RU2014681C1 - Adaptive array - Google Patents

Abstract

FIELD: radio communication systems. SUBSTANCE: adaptive array has N aerial elements, N hybrid units, N weight multipliers, common adder, unit of estimation of signal power, unit of estimation of noise power, comparator, control unit, first multiplier, 2N adaptive circuits, 2N second subtracters, unit of vector multiplication, unit of dividers, unit of matrix multiplication, adder. EFFECT: improved efficiency under conditions of complex noises. 4 dwg

Description

 The invention relates to radio engineering and can be used in radio communication systems, radar and radio navigation, operating in a complex jamming environment.

 Adaptive antenna arrays (AAR) are known for use in radio communications, radiolocation and radio navigation.

 AAR, the design of which is given in TIIER, 1967, t.55, N 12, p.78-95, implements an algorithm to minimize the mean square disconnection of the received signal from the reference.

 For the algorithm to work in the device, it is necessary to form a reference signal. This is possible if there is a priori information about the useful signal. And since such information is never complete, since the meaning in the transmission of the useful signal would be lost, the reference signal may significantly differ from the useful one, which causes a decrease in the noise immunity of AAR. In addition, the convergence time largely depends on the conditionality of the correlation matrix of the input signals and can be significant.

 AAP, the construction of which is described in the journal IEEE Trans Antennas and Propag, vol. AP-26, 1978 N 2 p. 228-235, implements an algorithm for minimizing the output power and has relatively good noise immunity characteristics. However, in the case when there is no interference or its power is less than the power of the useful signal, the useful signal can be suppressed due to minimization of the total output power. In addition, the adaptation time can be quite large with poor conditionality of the correlation matrix of input signals.

 Of the known AAPs, the lattice described in ed. USSR N 1548820. This device contains N antenna elements connected through hybrid devices, and weighted multipliers with the corresponding inputs of the common adder, the first multiplier and 2N adaptive circuits, each of which consists of an integrator, switch, amplifiers, adjustable inverting amplifier, subtraction unit, the second multiplier and the correlator, the first and second inputs of the correlator are connected respectively to the output of the hybrid device and the output of the common adder, and the output of the correlator is connected to the first input of the block and subtraction, the second input of which is connected to the output of the first multiplier, and the output of the subtraction unit is connected to the inputs of the amplifier and an adjustable inverting amplifier, the outputs of which are connected to the inputs of the switch, the output of which through the integrator is connected to the second input of the weight multiplier and to the first input of the second multiplier, and the second input of the second multiplier is connected to the output of the first multiplier, the first and second inputs of which are combined with the output of the common adder and the second input of the correlator, as well as a signal power estimation unit, bl ok estimates the interference power, the comparison unit and the control unit, and the inputs of the units for evaluating the signal power and interference power are connected to the output of the common adder, and the outputs are connected to the inputs of the comparison unit, the output of which is connected to the control inputs of the switches, the output of the control unit is connected to the control inputs of the adjustable inverting amplifiers, and the output is connected to the output of the interference power estimation unit.

 Depending on the signal-noise environment, the device operates by the method of minimizing or maximizing the output power, which prevents the suppression of the useful signal and increases the noise immunity of the grating. However, although the magnitude of the gain in the feedback circuit varies depending on the interference power, the adaptation time of the device can turn out to be rather long if the correlation matrix of the input signals is poorly conditioned (the spread of the eigenvalues of the correlation matrix). Moreover, the eigenvalues of the correlation matrix depend on the power and amount of interference. When there are several interference of different power, the eigenvalues of the correlation matrix are not equal to each other (there is some variation), the adaptive antenna array reacts quickly to strong interference and slowly to weak interference, and the adaptation time is determined by the slowest component of the adaptation process.

 The aim of the invention is to reduce the adaptation time of the device in the presence of several interference of different power.

This is achieved by the fact that in the AAR, containing N antenna elements connected through hybrid devices and weight multipliers with the corresponding inputs of the common adder, the first multiplier and 2N adaptive circuits, each of which consists of an integrator, commutator, amplifier, adjustable inverting amplifier, and a subtraction unit , the second multiplier and the correlator, the first and second inputs of the correlator are connected respectively to the output of the hybrid device and the output of the common adder, and the output of the correlator is connected to the first input subtraction lock, the second input of which is connected to the output of the second multiplier, and the output of the subtraction block is connected to the inputs of the amplifier and an adjustable inventory amplifier, the outputs of which are connected to the inputs of the switch, the output of which through the integrator is connected to the second input of the weight multiplier and to the first input of the second multiplier, and the second input of the second multiplier is connected to the output of the first multiplier, the first and second inputs of which are combined with the output of the common adder and the second input of the correlator, as well as a signal power estimator , an interference power estimator, a comparison unit, and a control unit, the inputs of the signal power and interference power estimator being connected to the output of the common adder, and the outputs connected to the inputs of the comparison unit whose output is connected to the control inputs of the switches, the input of the control unit is connected to the output of the unit estimates of the interference power, and the output to the control inputs of adjustable inverting amplifiers, a matrix multiplication block, a divider block, an adder, a vector multiplication block and 2N second subtraction blocks are additionally introduced eat first and second 2N input unit of the vector multiplication are connected to the outputs of hybrid devices, the first 4N ² vector multiplication unit outputs are connected to first inputs of the divider unit and the second 2N outputs via an adder connected to the second inputs of dividers unit which outputs are connected with the first 4N ² inputs a matrix multiplication block, the second 2N inputs of which are combined with the outputs of the first subtraction blocks and with the first inputs of the second subtraction blocks, and the 2N outputs of the matrix multiplication block are connected with the second inputs of the second subtraction blocks Ania, the outputs of which are connected to the input of the amplifier and adjustable inverting amplifiers.

 An analysis of similar technical solutions showed that the set of distinctive features that determine the achievement of the specified positive effect in the known devices of identical purpose was not found.

 Figure 1 presents the functional diagram of the AAP; figure 2 - block diagram of vector multiplication; figure 3 is a block diagram of matrix multiplication; figure 4 - simulation results AAR.

 AAR contains antenna elements 1, hybrid devices 2, weight multipliers 3, a common adder 4, a block 5 for estimating the signal power, a block 6 for estimating the interference power, a block 7 for comparison, a block 8 for control, the first multiplier 9, adaptive circuits 10 consisting of second multipliers 11, integrators 12, amplifiers 13, switches 14, adjustable inverting amplifiers 15, first subtraction blocks 16, second subtraction blocks 17 and correlators 18, as well as vector multiplication block 19, adder 20, divider block 21, matrix multiplication block 22.

 Adaptive antenna array operates as follows.

 Radio signals are received by the antenna elements 1 and fed to hybrid devices 2, in which they are divided into in-phase and quadrature components, and then weighed by weight multipliers 3 and summed in a common adder 4, the output of which is the output of the device. Using adaptive circuits 10, weight multipliers 3 are adjusted to increase the signal-to-noise ratio + noise at the output of the device. If the useful signal power is less than the interference power, then the output power is minimized, and if the signal power is greater than the interference power, the output power is maximized. The inclusion of one or another mode is carried out by the switch 14, depending on the ratio of the useful signal power and interference. The signal and interference powers are evaluated in the signal power estimation unit 5 and in the interference power evaluation unit 6 and equivalent voltages are supplied to the inputs of the comparison unit 7. Depending on the ratio of these voltages, the comparison unit 7 controls the switches 14 of the adaptive circuits 10.

(K) (k) на каждой итерации должна умножаться слева на оценку обратной корреляционной матрицы R^{^-1} .Adaptive circuits 10 implement an output power optimization algorithm based on Newton's method. In this case, the gradient estimate of the objective function

(K) (k) at each iteration should be multiplied on the left by the estimate of the inverse correlation matrix R ^{^ -1} .

(K+1)=

(K)+

(K), (1) где

(K),

(K+1) - значение вектора весовых коэффициентов соответственно на k-й и (k+1)-й итерации;
μ - коэффициент усиления в адаптивных контурах;

(K) - оценка градиента на k-й итерации.

(K + 1) =

(K) +

(K), (1) where

(K)

(K + 1) is the value of the vector of weighting coefficients at the kth and (k + 1) th iteration, respectively;
μ is the gain in adaptive circuits;

(K) is the gradient estimate at the kth iteration.

 This leads to the fact that all components of the adaptive process converge with the same time constant. Such an algorithm has quadratic convergence.

 Obtaining an estimate of the inverse correlation matrix is a difficult task. In the selective assessment of the correlation matrix, difficulties arise with its handling in case of poor conditionality; a large number of computational operations are necessary. Recursive methods of treatment are also associated with a rather large bulk of calculations. Therefore, such methods are complicated in technical implementation.

As an estimate of the inverse correlation matrix, we use the pseudoinverse matrix R _po , formed by one sample.

To obtain a pseudoinverse matrix, we use the spectral representation of the correlation matrix R and the inverse correlation matrix R ^-1 .

; (2)

, (3) где Q - унитарная (ортонормированная QQ^т=

= I матрица собственных векторов Q=[

,

...

] ;
Λ- диагональная матрицы собственных значений Λ = diag(λ₁,λ₂...λ_п);
Т - знак транспонирования. Запишем выражение (3) в виде
R

1-

=

-

=I-

, (4) где I - единичная матрица I = diag(1,1...1).R = QΛQ ^t =

; (2)

, (3) where Q is unitary (orthonormal QQ ^m =

= I eigenvector matrix Q = [

,

...

];
Λ is the diagonal matrix of eigenvalues Λ = diag (λ ₁ , λ ₂ ... λ _p );
T is the sign of transposition. We rewrite expression (3) in the form
R

1-

=

-

= I-

, (4) where I is the identity matrix I = diag (1,1 ... 1).

^≈

(5)
Отметим также, что значение λ_max не может превышать след матрицы R
λ_max≅ t_r[R] (6) С учетом (5) и (6) перепишем выражение (4)

≈ I_

=I_

(7) Полученный результат совпадает с представлением обратной корреляционной матрицы в виде двух первых членов ряда при разложении ее по степеням R.Note that λ _i varies in the range from λ _min to λ _max - the minimum and maximum eigenvalues. We use the linear approximation of the scalar factor in expression (4) in the region ((λ _min ÷ λ _max ))

^≈

(5)
We also note that the value of λ _max cannot exceed the trace of the matrix R
λ _max ≅ t _r [R] (6) Taking into account (5) and (6), we rewrite expression (4)

≈ I_

= I_

(7) The result obtained coincides with the representation of the inverse correlation matrix in the form of the first two members of the series when it is expanded in powers of R.

≈ 1; C₁=-

.R ^-1 = C ₀ I + C ₁ R + C ₂ R ² + ... + C _to R ^k , (8) where k ≅ N in which the coefficients C ₀ =

≈ 1; C ₁ = -

.

(K)= S(K)[

K)-

(K)S(K)] , (9) где S(K)=

(K)

(K) - сигнал на выходе общего сумматора 4.As an estimate of the gradient in (1), we use the well-known expression

(K) = S (K) [

K) -

(K) S (K)], (9) where S (K) =

(K)

(K) - signal at the output of the total adder 4.

(K+1)=

(K)-

I-

S(K)[

(K)-

(K)S(K)], (10) где

(K) - вектор весовых коэффициентов на выходах интеграторов 12;

(K) - вектор входных сигналов на выходах гибридных устройств 2;
S(K)=

(K)

(K) - сигнал на выходе общего сумматора 4;
μ- коэффициент усиления усилителей 13 и 15 в адаптивных контурах 10. Когда коэффициент μ > 0, осуществляется минимизация мощности на выходе общего сумматора 4 (подавляются помехи), а когда μ< 0, выходная мощность сигнала общего сумматора 4 максимизируется (выделяется полезный сигнал). Выбор знака μ осуществляется при помощи коммутатора 14, а установка величины μ осуществляется блоком 8 управления в зависимости от мощности помех.In view of the foregoing, the algorithm for setting the weight multipliers 3 has the form

(K + 1) =

(K) -

I-

S (K) [

(K) -

(K) S (K)], (10) where

(K) is the vector of weights at the outputs of the integrators 12;

(K) is the vector of input signals at the outputs of hybrid devices 2;
S (K) =

(K)

(K) is the signal at the output of the common adder 4;
μ is the gain of amplifiers 13 and 15 in adaptive circuits 10. When the coefficient is μ> 0, the power at the output of the common adder 4 is minimized (interference is suppressed), and when μ <0, the output power of the signal of the common adder 4 is maximized (a useful signal is extracted) . The choice of the sign μ is carried out using the switch 14, and the installation of the value μ is carried out by the control unit 8 depending on the interference power.

на оценку градиента

S(K)[

K)-

(K)S(K)] осуществляется следующим образом.Formation of the product of a pseudoinverse matrix R _by = I-

on the gradient estimate

S (K) [

K) -

(K) S (K)] is as follows.

(K) самого на себя

(K)

(K)=

(11)
Схема блока 19 векторного умножения, состоящего из 4N² умножителей 1-1-2N-2N, приведена на фиг.2.First, in block 19 of vector multiplication, the external product of the vector is calculated

(K) yourself

(K)

(K) =

(eleven)
The circuit block 19 of the vector multiplication, consisting of 4N ² multipliers 1-1-2N-2N, shown in figure 2.

(K) самого на себя (квадрат модуля) производится в сумматоре 20, как сумма диагональных элементов (след) матрицы (11):
<

(K)

(K)>=

(K)

= x $\genfrac{}{}{0ex}{}{\text{2}}{\text{1}}$ (K)+x $\genfrac{}{}{0ex}{}{\text{2}}{\text{2}}$ (K)+. ..x $\genfrac{}{}{0ex}{}{\text{2}}{\text{2}}$ _N(K)=t_r[

(K)

(K)] (12) Затем, разделяя каждый элемент матрицы (11) на ее след в блоке 21 делителей, вычисляется нормированная матрица

.Calculation of the internal product of a vector

(K) of itself (the square of the module) is produced in the adder 20, as the sum of the diagonal elements (trace) of the matrix (11):
<

(K)

(K)> =

(K)

= x $\genfrac{}{}{0ex}{}{\text{2}}{\text{1}}$ (K) + x $\genfrac{}{}{0ex}{}{\text{2}}{\text{2}}$ (K) +. ..x $\genfrac{}{}{0ex}{}{\text{2}}{\text{2}}$ _N (K) = t _r [

(K)

(K)] (12) Then, dividing each element of the matrix (11) into its trace in the block 21 of the dividers, the normalized matrix is calculated

.

(K) целевой функции производится на выходах блоков 16 вычитания аналогично прототипу согласно выражению (9).Gradient Estimation

(K) the objective function is performed at the outputs of the subtraction blocks 16 similarly to the prototype according to expression (9).

.В результате получаем 2N - мерный вектор

= [a₁. ..a_2N] a_i=

i=

где δ_m - компоненты вектора градиента

(K);
r_im - компоненты нормированной матрицы. Схема блока матричного умножения, состоящего соответственно из 4 умножителей 1-1-2N-2N и 2N сумматоров 3-1-3-2N, представлена на фиг.3.In block 22 of the matrix multiplication, the gradient estimate is multiplied on the left by the normalized matrix [τ _ik ] =

. As a result, we obtain a 2N - dimensional vector

= [a ₁ . ..a _2N ] a _i =

i =

where δ _m are the components of the gradient vector

(K);
r _im are the components of the normalized matrix. The block diagram of the matrix multiplication, consisting respectively of 4 multipliers 1-1-2N-2N and 2N adders 3-1-3-2N, shown in Fig.3.

(K) и компонент вектора

(K). Таким образом

(K+1) =

(K)m[

(K)-

(K)] (13) В результате движение осуществляется не по направлению антиградиента (градиента), а по направлению на точку экстремума, что значительно уменьшает время адаптации, когда линии уровня имеют вытянутый характер.The adaptive addition in the adaptation algorithm (10) is calculated in subtraction blocks 17 as the difference of the gradient components

(K) and the component of the vector

(K). In this way

(K + 1) =

(K) m [

(K) -

(K)] (13) As a result, the movement is carried out not in the direction of the antigradient (gradient), but in the direction to the extremum point, which significantly reduces the adaptation time when the level lines are elongated.

 The use of new elements - block 19 vector multiplication, adder 20, block 21 dividers, block 22 matrix multiplication favorably distinguishes the proposed AAR from the prototype, since the adaptation time constant becomes almost the same for all components of the adaptation process. As a result, the adaptation time will be reduced in the presence of several interference of different power, which corresponds to poor conditioning of the correlation matrix of input signals (a large spread of the eigenvalues of the correlation matrix).

 Figure 4 presents the simulation results of the claimed AAP and prototype. From the graphs it can be seen that both devices have the same efficiency in terms of obtaining the output ratio of the signal power to the sum of the interference and noise powers (SINR), but the adaptation time in the claimed AAR is much shorter.

 This advantage helps to increase the efficiency of the AAR in difficult interference environments.

Claims (1)

ADAPTIVE ANTENNA ARRAY, containing N antenna elements, the outputs of which are connected to the inputs of the corresponding hybrid units, the first and second outputs of which are connected through the respective weight multipliers to the corresponding inputs of the common adder, the output of which is the output of the array, the signal power estimation unit and the comparison unit connected in series, a series-connected unit for evaluating the interference power and a control unit, the input of which is connected to the second input of the comparison unit, the first multiplier, the first and second the first inputs of which are connected to the output of the common adder, the input of the signal power estimator and the input of the interference power estimator, 2 N adaptive circuits, each of which contains a serially connected controlled inverting amplifier, commutator, integrator, second multiplier and first subtraction unit, amplifier, input which is connected to the first input of the controlled inverting amplifier, and the output to the second input of the switch, a correlator, the output of which is connected to the second input of the first subtraction unit, and the first and second input accordingly, with the output of the corresponding hybrid unit and the first input of the first multiplier, the output of which is connected to the second input of the corresponding second multiplier, the first input of which is connected to the second input of the corresponding weight multiplier, while the output of the comparison unit is connected to the third input of the corresponding switch, and the output of the control connected to the second input of the corresponding controlled inverting amplifier, characterized in that, in order to reduce the adaptation time, in the presence of several an ommech of different power, series-connected vector multiplication block, divider block and matrix multiplication block are introduced, an adder whose output is connected to the first input of the divider block, and 2 N inputs to the corresponding first outputs of the vector multiplication block, the first and second 2 N inputs of which are connected with respective outputs of hybrid units 4 N ² inputs of block dividers are connected to respective second outputs of the vector multiplication unit, and N ² O 4 - to the corresponding first inputs of the matrix multiplying unit And in each adaptive loop is entered a second subtracter whose output is connected to the amplifier input and the first input - first output from the subtraction unit and the corresponding first input of the matrix multiplication outputs are connected to second inputs of the respective second subtracting units.

Images