CN106199588A - The multistation Radar Signal Fusion detection method quantified based on Pasteur's distance - Google Patents

Abstract

The invention belongs to Radar Technology field, disclose a kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance, it is possible to reduce the communication bandwidth between radar and signal fused center in the detection of multistation Radar Signal Fusion.Including: arranging netted radar system, described netted radar system includes a signal fused inspection center and N number of radar station；Determine the quantization digit b of each radar station echo-signal to receiving, and obtain quantized interval number M=2 according to quantization digit b^b；Its echo-signal received is quantified by each radar station, determines the quantized interval belonging to the echo-signal after quantization, and transmits label corresponding for this quantized interval to signal fused inspection center；Arranging desired false-alarm probability, target is detected by described signal fused inspection center according to the label that described desired false-alarm probability is corresponding with the respective quantized interval that N number of radar station sends.

Description

The multistation Radar Signal Fusion detection method quantified based on Pasteur's distance

Technical field

The present invention relates to Radar Technology field, particularly relate to a kind of multistation Radar Signal Fusion quantified based on Pasteur's distance Detection method, can be used for the communication bandwidth between radar and signal fused center in reduction multistation Radar Signal Fusion detection.

Background technology

Multistation radar fence is made up of a signal fused center and multiple radar station.Compared with single radar, multistation radar Net has the advantages such as bigger investigative range, higher positioning precision, higher survival ability and higher capacity of resisting disturbance.Cause This, the research of multistation radar fence becomes a new study hotspot.In the research of multistation radar fence, the signal of multistation radar Fusion detection method is an important research direction.

For current existing research, the signal fused detection method of multistation radar is broadly divided into three kinds: multistation radar The signal level fusion detection method of flight path level fusing method, the decision stage fusion method of multistation radar and multistation radar.Multistation thunder The flight path level fusing method reached refers to that the flight path of oneself is transmitted to signal fused center by each radar station, and signal fused center is by institute The Track Fusion having radar station becomes a flight path.The decision stage fusion method of multistation radar refers to that each radar station is by respective judgement Result is transmitted to signal fused center, and signal fused center finally judges having of target according to the court verdict of all radars Nothing.The signal level fusion detection method of multistation radar refers to that the echo-signal of oneself is transmitted to fusion center, letter by each radar station Number fusion center judges the presence or absence of target according to all radar station echo-signals.

In detection performance, the decision stage fusion method of multistation radar is better than the fusion method of flight path level, and multistation radar Signal level fusion detection method be better than again decision stage fusion method.At present, the flight path level fusing method of multistation radar has become Ripe, and it has been applied to the engineering system of reality.Next step goal in research is that signal level fusing method is applied to new one The netted radar system in generation.For the signal level fusion detection method of multistation radar, owing to each radar station needs all of Echo-signal is transmitted to signal fused center, it is therefore desirable to big communication bandwidth.But, between radar station and signal fused center Communication bandwidth limited, so required communication band when must study new quantization method to reduce radar echo signal transmission Wide.And while reducing communication bandwidth, it is ensured that the detection performance that signal fused center has had.Return for different radars Ripple signal model, it has been proposed that multiple quantization criterion and corresponding Quantitative fusion detection algorithm.But, existing multistation radar There is the problems such as detection performance loss is relatively big, quantization threshold optimization is not good enough in Quantitative fusion detection algorithm, needs research further.

Summary of the invention

Present invention aims to the deficiency of above-mentioned prior art, propose a kind of multistation quantified based on Pasteur's distance Radar Signal Fusion detection method, to ensure under conditions of the detection performance loss of signal fused inspection center is the least, with Bigization Pasteur's distance quantifies the echo-signal of radar station for criterion, when reducing the transmission of radar station echo-signal to signal fused center Required communication bandwidth.

For reaching above-mentioned purpose, embodiments of the invention adopt the following technical scheme that

A kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance, described method comprises the steps:

Step 1, arranges netted radar system, and described netted radar system includes a signal fused inspection center and N number of Radar station；

Step 2, determines the quantization digit b of each radar station echo-signal to receiving, and obtains according to quantization digit b Quantized interval number M=2^b, the label that each quantized interval is corresponding is expressed as m, m=0,1, and 2 ..., M-1, described quantization digit b Meet: 1≤b < C_h/f_s, wherein, C_hFor the maximum communication bandwidth between each radar station and signal fused inspection center, f_sFor often The highest sample frequency of individual radar station；

Step 3, its echo-signal received is quantified by i-th radar station, determines the institute of the echo-signal after quantization The quantized interval belonged to, and label corresponding for this quantized interval is transmitted to signal fused inspection center；The initial value of i is 1, i= 1,...,N；

Step 4, makes the value of i add 1, and performs step 3, until i > N；

Step 5, arranges desired false-alarm probability, described signal fused inspection center according to described desired false-alarm probability and Target is detected by label corresponding to the quantized interval belonging to echo-signal after the quantization that N number of radar station sends.

Advantages of the present invention is as follows: (1) due to the fact that and quantifies radar echo signal for criterion maximizing Pasteur's distance And use randomized detection criteria to adjudicate the presence or absence of target at signal fused center, therefore, it can in ensureing signal fused Under conditions of the detection performance loss of the heart is the least, reduces radar station echo-signal and transmit to communication required during signal fused center Bandwidth；(2) due to the fact that the sequential quadratic programming algorithm using improvement solves quantization threshold, therefore can obtain local optimum Quantization threshold value.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is only this Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to Other accompanying drawing is obtained according to these accompanying drawings.

Fig. 1 is a kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance that the embodiment of the present invention provides Realize schematic flow sheet；

Fig. 2 is that the present invention realizes schematic flow sheet when using the sequential quadratic programming algorithm of improvement to solve quantization threshold；

Fig. 3 be the sequential quadratic programming algorithm of the improvement that the present invention uses realize schematic flow sheet；

Fig. 4 is the detection probability contrast schematic diagram figure of the inventive method and non-quantized signal fusion detection method；

Fig. 5 is that the inventive method solves the quantization threshold curve synoptic diagram obtained；

Fig. 6 is that existing method solves the quantization threshold curve synoptic diagram obtained.

Detailed description of the invention

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Describe, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments wholely.Based on Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under not making creative work premise Embodiment, broadly falls into the scope of protection of the invention.

The embodiment of the present invention provides a kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance, such as Fig. 1 Shown in, described method comprises the steps:

Step 1, arranges netted radar system, and described netted radar system includes a signal fused inspection center and N number of Radar station.

Concrete, a given netted radar system, netted radar system includes a signal fused center and N number of thunder Reach station；N number of radar station is designated as 1 to n-th radar station, the maximum allowed between each radar station and signal fused center Communication bandwidth is the most identical；The maximum communication bandwidth allowed between radar station and signal fused center is designated as C_h, maximum communication band Wide C_hUnit be bps；The highest sample frequency in N number of radar station is designated as f_s, the highest sample frequency f_sUnit be conspicuous Hereby.

Step 2, determines the quantization digit b of each radar station echo-signal to receiving, and obtains according to quantization digit b Quantized interval number M=2^b, the label that each quantized interval is corresponding is expressed as m, m=0,1, and 2 ..., M-1, described quantization digit b Meet: 1≤b < C_h/f_s, wherein, C_hFor the maximum communication bandwidth between each radar station and signal fused inspection center, f_sFor often The highest sample frequency of individual radar station.

On the basis of meeting above formula, the value of quantization digit b is the biggest, and the detection performance at signal fused center is the best；But It is, when quantization digit b increases to a certain degree, to increase quantization digit b further and the detection performance at signal fused center is changed Kind the least, therefore, the experience value of quantization digit b is 3～5.

Step 3, its echo-signal received is quantified by i-th radar station, determines the institute of the echo-signal after quantization The quantized interval belonged to, and label corresponding for this quantized interval is transmitted to signal fused inspection center；The initial value of i is 1, i= 1,...,N；

Step 3 specifically includes following sub-step:

(3a) echo-signal that i-th radar station receives is designated as x_i, the signal to noise ratio of echo-signal is designated as λ_i, echo is believed Number statistic of test be designated as yi=x_i/μ_i, wherein, μ_iNoise power for i-th radar station；In practice, if signal to noise ratio λ_iWith noise power μ_iActual value unknown, its value can be obtained by the method for estimation；

(3b) M-1 quantization threshold of i-th radar station is designated as T_i,1,T_i,2,…,T_i,M-1, according to signal to noise ratio λ_iObtain Pasteur's distance B of i-th radar station_i:

$B_i=-ln\&lsqb;\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}\sqrt{P(v_i=m|H_1)P(v_i=m|H_0)}\&rsqb;$

Wherein, v_iFor statistic of test y_iLabel after quantization, H₁Represent and assume that target exists, H₀Represent that hypothesis does not has mesh Mark, P () represents probit, remembers statistic of test y_iLower limit be T_i,0=0, statistic of test y_iThe upper limit be T_i,M=+∞, Then:

$\left\{\begin{array}{c}P(v_i=m|H_1)=\exp \&lsqb;-T_{i,m}/(1+{\mathrm{\&lambda;}}_i)\&rsqb;-\exp \&lsqb;-T_{i,m+1}/(1+{\mathrm{\&lambda;}}_i)\&rsqb;\\ P(v_i=m|H_0)=\exp (-T_{i,m})-\exp (-T_{i,m+1})\end{array}\right.$

(3c) according to Pasteur's distance B of i-th radar station_i, obtain the Pasteur coefficient ρ of i-th radar station_iFor:

${\mathrm{\&rho;}}_i=\underset{m=0}{\overset{M-1}{\&Sigma;}}\sqrt{P(v_i=m|H_1)P(v_i=m|H_0)};$

(3d) to minimize the Pasteur coefficient ρ of i-th radar station_iFor criterion, build and solve quantization threshold T_i,1,T_i,2,…, T_i,M-1Optimized model be:

$\begin{array}{cc}\underset{T_{i,1},T_{i,2},...,T_{i,M-1}}{\min }& {\mathrm{\&rho;}}_1\\ s.t.& T_{i,1}\&GreaterEqual;0,T_{i,2}\&GreaterEqual;T_{i,1},...,T_{i,M-1}\&GreaterEqual;T_{i,M-2}\end{array},$

Wherein, s.t. represents constraints, thus solves M-1 the quantization threshold T obtaining i-th radar station_i,1, T_i,2,…,T_i,M-1；

The nonlinear programming problem that Optimized model is belt restraining in step (3d), it is possible to use existing nonlinear optimization Algorithm for Solving；Owing to sequential quadratic programming algorithm has, optimal speed is fast, solve the advantages such as effective, and therefore, this example uses The sequential quadratic programming algorithm improved solves and obtains quantization threshold T_i,1,T_i,2,…,T_i,M-1。

As in figure 2 it is shown, in sub-step (3d), according to quantization threshold T_i,1,T_i,2,…,T_i,M-1Optimized model solve and obtain M-1 quantization threshold T of i-th radar station_i,1,T_i,2,…,T_i,M-1Detailed process be:

(3d1) cycle-index N is set_c, arrange a M-1 dimension stores vector T temporarily, arranges minimum Pasteur's coefficient value ρ_min=M, arranges cycle-index labelling k=1；

(3d2) random quantization threshold initial value is set

(3d3) according to quantization threshold initial valueUse the sequential quadratic programming algorithm improved, specifically Algorithm refers to document [A.Antoniou and W.-S.Lu, Practical optimization:algorithms and engineering applications:Springer,2007]；The thought of this algorithm is: in each iteration, by solving One quadratic programming subproblem obtains the descent direction of optimized variable, then uses one dimensional line searching algorithm to solve and is optimized The decline step-length of variable, according to descent direction and decline step-length more new variables, then uses BFGS Quasi-Newton algorithm to update secondary Hessian matrix in planning, finally starts next iteration；The flow process of this algorithm is as shown in Figure 3；According in step (3d) Optimized model solves the quantization threshold value obtaining this circulationWith minimize after Pasteur's coefficient value

If the Pasteur's coefficient value after (3d4) minimizingThen make and store vector temporarily Minimum Pasteur's coefficient value

If (3d5) cycle-index labelling k < N_c, then the value of cycle-index labelling k adds 1, and goes to step (3d2)；No Then, step (3d6) is gone to；

(3d6) according to storing vector T, the quantization threshold T after being solved temporarily_i,1=T (1), T_i,2=T (2) ..., T_i,M-1=T (M-1)；T (j) represents interim one element of jth storing vector T.

(3e) according to M-1 quantization threshold T of i-th radar station_i,1,T_i,2,…,T_i,M-1, to statistic of test y_iCarry out Quantify；If statistic of test y_iMeet T_i,m≤y_i<T_i,m+1, then statistic of test y_iLabel v after quantization_i=m；I-th thunder Reach the label v after quantifying that stands_iTransmission is to signal fused inspection center.

Step 4, makes the value of i add 1, and performs step 3, until i > N；Thus complete the label after N number of radar station will quantify Transmit the process to signal fused inspection center.

Step 5, arranges desired false-alarm probability, described signal fused inspection center according to described desired false-alarm probability and Target is detected by label corresponding to the quantized interval belonging to echo-signal after the quantization that N number of radar station sends.

Step 5 specifically includes following sub-step:

(5a) according to the label m=0 that quantized interval number M is corresponding with quantized interval, 1,2 ..., M-1, obtain signal fused Institute's likely value v of inspection center's label summation_sum=0,1,2 ..., N × (M-1) }, wherein, × represent that multiplication sign, note are assumed When not having a target, signal fused inspection center label summation value is v_sumProbability be P (v_sum|H₀)；

(5b) according to desired false-alarm probability P_fa, by following formula

$P_{fa}=\underset{v_{sum}>g}{\&Sigma;}P\left(v_{sum}\right|H_0)+\mathrm{\&gamma;}\&times;\underset{v_{sum}=g}{\&Sigma;}P\left(v_{sum}\right|H_0)$

Solve and obtain detection threshold g and probit γ；

(5c) the label v of N number of radar station that signal fused inspection center will receive₁,v₂,…,v_NSummation: If label sum v_f> g, then judgement has target；If label sum v_f=g, then have target with the judgement of probit γ；If mark Number sum v_f< g then adjudicates driftlessness.

The effect of the present invention is further illustrated by the test of following simulation comparison:

1. simulation parameter is arranged: multistation radar fence includes 1 signal fused center and 5 radar stations, these 5 radar stations Echo-signal there is identical signal to noise ratio, the span of signal to noise ratio is 0dB to 20dB, and the value of signal to noise ratio is spaced apart 0.5dB；Quantization digit b=4, i.e. quantized interval number M=16；The false-alarm probability at signal fused center is set to 10-6；Asking When solving the quantization threshold of each radar station, cycle-index N is set_c=20.

2. emulation content:

Arrange according to simulation parameter, detection probability curve such as the circle in Fig. 4 corresponding in calculating signal to noise ratio span Shown in mark line；In order to contrast, give the detection probability curve such as the square marks in Fig. 4 when radar echo signal does not quantifies Shown in line.

The sequential quadratic programming algorithm improved is used to solve the thresholding obtained as shown in Figure 5；In order to contrast, use document [M.Longo,T.D.Lookabaugh,and R.M.Gray,"Quantization for decentralized hypothesis testing under communication constraints,"IEEE Transactions on Information Theory, vol.36, pp.241-255,1990.] in the thresholding tried to achieve of two step descent methods as shown in Figure 6.

3. analysis of simulation result:

As shown in Figure 4, the detection probability of the present invention is close to detection probability when not quantifying；When detection probability is 0.5, with Detection performance when not quantifying is compared, and the detection performance loss of the present invention is 0.157dB, so the detection performance loss of the present invention The least.

As shown in Figure 5, the present invention solves the quantization threshold curve obtained to be increased with the increase of signal to noise ratio, and quantifies door Limit line smoothing；And there is undulation, consistent Changing Pattern useless in the quantization threshold curve in Fig. 6.Therefore, the present invention makes Solve with the sequential quadratic programming algorithm improved and obtained more preferable quantization threshold.

In radar, traditional quantization digit is generally 12, and the quantization digit of the present invention is 4, therefore, with traditional Quantization method is compared, and the communication bandwidth needed for the present invention reduces to original 1/3rd.

Summary emulation experiment it can be seen that the present invention use the sequential quadratic programming algorithm of improvement solve obtained good Quantization threshold, ensure signal fused center detection performance loss the least under conditions of, reduce radar station echo-signal Transmission is to communication bandwidth required during signal fused center.

The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any Those familiar with the art, in the technical scope that the invention discloses, can readily occur in change or replace, should contain Cover within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with described scope of the claims.

Claims (4)

1. the multistation Radar Signal Fusion detection method quantified based on Pasteur's distance, it is characterised in that described method includes Following steps:

Step 1, arranges netted radar system, and described netted radar system includes a signal fused inspection center and N number of radar Stand；

Step 2, determines the quantization digit b of each radar station echo-signal to receiving, and is quantified according to quantization digit b Interval number M=2^b, the label that each quantized interval is corresponding is expressed as m, m=0,1, and 2 ..., M-1, described quantization digit b are full Foot: 1≤b ＜ C_h/f_s, wherein, C_hFor the maximum communication bandwidth between each radar station and signal fused inspection center, f_sFor often The highest sample frequency of individual radar station；

Step 3, its echo-signal received is quantified, determines belonging to the echo-signal after quantization by i-th radar station Quantized interval, and by label transmission corresponding for the quantized interval belonging to the echo-signal after this quantization to signal fused detection The heart；The initial value of i is 1, i=1 ..., N；

Step 4, makes the value of i add 1, and performs step 3, until i ＞ N；

Step 5, arranges desired false-alarm probability, and described signal fused inspection center is according to described desired false-alarm probability and N number of Target is detected by label corresponding to the quantized interval belonging to echo-signal after the quantization that radar station sends.

A kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance the most according to claim 1, it is special Levying and be, step 3 specifically includes following sub-step:

(3a) echo-signal that i-th radar station receives is designated as x_i, the signal to noise ratio of echo-signal is designated as λ_i, echo-signal Statistic of test is designated as y_i=x_i/μ_i, wherein, μ_iNoise power for i-th radar station；

(3b) M-1 quantization threshold of i-th radar station is designated as T_{I, 1}, T_{I, 2}..., T_{I, M-1}, according to signal to noise ratio λ_iObtain i-th Pasteur's distance B of radar station_i:

$B_i=-ln\&lsqb;\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}\sqrt{P(v_i=m|H_1)P(v_i=m|H_0)}\&rsqb;$

Wherein, v_iFor statistic of test y_iLabel after quantization, H₁Represent and assume that target exists, H₀Represent that hypothesis does not has target, P () represents probit, remembers statistic of test y_iLower limit be T_i,0=0, statistic of test y_iThe upper limit be T_i,M=+∞, then:

$\left\{\begin{array}{c}P(v_i=m|H_1)=\exp \&lsqb;-T_{i,m}/(1+{\mathrm{\&lambda;}}_i)\&rsqb;-\exp \&lsqb;-T_{i,m+1}/(1+{\mathrm{\&lambda;}}_i)\&rsqb;\\ P(v_i=m|H_0)=\exp (-T_{i,m})-\exp (-T_{i,m+1})\end{array}\right.$

(3c) according to Pasteur's distance B of i-th radar station_i, obtain the Pasteur coefficient ρ of i-th radar station_iFor:

${\mathrm{\&rho;}}_i=\underset{m=0}{\overset{M-1}{\&Sigma;}}\sqrt{P(v_i=m|H_1)P(v_i=m|H_0)};$

(3d) to minimize the Pasteur coefficient ρ of i-th radar station_iFor criterion, build and solve quantization threshold T_i,1,T_i,2,…,T_i,M-1 Optimized model be:

$\begin{array}{cc}\underset{T_{i,1},T_{i,2},...,T_{i,M-1}}{\min }& {\mathrm{\&rho;}}_1\\ s.t.& T_{i,1}\&GreaterEqual;0,T_{i,2}\&GreaterEqual;T_{i,1},...,T_{i,M-1}\&GreaterEqual;T_{i,M-2}\end{array},$

Wherein, s.t. represents constraints, thus solves M-1 the quantization threshold T obtaining i-th radar station_i,1,T_i,2,…, T_i,M-1；

(3e) according to M-1 quantization threshold T of i-th radar station_i,1,T_i,2,…,T_i,M-1, to statistic of test y_iQuantify； If statistic of test y_iMeet T_i,m≤y_i<T_i,m+1, then statistic of test y_iLabel v after quantization_i=m；I-th radar station will Label v after quantization_iTransmission is to signal fused inspection center.

A kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance the most according to claim 1, it is special Levy and be, in sub-step (3d), according to quantization threshold T_i,1,T_i,2,…,T_i,M-1Optimized model solve and obtain i-th radar station M-1 quantization threshold T_i,1,T_i,2,…,T_i,M-1Detailed process be:

(3d1) cycle-index N is set_c, arrange a M-1 dimension stores vector T temporarily, arranges minimum Pasteur coefficient value ρ_min=M, Cycle-index labelling k=1 is set；

(3d2) random quantization threshold initial value is set

(3d3) according to quantization threshold initial valueOptimized model according to quantization threshold:Solve the quantization threshold value obtaining this circulationWith minimize After Pasteur's coefficient value

If the Pasteur's coefficient value after (3d4) minimizingThen make and store vector temporarily Minimum Pasteur's coefficient value

If (3d5) cycle-index labelling k ＜ N_c, then the value of cycle-index labelling k adds 1, and goes to step (3d2)；Otherwise, turn To step (3d6)；

(3d6) according to storing vector T, the quantization threshold T after being solved temporarily_{I, 1}=T (1), T_{I, 2}=T (2) ..., T_{I, M-1}= T(M-1)；T (j) represents interim one element of jth storing vector T.

A kind of multistation Radar Signal Fusion detection method quantified based on Pasteur's distance the most according to claim 1, it is special Levying and be, step 5 specifically includes following sub-step:

(5a) according to the label m=0 that quantized interval number M is corresponding with quantized interval, 1,2 ..., M-1, obtain signal fused detection Institute's likely value v of center label summation_sum=0,1,2 ..., N × (M-1) }, wherein, × represent that multiplication sign, note hypothesis do not have During target, signal fused inspection center label summation value is v_sumProbability be P (v_sum|H₀)；

(5b) according to desired false-alarm probability P_fa, by following formula

$P_{fa}=\underset{v_{sum}>g}{\&Sigma;}P\left(v_{sum}\right|H_0)+\mathrm{\&gamma;}\&times;\underset{v_{sum}=g}{\&Sigma;}P\left(v_{sum}\right|H_0)$

Solve and obtain detection threshold g and probit γ；

(5c) the label v of N number of radar station that signal fused inspection center will receive₁, v₂..., v_NSummation:If Label sum v_f＞ g, then judgement has target；If label sum v_f=g, then have target with the judgement of probit γ；If label it And v_f＜ g, then adjudicate driftlessness.