y_1^{*}(t-\mathrm{n\&Delta;t})y_i\left(t\right)$

$=\mathrm{ABexp}\left\{j\right[2\mathrm{\&pi;}{f}_{\mathrm{dm}}t+2\mathrm{\&pi;}{f}_c({\mathrm{<figure-callout\; id="1"\; label="D"\; filenames="HDA00002987011200014.png,HDA00002987011200021.png"\; state="\{\{state\}\}">D</figure-callout>}}_1+\mathrm{n\&Delta;t}-D_m)-2\mathrm{\&pi;}{f}_d{D}_m+\mathrm{\&theta;}(t-D_m)-\mathrm{\&theta;}(t-D_1-\mathrm{n\&Delta;t})\left]\right\}+N\left(t\right)$

$=\mathrm{ABexp}\left\{j\right[2\mathrm{\&pi;}{f}_{\mathrm{dm}}t+\mathrm{\&theta;}(t-D_m)-\mathrm{\&theta;}(t-D_1-\mathrm{n\&Delta;t})+{\mathrm{\&theta;}}_1\left]\right\}+N\left(t\right)$

* represent conjugate matrices, can see,, work as D by adjusting the time delay of direct wave signal ₂=D ₁During+n △ t, two phase places in the following formula in the exponential term will disappear mutually, thereby following formula will become Doppler shift item and noise item, that is:

${\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="HDA00002987011200014.png,HDA00002987011200021.png"\; state="\{\{state\}\}">y</figure-callout>}}_1^{*}(t-\mathrm{n\&Delta;t})y_i\left(t\right)=\mathrm{ABexp}\left\{j\right[2\mathrm{\&pi;}{f}_{\mathrm{dm}}t+{\mathrm{\&theta;}}_1\left]\right\}+N\left(t\right)$

Following formula is done the FFT computing, can obtain the time delay and the Doppler shift of target.

The M road signal that step 3 is obtained the 3 the road to M road signal and the 1 road signal y ₁(t) do time-frequency two-dimensional and be correlated with, I=3 ..., M, (τ f) is the time delay and the Doppler frequency of i road signal correspondence.When having target, on the i road | Ψ _i(τ, f) | can produce a peak point | Ψ _i(τ _d, f _d) |, search peak point after, with (τ _d, f _d) as a useful signal, calculate signal to noise ratio (S/N ratio),

Represent to remove on the signal of i road | Ψ _i(τ _d, f _d) | outer, all | Ψ _i(there are corresponding (τ in τ, f) | average if its value, is then represented target greater than the threshold value 13dB that sets in advance _d, f _d) be the time delay and the Doppler frequency of a target; If its value is then represented not have target on the signal of i road less than this threshold value.

In order to verify effect of the present invention, adopt respectively two sections independently at random sound signal as modulation signal in order to produce fm broadcast signal, as direct wave signal and co-channel interference in the simulating scenes, carrier frequency is 80kHz with this, maximum frequency deviation is ± 75kHz that sample frequency is 200kHz.Emulation platform operating system is WIN7, adopts MATLAB to compile.

Table 1 provides each simulation parameter of mixed signal, and relatively under different condition classic method and the present invention the detection performance to echo signal of method is proposed, also compared the operand of classic method and institute of the present invention extracting method at last.

Table 1 simulate signal parameter

Signal	Time delay (μ s)	Power attenuation (dB)	The angle of arrival (°)	Doppler frequency (Hz)
					Main platform direct wave s1	0	0	-50	0
Main platform multipath s2	75	-10	-20	0
					Main platform multipath s3	40	-20	30	0
Lin Tai disturbs s4	0	-10～-30	8	0
					Target echo s5	439.12	-60～-85	5	119.37

The receiving array array number is 12.Additional noise is-50dB to be higher than target echo signal 10dB～35dB.The time-frequency two-dimensional coherent accumulation time is 1s.Digital beam forms DBF and adopts Chebyshev's weighting, and beam area is from-45 ° to 45 °, and beam separation is 10 °.Constant false alarm detector adopts the unit to select big constant false alarm detector, and it is 4.5 that CFAR thresholding coefficient is set.To classic method performance such as table 2, the present invention proposes method performance such as table 3:

Table 2 traditional technique in measuring performance table

Table 3 institute of the present invention extracting method detects the performance table

If target time delay, the relevant peak-peak of Doppler frequency place time-frequency two-dimensional are A, except that handle by constant false alarm rate CFAR can detected target peak, the average out to B that time-frequency two-dimensional correlogram upper base is made an uproar, the peak-peak discrimination is defined as 10log (A/B) in then upward showing.

Can illustrate that by the emulated data that table 2 and table 4 provided method proposed by the invention is having strong co-channel interference, echo signal and interference are in same wave beam, and can both effectively detect echo signal under the very faint situation of echo signal, and extract time delay, the Doppler frequency parameter of target.

Target decay-60dB, under the condition of interference attenuation-10dB, classic method and echo signal time-frequency two-dimensional correlogram of the present invention are respectively Fig. 4, Fig. 5; Classic method and echo signal time-frequency two-dimensional correlogram time delay of the present invention dimension are respectively Fig. 6, Fig. 7; Classic method and echo signal time-frequency two-dimensional correlogram frequency dimension of the present invention are respectively Fig. 8, Fig. 9.Can be respectively from, classic method and time-frequency two-dimensional correlogram time delay peacekeeping Doppler of the present invention tie up and observe the contrast target, can find by contrast, the end of classic method time-frequency two-dimensional correlogram, make an uproar very high, the peak-peak discrimination of target is lower, and to make an uproar in the end of time-frequency two-dimensional correlogram of the present invention less, and target peak-peak discrimination is higher, detects effective.

Because it is relevant that classic method and the present invention will use time-frequency two-dimensional, so relatively calculated amount only need compare the relevant preceding operand of time-frequency two-dimensional, establishing the receiving array element number of array is M, and umber of beats is N soon, and the wave beam number is L, and then comparative result is shown in table 4, table 5:

Table 4 classic method computing scale

Table 5 institute of the present invention extracting method computing scale

By above calculating contrast as can be seen, the present invention has a certain amount of reducing than the operand of classic method.And this method does not need to use reference antenna collimation arrived wave signal, has saved the cost of hardware to a certain extent.

Claims (5)

1. external radiation source method for detecting weak signals based on PCA may further comprise the steps:

Step 1: the take over party carries out bandpass sampling to the pending main channel array digital modulation signals that receives and obtains sample sequence x (t), and t represents sampling instant;

Step 2: the autocorrelation matrix R that asks sample sequence x (t);

Step 3: autocorrelation matrix R is carried out svd SVD obtain R=Q Λ Q ^H, utilize to decompose the diagonal matrix Λ that obtains and unitary matrix Q to sample sequence x (t) carry out linear transformation obtain received signal from strong to weak arrangement y (t),

Y (t)=[y wherein ₁(t) y ₂(t) ... y _M(t)] ^T, M is a receiving antenna array element number;

Step 4: with among the received signal y (t) the 3 the road to M road signal respectively with the 1 road signal y ₁(t) do the time-frequency distributions two-dimensional correlation and obtain corresponding correlation | Ψ _i(τ, f) |,

I=3 ..., M, L are bandpass sampling length, τ is time delay, f Doppler shift;

Step 5: respectively in the two-dimensional correlation value of 3 to M each road signal correspondence | Ψ _i(τ, f) | in search plain peak point, with the time delay of each road signal peak value point correspondence and Doppler shift as a useful signal, calculate each road signal to noise ratio (S/N ratio), as a certain road signal to noise ratio (S/N ratio) during more than or equal to threshold value, time delay and the Doppler shift that the time delay and the Doppler of its peak point correspondence is defined as a target frequently then, otherwise judge in this road signal and do not have target.

2. a kind of according to claim 1 external radiation source method for detecting weak signals based on PCA is characterized in that, autocorrelation matrix R=E{x (t) x ^H(t) }, E{} represents to average.

3. a kind of according to claim 1 external radiation source method for detecting weak signals based on PCA is characterized in that, $R=\frac{1}{L}\underset{t}{\mathrm{\&Sigma;}}x\left(t\right)x^H\left(t\right),$ H represents conjugate transpose.

4. as a kind of external radiation source method for detecting weak signals as described in the claim 1,2 or 3, it is characterized in that the method for calculating each road signal to noise ratio (S/N ratio) in the step 5 is based on PCA:

| Ψ _i(τ _d, f _d) | be the two-dimensional correlation value on the i road | Ψ _i(τ, f) | peak point,

Represent to remove on the signal of i road | Ψ _i(τ _d, f _d) | outer, all | Ψ _i(τ, f) | average.

5. as a kind of external radiation source method for detecting weak signals as described in the claim 4, it is characterized in that described threshold value is 13dB based on main component analysis PCA.

Images