CN104616059A - DOA (Direction of Arrival) estimation method based on quantum-behaved particle swarm - Google Patents

Abstract

The invention discloses a DOA (Direction of Arrival) estimation method based on a quantum-behaved particle swarm; the method comprises the steps: establishing an array signal model and setting relevant parameters of an array; establishing maximum likelihood estimation according to the array signal data; then performing initialization, calculating a fitness function for each particle, updating the speed and position of the quantum-behaved particle swarm; judging whether the maximum iteration times is obtained; and finally, outputting the vector estimated value of information message incidence angles and calculating the mean square error of incidence direction angles. The quantum behaviors of the particles are utilized in the optimization process; the DOA estimation method overcomes the disadvantage that the standard particle swarm algorithm is easy to fall into the partial minimum value; the standard particle swarm algorithm is high in estimation precision and has better effect for solving the DOA estimation problem; and a new solving idea is provided to the high-resolution DOA estimation.

Description

A kind of Wave arrival direction estimating method based on quantum particle swarm

Technical field

The present invention relates to Array Signal Processing and technical field of intelligence, specifically a kind of Mutual coupling technology based on quantum particle swarm.

Background technology

At present, the Mutual coupling based on array mainly contains classic method, subspace method, maximum likelihood method and MUSIC algorithm.Classic method needs more array element guarantee high resolving power, thus limits its application; Subspace method takes full advantage of the space structure receiving data, and be signal subspace and noise subspace by data decomposition, its performance is better than classic method; Maximum likelihood method has good robustness, even if also can obtain good performance when signal to noise ratio (S/N ratio) is lower, but it needs huge calculated amount, makes in engineering, limit its application; There is the difficulty of multidimensional spectrum peak search in MUSIC algorithm.

Quanta particle swarm optimization (QPSO), a kind of colony intelligence computing method of new rise in recent years, its incorporating quantum physics basic theories, quantum theory is applied to the particle swarm optimization algorithm of standard, from a kind of new particle swarm optimization algorithm model that the angle of Wave Functions of Quantum Mechanics proposes, it is the brand-new improvement to PSO algorithm, and the search performance of particle is far superior to basic PSO algorithm, is applied in the field such as function optimization, neural metwork training.

The defect existed for the above-mentioned existing Mutual coupling technology based on array or deficiency, the applicant is engaged in the technology accumulation of the industry then for many years with it, study energetically and how quanta particle swarm optimization is applied in the estimation of direction of arrival, to the disappearance of prior art can be improved, finally under the discretion of each side's condition is considered, develop the present invention.

Summary of the invention

The object of the invention is to overcome non-linear complicated optimum problem in existing DOA estimation algorithm, a kind of Wave arrival direction estimating method based on quantum particle swarm is proposed, the quantum behavior of particle is make use of in searching process, and search for particle and combine collaborative between individual local extremum and global extremum switching, give full play to the interparticle collaboratively searching mechanism of search, there is the ability of effective search and global search, realize best direction of arrival effect to reach.

In order to achieve the above object and effect, the present invention adopts following technology contents:

Based on a Wave arrival direction estimating method for quantum particle swarm, comprise the steps:

(1) array number is determined plane wave signal number P, signal center's wavelength X, fast umber of beats adjacent array element distance d, carries out feature decomposition in conjunction with Received signal strength to the maximal possibility estimation of data covariance matrix and obtains signal subspace S, noise subspace G;

(2) initialization; Determine the population scale M of population, the initial position vector of particle is z _i, the velocity vector that particle is corresponding is v _i, i=1,2 ..., M, maximal rate V _max, iterations k=1, maximum iteration time K _max, the local optimum position of each particle and the global optimum position of whole population

(3) fitness function of each particle is calculated; By M particle position z _i, i=1,2 ..., M, as the estimated value of direction of arrival angle θ, substitutes into the Power estimation fitness function of MUSIC algorithm

$f\left(\mathrm{\θ}\right)=\frac{a^H\left(\mathrm{\θ}\right)a\left(\mathrm{\θ}\right)}{a^H\left(\mathrm{\θ}\right)(I-{\mathrm{SS}}^H)a\left(\mathrm{\θ}\right)}$

Obtain each particle ideal adaptation degree functional value;

Wherein, I is unit matrix, $a\left(\mathrm{\θ}\right)={[1,e^{\mathrm{j\φ}\left(\mathrm{\θ}\right)},...,e^{j(\stackrel{\‾}{N}-1)\mathrm{\φ}\left(\mathrm{\θ}\right)}]}^T,\mathrm{\φ}\left(\mathrm{\θ}\right)=\frac{2\mathrm{\π}}{\mathrm{\λ}}d\sin \mathrm{\θ};$

(4) quanta particle group velocity and location updating; Kth+1 iteration, particle is according to following formula renewal speed and position:

$v_i^{k+1}=\frac{{2r}_1{p}_i^k+2.1r_2{p}_g^k}{{2r}_1+2.1r_2};$

Wherein: i=1,2 ..., M, α are shrinkage expansion coefficient; r ₁, r ₂, r ₃, r ₄get equally distributed random number between [0,1], be the desired positions of i-th particle experience, for the desired positions that all particle populations experience;

(5) if reach maximum iteration time k=K _max, then optimizing terminates, and front P the global optimum's position vector obtained is optimum direction of arrival angular estimation value, namely exports information source incident angle vector estimated value, and then calculates the mean square deviation at incident direction angle; Otherwise k:=k+1, forwards step 3 to.

The present invention at least has following beneficial effect:

The present invention has more state than standard particle group algorithm, and the movement locus that particle neither one is determined, the optional position in search volume can be appeared at a certain probability determined, and this position can be away from local attraction's point, even may be better than global optimum position in current population, thus substantially increase the diversity of particle, avoid algorithm Premature Convergence, thus substantially increase Optimizing Search efficiency and performance, enhance the ability of searching optimum of algorithm.Therefore, the present invention can realize Mutual coupling better.

Other objects of the present invention and advantage can be further understood from the technology contents disclosed by the present invention.In order to above and other object of the present invention, feature and advantage can be become apparent, special embodiment below also coordinates institute's accompanying drawings to be described in detail below.

Accompanying drawing explanation

Fig. 1 is Wave arrival direction estimating method process flow diagram of the present invention.

Fig. 2 is the graph of a relation that embodiment of the present invention medium wave reaches between bearing estimate mean square deviation and signal to noise ratio (S/N ratio).

Embodiment

Next will coordinate institute's accompanying drawings through embodiment, and illustrate that the present invention has the unique technology parts such as innovation, progressive or effect compared with prior art, those of ordinary skill in the art can be realized according to this.Should be noted that, the modification that those of ordinary skill in the art carry out under not departing from spirit of the present invention and change, all do not depart from protection category of the present invention.

Consider one the linear homogeneous array of array element, there is P arrowband plane wave signal in space, and the Received signal strength baseband envelope of i-th array element can be expressed as

$x_i\left(t\right)=\underset{k=1}{\overset{P}{\mathrm{\Σ}}}{s}_k\left(t\right)\exp \left[j\frac{2\mathrm{\π}}{\mathrm{\λ}}d(i-1)\sin {\mathrm{\θ}}_k\right]+w_i\left(t\right)---\left(1\right)$

Wherein, λ is signal center's wavelength, and d is adjacent array element distance, θ _kfor a kth signal source direction and Array Method to angle (i.e. direction of arrival angle), s _kfor the baseband envelope of a kth signal source, w _it () is that the additivity in i-th array element receives noise.

Formula (1) is expressed as vector form:

X(t)＝A(θ)s(t)+w(t) (2)

Wherein, $X\left(t\right)={[x_1\left(t\right),...,x_{\stackrel{\‾}{N}}\left(t\right)]}^T,$ s(t)＝[s ₁(t),…,s _P(t)] ^T， $w\left(t\right)={[w_1\left(t\right),...,w_{\stackrel{\‾}{N}}\left(t\right)]}^T,$ T represents transposition.A (θ) is rank matrix, its column vector is

$a\left({\mathrm{\θ}}_k\right)={[1,e^{\mathrm{j\φ}\left({\mathrm{\θ}}_k\right)},...,e^{j(\stackrel{\‾}{N}-1)\mathrm{\φ}\left({\mathrm{\θ}}_k\right)}]}^T,k=\mathrm{1,2},...,P,---\left(3\right)$

Wherein, $\mathrm{\φ}\left({\mathrm{\θ}}_k\right)=\frac{2\mathrm{\π}}{\mathrm{\λ}}d\sin {\mathrm{\θ}}_k,k=\mathrm{1,2},...,P.$

Maximal possibility estimation according to covariance matrix is:

$R=\frac{1}{\stackrel{\‾}{M}}\underset{m=1}{\overset{\stackrel{\‾}{M}}{\mathrm{\Σ}}}X\left(t_m\right)X^H\left(t_m\right)---\left(4\right)$

Wherein, for fast umber of beats, H is complex-conjugate transpose.Carry out feature decomposition to R to have:

R＝UΣU ^H＝SΣ _SS ^H+GΣ _NG ^H(5)

Wherein, U=[S|G], S are the subspaces of being opened by large eigenwert characteristic of correspondence vector, i.e. signal subspace; G is the subspace of being opened by little eigenwert characteristic of correspondence vector, i.e. noise subspace.Due to the existence of noise, the steering vector of signal subspace and noise subspace can not be completely orthogonal.Therefore, direction of arrival realizes to minimize Optimizing Search: according to character SS ^h+ GG ^h=I, in actual optimization algorithm, calculates the Power estimation fitness function of MUSIC algorithm

$f\left(\mathrm{\θ}\right)=\frac{a^H\left(\mathrm{\θ}\right)a\left(\mathrm{\θ}\right)}{a^H\left(\mathrm{\θ}\right)(I-{\mathrm{SS}}^H)a\left(\mathrm{\θ}\right)}---\left(6\right)$

Wherein, I is unit matrix.Angle θ corresponding to f (θ) minimum value is estimated direction of arrival angle.

In one embodiment, one is considered the linear homogeneous array of array element, adjacent array element distance d=1, there is P=1 arrowband plane wave signal in space, signal center wavelength X=2, and its signal amplitude is signal angular frequency is 1.0, and signal initial phase is π/3, and this signal is respectively π/4 relative to the actual direction of arrival angle of array.Utilize Received signal strength baseband envelope, feature decomposition is done to data covariance matrix and can obtain signal subspace S, noise subspace G.

As shown in Figure 1, embodiment can be divided into the following steps to the workflow of the inventive method:

Step 1: fast umber of beats is set according to given array number plane wave signal number P, signal center's wavelength X, adjacent array element distance d, carry out feature decomposition in conjunction with Received signal strength to the maximal possibility estimation of data covariance matrix and obtain signal subspace S, noise subspace G.

Step 2: initialization.Arrange the population scale M=20 of population, dimension size D=1, the initial position vector of particle is z _i∈ [0, pi/2], the velocity vector that particle is corresponding is v _i∈ (0,0.01), i=1,2 ..., M, maximal rate V _max=0.01, iterations k=1, maximum iteration time K _max=80, determine the local optimum position of each particle and the global optimum position of whole population

Step 3: the fitness function calculating each particle.By M particle position z _i(i=1,2 ..., M) as direction of arrival angle θ estimated value substitute into (6), obtain each particle ideal adaptation degree functional value;

$f\left(\mathrm{\θ}\right)=\frac{a^H\left(\mathrm{\θ}\right)a\left(\mathrm{\θ}\right)}{a^H\left(\mathrm{\θ}\right)(I-{\mathrm{SS}}^H)a\left(\mathrm{\θ}\right)}---\left(6\right)$

Wherein, I is unit matrix, $a\left(\mathrm{\θ}\right)={[1,e^{\mathrm{j\φ}\left(\mathrm{\θ}\right)},...,e^{j(\stackrel{\‾}{N}-1)\mathrm{\φ}\left(\mathrm{\θ}\right)}]}^T,\mathrm{\φ}\left(\mathrm{\θ}\right)=\frac{2\mathrm{\π}}{\mathrm{\λ}}d\sin \mathrm{\θ}.$

Step 4: quanta particle group velocity and location updating.Kth+1 iteration, particle is according to following formula renewal speed and position:

$v_i^{k+1}=\frac{{2r}_1{p}_i^k+2.1r_2{p}_g^k}{{2r}_1+2.1r_2}$

Wherein: i=1,2 ..., M, M are population size, usually get 20 ~ 40; α is shrinkage expansion coefficient, r ₁, r ₂, r ₃, r ₄get equally distributed random number between [0,1], be the desired positions (individual extreme value) of i-th particle experience, for the desired positions (global extremum) that all particle populations experience.

Step 5: if reach maximum iteration time (k=K _max), then optimizing terminates, and front P the global optimum's position vector obtained is optimum direction of arrival angular estimation value, namely exports information source incident angle vector estimated value, and then calculates the mean square deviation at incident direction angle; Otherwise k:=k+1, goes to step 3.

Give embodiment of the present invention medium wave in Fig. 2 and reach graph of a relation between bearing estimate mean square deviation and signal to noise ratio (S/N ratio).Visible, utilize the present invention program to carry out Mutual coupling precision high.

In sum, the present invention make use of the quantum behavior of particle in searching process, and overcome standard particle group algorithm and be easily absorbed in local minimum shortcoming, estimated accuracy is high, achieves good effect for solution Mutual coupling problem.High-resolution Mutual coupling is the bottleneck improving communication system performance always, and the present invention obtains high-resolution Mutual coupling to provide a kind of new resolving ideas.

Above-described is only the preferred embodiment of the present invention, the invention is not restricted to above embodiment.Be appreciated that the oher improvements and changes that those skilled in the art directly derive without departing from the spirit and concept in the present invention or associate, all should think and be included within protection scope of the present invention.

Claims (1)

1. based on a Wave arrival direction estimating method for quantum particle swarm, it is characterized in that, described method comprises the steps:

(1) array number is determined plane wave signal number P, signal center's wavelength X, fast umber of beats adjacent array element distance d, carries out feature decomposition in conjunction with Received signal strength to the maximal possibility estimation of data covariance matrix and obtains signal subspace S, noise subspace G;

(2) initialization; Determine the population scale M of population, the initial position vector of particle is z _i, the velocity vector that particle is corresponding is v _i, i=1,2 ..., M, maximal rate V _max, iterations k=1, maximum iteration time K _max, the local optimum position of each particle and the global optimum position of whole population

(3) fitness function of each particle is calculated; By M particle position z _i, i=1,2 ..., M, as the estimated value of direction of arrival angle θ, substitutes into the Power estimation fitness function of MUSIC algorithm

$f\left(\mathrm{\θ}\right)=\frac{a^H\left(\mathrm{\θ}\right)a\left(\mathrm{\θ}\right)}{a^H\left(\mathrm{\θ}\right)(I-{\mathrm{SS}}^H)a\left(\mathrm{\θ}\right)}$

Obtain each particle ideal adaptation degree functional value;

Wherein, I is unit matrix, $a\left(\mathrm{\θ}\right)={[1,e^{\mathrm{j\φ}\left(\mathrm{\θ}\right)},...,e^{j(\stackrel{\‾}{N}-1)\mathrm{\φ}\left(\mathrm{\θ}\right)}]}^T,\mathrm{\φ}\left(\mathrm{\θ}\right)=\frac{2\mathrm{\π}}{\mathrm{\λ}}d\sin \mathrm{\θ};$

(4) quanta particle group velocity and location updating; Kth+1 iteration, particle is according to following formula renewal speed and position:

$v_i^{k+1}=\frac{2r_1{p}_i^k+2.1r_2{p}_g^k}{{2r}_1+2.1r_2};$

Wherein: i=1,2 ..., M, α are shrinkage expansion coefficient; r ₁, r ₂, r ₃, r ₄get equally distributed random number between [0,1], be the desired positions of i-th particle experience, for the desired positions that all particle populations experience;

(5) if reach maximum iteration time k=K _max, then optimizing terminates, and front P the global optimum's position vector obtained is optimum direction of arrival angular estimation value, namely exports information source incident angle vector estimated value, and then calculates the mean square deviation at incident direction angle; Otherwise k:=k+1, forwards step 3 to.