CN100592622C - Method and device for realizing automatic gain control - Google Patents

Abstract

The invention provides a method for reaching auto-gain control and its device. The method comprises that using the method of quadratic function matching to attain the observable estimated value and using the method of least mean square error to calculate the total gain control signal; the device mainly supplies a quadratic weighted detecting module which can attain the observable estimated value by the method of quadratic function matching and a self-adaptive control signal calculate module which can calculate the total gain control signal by the method of least mean square error. The invention increases the accuracy of attained observable estimated value and the total gain control signal, to improve the properties of catching time, traceability, and the signal-to-noise ratio of said system, therefore, the whole property of system is improved.

Description

Realize the method and the device of automatic gain control

Technical field

The present invention relates to the automatic gain control field, be meant method and the device of realizing automatic gain control especially.

Background technology

In mobile communication system, the dynamic change scope of the feasible received signal that arrives of abominable wireless propagation environment is very big, can reach about 80dB.And the wide variation of received signal will be brought many difficulties to system design, therefore, in the receiver of communication system, generally adopt the automatic gain control signal that the dynamic change scope is bigger to be adjusted into the less signal of dynamic change scope.Automatic gain control (AGC, Automatic Gain Control) can have a lot of middle implementations.From the circuit angle, can adopt analog circuit or digital circuit, from the feedback kind angle, can adopt feedback-type AGC and feed-forward type AGC, from realizing the position, can in radio frequency, intermediate frequency or base band, realize, say, can adopt single-stage control structure or Multistage Control structure from control structure.

The method of realization automatic gain control has multiple, but its core mainly is following several: feed-forward type AGC, feedback-type AGC, feedforward feedback cascade connection type AGC and feedforward feedback mixed type AGC.In general, have only independent feed-forward type AGC or feedback-type AGC, to the adjustment or more coarse of gain, perhaps tracking velocity is slow, thereby causes the fluctuation range of AGC output signal still bigger.Therefore, to the stability requirement of signal level than higher occasion, generally all can adopt feedback AGC and feed-forward AGC that output signal is adjusted simultaneously, promptly adopt feedforward feedback cascade connection type AGC or feedforward feedback mixed type AGC usually.

So-called feedforward feedback cascade connection type AGC is meant signal independently output again after feedforward and the adjustment of feedback two-stage respectively.Because the type has increased the one-level adjustment, performance can be adjusted better than only using one-level.So-called feedforward feedback mixed type AGC adopts feedforward and feedback information that output signal is adjusted simultaneously, its method of adjustment be with the feedforward and feedback signal according to certain algorithm synthesis together after, control a controllable gain amplifier again input signal adjusted.This structure has been utilized the control information of feedback and feedforward simultaneously, and only adopts a controllable gain amplifier, therefore is widely adopted.

At paper " forward and feedback AGC for fast fading channels ", in, provided the instantiation of feedforward feedback mixed type automatic gain control circuit, the envelope detected module adopts the algorithm that linear weighted function is estimated in this example, and the adder that the control signal computing module then adopts is carried out simple computation.The principle of this example as shown in Figure 1.The 13rd phase Electronicsletters magazine that above-mentioned paper was published June 22, the 1029-1030 page or leaf among volume (Volume) 31, its author is M.Fujii, N.Kawaguchi, M.Nakamura and T.Ohsawa.

Figure 1 shows that the circuit basic structure schematic diagram of the feedforward feedback mixed type automatic gain control of prior art.Its implementation procedure is as follows:

Input signal z (t) postpones T through time delay (Delay) module 101 ₀Obtain delay input signal z (t-T after time ₀); Input signal z (t) sends into feedforward envelope detected module 103 after 102 conversion of mould/number (A/D) modular converter, obtain observable feedforward envelope estimated value v _D1(n); The output signal of feedforward envelope detected module 103 obtains feedfoward control gain signal v after the Filtering Processing of feedforward filter module 104 _F(n), feedforward envelope detected module 103 is T with total processing delay of feedforward filter module 104 ₀

Output signal y (t) sends into feedback envelope detection module 106 after 105 conversion of A/D modular converter, obtain observable feedback envelope estimated value v _D2(n); The output signal of feedback envelope detection module 106 is sent into error calculating module 107, calculates desirable incoming signal level V by error calculating module 107 _RObservable feedback envelope estimated value v with 106 outputs of feedback envelope detection module _D2(n) difference between; This difference obtains feedback control gain signal v after 108 filtering of feedback filter module _B(n).

Control signal computing module 109 is according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n), calculate total gain control signal v _C(n), this total gain control signal v _C(n) after 110 conversion of D/A (D/A) modular converter, obtain analog control voltage v _C(t).

According to analog control voltage v _C(t), and the amplification characteristic of controllable gain amplifier module 111, gain amplifier signal G (v obtained _C(t)), afterwards, with this gain amplifier signal G (v _C(t)) with from the delay input signal z (t-T of Postponement module 101 ₀) multiply each other, obtain output signal y (t), i.e. y (t)=G (v through automatic gain control _C(t)) z (t-T ₀).

In implementation procedure shown in Figure 1, feedforward envelope detected module 103 and feedback envelope detection module 106 all adopt the method for linear weighted function to calculate, and promptly at the 2N+1 of statistics in the time, N is a natural number, and envelope is the linear function of time, and its implementation is as follows:

With current time n is the benchmark moment, makes independent variable x=i-n, and i ∈ [n-N, n+N], x ∈ [N, N], and in the time [n-N, n+N], observable envelope estimated value v _D(n) calculating is as shown in Equation (1):

$v_D\left(n\right)=\frac{1}{(2N+1)}\underset{i=N}{\overset{N}{\mathrm{\Σ}}}\left|z(n+x)\right|---\left(1\right)$

Wherein, | z (n+x) | the envelope INSTANTANEOUS OBSERVATION value of expression (n+x) point, as can be seen, the weight of each INSTANTANEOUS OBSERVATION value all is from formula (1)

The weight that is each instantaneous envelope measured value is identical.

Because what the envelope estimated value was calculated is the voltage of signals value, and the energy value of signal is closely connected with magnitude of voltage, therefore, if v _D(n) the observable Energy Estimation value of expression, then $v_D\left(n\right)=\frac{1}{(2N+1)}\underset{i=-N}{\overset{N}{\mathrm{\Σ}}}{\left|z\right|(n+x)}^2,$ Wherein | z (n+x) | ²The instantaneous energy measured value of expression (n+x) point.Certainly the feedforward envelope detected module 103 of this moment and feedback envelope detection module 106 are respectively feedforward energy detection module and feedback energy detection module.The characteristics of this observable Energy Estimation value also are that the weight of each INSTANTANEOUS OBSERVATION value is all identical, promptly all are

In implementation procedure shown in Figure 1, control signal computing module 109 is simple adder, its computational process as shown in Equation (2):

v _C(n)＝v _B(n)+v _F(n) (2)

From above-mentioned implementation procedure as can be seen, in existing feedforward feedback mixed type automatic gain control, no matter be observable envelope estimated value or observable Energy Estimation value, all be to adopt the method for linear function to calculate in its computational process, the weight that is to say each INSTANTANEOUS OBSERVATION value all is identical, because these computational methods can not be upgraded weight according to characteristics of signals, thereby the accuracy of this may observe estimated value is lower.In addition, in existing feedforward feedback mixed type automatic gain control, the control signal computing module only is simple adder, and it calculates total gain control signal v _C(n) accuracy is also lower.For these reasons, the entire system performance will be affected.

Summary of the invention

In view of this, an object of the present invention is to obtain may observe estimated value more accurately, another object of the present invention is to obtain total more accurately gain control signal, for this reason, the invention provides two kinds of methods that realize automatic gain control, the device of realizing said method also is provided simultaneously.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of method that realizes automatic gain control, this method may further comprise the steps:

A, use the may observe estimated value that mode that quadratic function fits is calculated input signal and output signal respectively, this may observe estimated value is obtained the feedfoward control gain signal v of input signal respectively after the filtering _F(n) and the feedback control gain signal v of output signal _B(n);

B, according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v _C(n);

C, according to the described total gain control signal v of step b _C(n) obtain gain amplifier signal G (v _C(t)), afterwards, with this gain amplifier signal G (v _C(t)) with the delay input signal z (t-T that handles through delay ₀) multiply each other, obtain output signal y (t) through automatic gain control.

Preferably, described may observe estimated value is observable envelope estimated value, or observable Energy Estimation value.

Preferably, the mode that fits of the described application quadratic function of the step a method of calculating observable envelope estimated value may further comprise the steps:

A1, order treat that timing statistics length is 2N+1, and wherein N is a natural number, and quadratic function is y=a ₀+ a ₁X+a ₂x ², and make independent variable x=i-n, and i ∈ [n-N, n+N], x ∈ [N, N], the INSTANTANEOUS OBSERVATION value y in i ∈ [n-N, the n+N] time _xFor: y _x=| z (i) |=| z (n+x) |, and make the vector of independent variable x be: $\stackrel{\→}{x}=[-N,-N+1,...,N-1,N]$ , INSTANTANEOUS OBSERVATION value y then _xVector be:

$\stackrel{\→}{y}=\left[\right|z(n-N)|,...,|z\left(n\right)|,...,|z(n+N)\left|\right];$

A2, according to step a1 obtains following equation group with 2N+1 to INSTANTANEOUS OBSERVATION value substitution quadratic function: ${\stackrel{\→}{y}}^T=A\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=A\stackrel{\→}{a};$

Wherein, $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]$ Be the vector of quadratic function coefficient, $A=\left[\begin{array}{ccc}1& {\left[\stackrel{\→}{x}\right]}^T& {\left[{\stackrel{\→}{x}}^2\right]}^T\end{array}\right]=\left[\begin{array}{ccc}1& -N& N^2\\ 1& -(N-1)& {(N-1)}^2\\ 1& -(N-2)& {(N-2)}^2\\ .& .& .\\ .& .& .\\ .& .& .\\ 1& N-2& {(N-2)}^2\\ 1& N-1& {(N-1)}^2\\ 1& N& N^2\end{array}\right]$ Matrix for independent variable x;

The least square solution of a3, the described equation group of calculation procedure a2 obtains $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=W{\stackrel{\→}{y}}^T;$ Wherein, W is the generalized inverse of A, $W={\left(A^{T}A\right)}^{-1}{A}^T=\left[\begin{array}{c}{\stackrel{\→}{w}}_1\\ {\stackrel{\→}{w}}_2\\ {\stackrel{\→}{w}}_3\end{array}\right];$

A4, according to the described least square solution of step a3, obtain a ₀Estimated value: $a_0={\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T;$

A5, to a ₀Estimated value carry out the weight coefficient normalizing, obtain the observable envelope estimated value v of n constantly _D(n), $v_D\left(n\right)=\frac{a_0}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{{\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^Y}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,i}\left|z(n+x)\right|}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}.$

Preferably, step b is described according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v _C(n) method is: with feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) carry out simple addition.

Preferably, step b is described according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v _C(n) method is: with feedforward ride gain signal is desired value, by adaptively the feedback control gain signal being weighted, obtains total gain control signal.

Preferably, described is desired value with feedforward ride gain signal, by adaptively the feedback control gain signal being weighted, obtains total gain control signal v _C(n) method may further comprise the steps:

B1, order $\stackrel{\→}{w}\left(n\right)=[w_1\left(n\right),w_2\left(n\right),...,w_M\left(n\right)],$ Wherein, The weighting column vector of expression M rank sef-adapting filter; Order ${\stackrel{\→}{v}}_B=[v_B\left(n\right),v_B(n-1),...,v_B(n-M+1)],$ Wherein,

Represent n point and former continuous N-1 thereof the M that the feedback control gain signal is formed a dimension FEEDBACK CONTROL vector;

B2, negate feedback initial weight vector $\stackrel{\→}{w}\left(0\right)=[\frac{1}{N},\frac{1}{N},...\frac{1}{N}],$ According to this feedback initial weight vector, and the described feedfoward control gain signal of step a v _F(n) and feedback control gain signal v _BWeight vector when (n) calculating n step feedback iteration $\stackrel{\→}{w}\left(n\right),\stackrel{\→}{w}\left(n\right)=\stackrel{\→}{w}(n-1)+2\mathrm{\μ}{\stackrel{\→}{v}}_B\left(n\right)(v_F\left(n\right)-\stackrel{\→}{w}(n-1){{\stackrel{\→}{v}}_B}^T\left(n\right)),$

Wherein, μ is for upgrading the factor, and its value has determined the convergence rate of update algorithm,

Weight vector when being n-1 step feedback iteration;

B3, according to the FEEDBACK CONTROL vector that the iteration weight vector and the n in the described n of step b2 step are ordered, calculate total gain control signal v adaptively _C(n), $v_c\left(n\right)=\stackrel{\→}{w}\left(n\right){{\stackrel{\→}{v}}_B}^T\left(n\right).$

Preferably, the mode that the described application quadratic function of step a fits is calculated respectively before the may observe estimated value of input signal and output signal, and this method further comprises: input signal and output signal are carried out analog-to-digital conversion process respectively; The gain control signal v that the described calculating of step b is total _C(n) afterwards, further comprise: this total gain control signal v _C(n) carry out digital-to-analogue conversion and handle, and then execution in step c.

A kind of method that realizes automatic gain control, this method may further comprise the steps:

A, obtain the feedfoward control gain signal v of input signal respectively _F(n) and the feedback control gain signal v of output signal _B(n);

B, according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v adaptively _C(n);

C, according to the described total gain control signal v of step B _C(n) obtain gain amplifier signal G (v _C(t)), afterwards, with this gain amplifier signal G (v _C(t)) with the delay input signal z (t-T that handles through delay ₀) multiply each other, obtain output signal y (t) through automatic gain control.

Preferably, described may observe estimated value is observable envelope estimated value, or observable Energy Estimation value.

Preferably, the described feedfoward control gain signal v that obtains input signal of steps A _F(n) and the feedback control gain signal v of output signal _B(n) method may further comprise the steps:

A1, make the independent variable x=i-n of linear function, and i ∈ [n-N, n+N], x ∈ [N, N], wherein N is a natural number, then with moment n be benchmark constantly, in the time [n-N, n+N], calculate the observable envelope estimated value v of input signal and output signal respectively _D(n), $v_D\left(n\right)=\frac{1}{(2N+1)}\underset{i=N}{\overset{N}{\mathrm{\Σ}}}\left|z(n+x)\right|,$

Wherein, | z (n+x) | the INSTANTANEOUS OBSERVATION value during for (n+x);

Obtain the feedfoward control gain signal v of input signal after A2, the may observe envelope estimated value difference filtering with steps A 1 described input and output signal _F(n) and the feedback control gain signal v of output signal _B(n).

Preferably, the described calculating input signal of steps A feedfoward control gain signal v _F(n) and the feedback control gain signal v of output signal _B(n) method is: use mode that quadratic function fits and calculate the considerable envelope of input signal and output signal respectively and survey estimated value, this may observe estimated value is obtained the feedfoward control gain signal v of input signal respectively after the filtering _F(n) and the feedback control gain signal v of output signal _B(n).

Preferably, the mode that fits of the described application quadratic function method of calculating observable envelope estimated value may further comprise the steps:

A01, order treat that timing statistics length is 2N+1, and wherein N is a natural number, and quadratic function is y=a ₀+ a ₁X+a ₂x ², and make independent variable x=i-n, and i ∈ [n-N, n+N], x ∈ [N, N], the INSTANTANEOUS OBSERVATION value y in i ∈ [n-N, the n+N] time _xFor: y _x=| z (i) |=| z (n+x) |, and make the vector of independent variable x be: $\stackrel{\→}{x}=[-N,-N+1,...,N-1,N],$ INSTANTANEOUS OBSERVATION value y then _xVector be:

$\stackrel{\→}{y}=\left[\right|z(n-N)|,...,|z\left(n\right)|,...,|z(n+N)\left|\right];$

A02, according to steps A 01 obtains following equation group with 2N+1 to INSTANTANEOUS OBSERVATION value substitution quadratic function: ${\stackrel{\→}{y}}^T=A\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=A\stackrel{\→}{a};$

Wherein, $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]$ Be the vector of quadratic function coefficient, $A=\left[\begin{array}{ccc}1& {\left[\stackrel{\→}{x}\right]}^T& {\left[{\stackrel{\→}{x}}^2\right]}^T\end{array}\right]=\left[\begin{array}{ccc}1& -N& N^2\\ 1& -(N-1)& {(N-1)}^2\\ 1& -(N-2)& {(N-2)}^2\\ .& .& .\\ .& .& .\\ .& .& .\\ 1& N-2& {(N-2)}^2\\ 1& N-1& {(N-1)}^2\\ 1& N& N^2\end{array}\right]$ Matrix for independent variable x;

The least square solution of A03, the described equation group of calculation procedure A02: $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=W{\stackrel{\→}{y}}^T;$ Wherein, W is the generalized inverse of A, $W={\left(A^{T}A\right)}^{-1}{A}^T=\left[\begin{array}{c}{\stackrel{\→}{w}}_1\\ {\stackrel{\→}{w}}_2\\ {\stackrel{\→}{w}}_3\end{array}\right];$

A04, according to steps A 03 described least square solution, obtain a ₀Estimated value: $a_0={\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T;$

A05, to a ₀Estimated value carry out the weight coefficient normalizing, obtain the observable envelope estimated value v of n constantly _D(n), $v_D\left(n\right)=\frac{a_0}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{{\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^Y}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,i}\left|z(n+x)\right|}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}.$

Preferably, step B is described calculates total gain control signal v adaptively _C(n) method is: with feedforward ride gain signal is desired value, by adaptively the feedback control gain signal being weighted, obtains total gain control signal.

Preferably, described is desired value with feedforward ride gain signal, by adaptively the feedback control gain signal being weighted, obtains total gain control signal v _C(n) method may further comprise the steps:

B1, order $\stackrel{\→}{w}\left(n\right)=[w_1\left(n\right),w_2\left(n\right),...,w_M\left(n\right)],$ Wherein,

The weighting column vector of expression M rank sef-adapting filter; Order ${\stackrel{\→}{v}}_B=[v_B\left(n\right),v_B(n-1),...,v_B(n-M+1)],$ Wherein,

Represent n point and former continuous N-1 thereof the M that the feedback control gain signal is formed a dimension FEEDBACK CONTROL vector;

B2, negate feedback initial weight vector $\stackrel{\→}{w}\left(0\right)=[\frac{1}{N},\frac{1}{N},...\frac{1}{N}],$ According to this feedback initial weight vector, and the described feedfoward control gain signal of step a v _F(n) and feedback control gain signal v _BWeight vector when (n) calculating n step feedback iteration

$\stackrel{\→}{w}\left(n\right)=\stackrel{\→}{w}(n-1)+2\mathrm{\μ}{\stackrel{\→}{v}}_B\left(n\right)(v_F\left(n\right)-\stackrel{\→}{w}(n-1){{\stackrel{\→}{v}}_B}^T\left(n\right)),$

Wherein, μ is for upgrading the factor, and its value has determined the convergence rate of update algorithm,

Weight vector when being n-1 step feedback iteration;

B3, according to the FEEDBACK CONTROL vector that the iteration weight vector and the n in the described n of step B2 step are ordered, calculate total gain control signal v adaptively _C(n), $v_c\left(n\right)=\stackrel{\→}{w}\left(n\right){{\stackrel{\→}{v}}_B}^T\left(n\right).$

Preferably, before the described may observe estimated value of obtaining input signal and output signal, this method further comprises: input signal and output signal are carried out analog-to-digital conversion process respectively; The gain control signal v that the described calculating of step B is total _C(n) afterwards, further comprise: this total gain control signal v _C(n) carry out digital-to-analogue conversion and handle, and then execution in step C.

A kind of device of realizing automatic gain control, comprise the time delay module, feedforward filter, feedback filter, the error signal calculation module, control signal computing module and controllable gain amplifier, wherein, the control signal computing module is according to the feedfoward control gain signal from feedforward filter filtering, with feedback control gain signal from feedback filter, calculate total gain control signal, and this total gain control signal sent into controllable gain amplifier, calculate the gain amplifier signal by controllable gain amplifier, and with this gain amplifier signal with multiply each other through the delay input signal of time delay module, acquisition is through the output signal of automatic gain control, this device also comprises secondary weighted detection module of feedforward and the secondary weighted detection module of feedback, wherein

The secondary weighted detection module of described feedforward carries out the quadratic function weighted to input signal, and the may observe estimated value that is obtained is sent into feedforward filter;

The secondary weighted detection module of described feedback, output signal is carried out the quadratic function weighted, the may observe estimated value that is obtained is sent into the error signal calculation module, after calculating error amount between this may observe estimated value and the desired value by the error signal calculation module, this error amount is sent into feedback filter.

Preferably, described control signal computing module is the adder that is used for feedfoward control gain signal and feedback control gain signal plus.

Preferably, described control signal computing module is desired value for what adopt least-mean-square error algorithm with feedforward ride gain signal, adaptively the auto-adaptive controling signal computing module that the feedback control gain signal is weighted.

Preferably, described secondary weighted detection module is a quadratic function weighting envelope detected module; Perhaps, described secondary weighted detection module is secondary weighted energy detection module.

Preferably, this device also comprises feedforward analog-to-digital conversion module, feedback modulus modular converter and D/A converter module, wherein,

Described feedforward analog-to-digital conversion module carries out input signal to send into the secondary weighted detection module of feedforward after the analog-to-digital conversion, described feedback modulus modular converter carries out input signal to send into the secondary weighted detection module of feedback after the analog-to-digital conversion, and described D/A converter module will be carried out sending into controllable gain amplifier after the digital-to-analogue conversion from the signal of control calculated signals module.

A kind of device of realizing automatic gain control, comprise the time delay module, feedforward filter, feedback filter, the error signal calculation module, the feed forward detection module, feedback detection module, the error signal calculation module, and controllable gain amplifier, wherein, the feed forward detection module is sent into feedforward filter after input signal is converted to the may observe estimated value, feedback detection module is sent into the error signal calculation module after output signal is converted to the may observe estimated value, after calculating error amount between this feedback may observe estimated value and the desired value by the error signal calculation module, this error amount is sent into feedback filter, this device also comprises the auto-adaptive controling signal computing module

Described auto-adaptive controling signal computing module is according to the feedfoward control gain signal from feedforward filter, with feedback control gain signal from feedback filter, adopt least-mean-square error algorithm to calculate total gain control signal, and this total gain control signal sent into controllable gain amplifier, calculate the gain amplifier signal by controllable gain amplifier, and with this gain amplifier signal with multiply each other through the delay input signal of time delay module, obtain output signal through automatic gain control.

Preferably, described auto-adaptive controling signal computing module is desired value for what adopt least-mean-square error algorithm with feedforward ride gain signal, adaptively the auto-adaptive controling signal computing module that the feedback control gain signal is weighted.

Preferably, this device also comprises feedforward analog-to-digital conversion module, feedback modulus modular converter and D/A converter module, wherein,

Described feedforward analog-to-digital conversion module carries out input signal to send into the feed forward detection module after the analog-to-digital conversion, described feedback modulus modular converter carries out output signal to send into feedback detection module after the analog-to-digital conversion, and described D/A converter module will carry out from the signal of auto-adaptive controling signal computing module sending into controllable gain amplifier after the digital-to-analogue conversion.

Preferably, described feed forward detection module is feedforward envelope detected module or feedforward energy detection module, and described feedback detection module is feedback envelope detection module or feedback energy detection module.

Preferably, described feedforward envelope detected module fits the secondary weighted envelope detected module of feedforward of computing for the execution quadratic function, or carries out the feedforward envelope detected module of linear weighted function computing; Described feedforward energy detection module fits the secondary weighted energy detection module of feedforward of computing for the execution quadratic function, or carries out the feedforward energy detection module of linear weighted function computing;

Described feedback envelope detection module fits the secondary weighted envelope detected module of feedback of computing for the execution quadratic function, or carries out the feedback envelope weight detection module of linear weighted function computing; Described feedback energy detection module fits the secondary weighted energy detection module of feedback of computing for the execution quadratic function, or carries out the feedback energy detection module of linear weighted function computing.

The present invention uses the mode that quadratic function fits and obtains the may observe estimated value, use the method for least mean-square error and calculate total gain control signal, improved the accuracy of observable estimated value and total gain control signal, thereby the system that makes is all making moderate progress than prior art aspect the signal to noise ratio of capture time, tracking performance and signal, and then has improved the overall performance of system.

Description of drawings

Figure 1 shows that the circuit basic structure schematic diagram of the realization feedforward feedback mixed type automatic gain control of prior art;

Figure 2 shows that the circuit basic structure schematic diagram of using realization feedforward feedback mixed type automatic gain control of the present invention;

Fig. 3 a is depicted as the step response waveform figure of prior art;

Fig. 3 b is depicted as step response waveform figure of the present invention;

Fig. 4 a is depicted as the envelope diagram of input time varying signal;

Fig. 4 b is the tracking oscillogram to time varying signal of prior art output;

Fig. 4 c is for using the tracking oscillogram to time varying signal of the present invention's output;

Fig. 5 a is depicted as quadriphase PSK (QPSK) the demodulation planisphere of prior art;

Fig. 5 b is depicted as and uses QPSK demodulation planisphere of the present invention.

Embodiment

Below in conjunction with accompanying drawing the present invention is done detailed description further again.

Thinking of the present invention is: the mode that the application quadratic function fits is obtained observable estimated value, use the method for least mean-square error and calculate total gain control signal, thereby improve the accuracy of observable estimated value and total gain control signal, and then improve the overall performance of system.

Figure 2 shows that the circuit basic structure schematic diagram of using realization feedforward feedback mixed type automatic gain control of the present invention.In the present embodiment, be that example is specifically described with observable envelope estimated value.

Input signal z (t) postpones T through Delay module 101 ₀Obtain delay input signal z (t-T after time ₀); Input signal z (t) sends into the secondary weighted envelope detected module 203 of feedforward after 102 conversion of A/D modular converter, obtain observable feedforward envelope estimated value v _D1(n); The output signal of the secondary weighted envelope detected module 203 that feedovers obtains feedfoward control gain signal v after feedforward filter module 104 Filtering Processing _F(n), feedover total processing delay of secondary weighted envelope detected module 203 and feedforward filter module 104 is T ₀

Output signal y (t) sends into the secondary weighted envelope detected module 206 of feedback after 105 conversion of A/D modular converter, obtain observable feedback envelope estimated value v _D2(n); Feed back the output signal of secondary weighted envelope detected module 206 and send into error calculating module 107, error calculating module 107 calculates desirable incoming signal level V _RObservable feedback envelope estimated value v with secondary weighted envelope detected module 206 outputs of feedback _D2(n) difference between; This difference obtains feedback control gain signal v after 108 filtering of feedback filter module _B(n).

Auto-adaptive controling signal computing module 209 is according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n), calculate total gain control signal v _C(n), this total gain control signal v _C(n) after 110 conversion of D/A modular converter, obtain analog control voltage v _C(t).

According to analog control voltage v _C(t), and the amplification characteristic of controllable gain amplifier module 111, gain amplifier signal G (v obtained _C(t)), afterwards, with this gain amplifier signal G (v _C(t)) with from the delay input signal z (t-T of Postponement module 101 ₀) multiply each other, obtain output signal y (t), i.e. y (t)=G (v through automatic gain control _C(t)) z (t-T ₀).

In the above-described embodiments, the secondary weighted envelope detected module 203 that feedovers and the secondary weighted envelope detected module 206 of feedback are that the mode of utilizing quadratic function to fit the instantaneous value in a period of time is obtained the may observe estimated value.Because the computational methods that the secondary weighted envelope detected module 206 of secondary weighted envelope detected module 203 of feedforward and feedback realizes are identical, therefore, be example only below, specify the implementation of utilizing quadratic function to fit with the secondary weighted envelope detected module 203 that feedovers.

The time span of as if statistics is 2N+1, and wherein N is a natural number, and quadratic function is y=a ₀+ a ₁X+a ₂x ², wherein, a ₀, a ₁And a ₂Be respectively the coefficient of zero degree side, first power and the quadratic power of quadratic function.With current time n is the benchmark moment, in the time [n-N, n+N], utilizes quadratic function y=a ₀+ a ₁X+a ₂x ²The 2N+1 that fits in [n-N, the n+N] time is as follows to the concrete grammar of INSTANTANEOUS OBSERVATION value:

Make independent variable x=i-n, and i ∈ [n-N, n+N], x ∈ [N, N], the INSTANTANEOUS OBSERVATION value y in i ∈ [n-N, the n+N] time _xFor: y _x=| z (i) |=| z (n+x) |.

The vector of definition independent variable x is: $\stackrel{\→}{x}=[-N,-N+1,...,N-1,N],$ INSTANTANEOUS OBSERVATION value y then _xVector be: $\stackrel{\→}{y}=\left[\right|z(n-N)|,...,|z\left(n\right)|,...,|z(n+N)\left|\right].$

According to above-mentioned definition, the quadratic function fitting problems is to utilize 2N+1 that the INSTANTANEOUS OBSERVATION value is estimated the coefficient of quadratic function 2N+1 to INSTANTANEOUS OBSERVATION value substitution quadratic function, can be obtained 2N+1 equation, is write all equations as following equation group:

${\stackrel{\→}{y}}^T=A\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=A\stackrel{\→}{a}---\left(3\right)$

Wherein, in the formula (3) $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]$ Be the vector of quadratic function coefficient, A is the matrix of independent variable x, the form of A as the formula (4):

$A=\left[\begin{array}{ccc}1& {\left[\stackrel{\→}{x}\right]}^T& {\left[{\stackrel{\→}{x}}^2\right]}^T\end{array}\right]=\left[\begin{array}{ccc}1& -N& N^2\\ 1& -(N-1)& {(N-1)}^2\\ 1& -(N-2)& {(N-2)}^2\\ .& .& .\\ .& .& .\\ .& .& .\\ 1& N-2& {(N-2)}^2\\ 1& N-1& {(N-1)}^2\\ 1& N& N^2\end{array}\right]---\left(4\right)$

Formula (3) is asked least square solution:

$\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=W{\stackrel{\→}{y}}^T---\left(5\right)$

Wherein, the W in the formula (5) is the generalized inverse of A, and its form is suc as formula (6):

$W={\left(A^{T}A\right)}^{-1}{A}^T=\left[\begin{array}{c}{\stackrel{\→}{w}}_1\\ {\stackrel{\→}{w}}_2\\ {\stackrel{\→}{w}}_3\end{array}\right]---\left(6\right)$

From formula (4), (5) and (6) as can be seen, after N was given, matrix A and W determined thereupon.Since present embodiment be the quadratic function match value of correspondence during with x=0 as the envelope estimated value, and the quadratic function match value of correspondence is a during x=0 ₀Can obtain a according to formula (5) and formula (6) ₀Estimated value:

$a_0={\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T---\left(7\right)$

To a ₀Carry out the weight coefficient normalizing, obtain the envelope estimated value v of n correspondence constantly _D(n), its form is as the formula (8):

$v_D\left(n\right)=\frac{a_0}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{{\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^Y}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,i}\left|z(n+x)\right|}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}---\left(8\right)$

From formula (8) as can be seen, present embodiment adopts the mode of quadratic function match to realize calculating observable envelope estimated value, and the formula (8) that is used to calculate may observe envelope estimated value is the simple form of a weighted sum.With respect to existing by asking average mode to calculate the method for may observe envelope estimated value, though increased amount of calculation seldom, but make the weight of each INSTANTANEOUS OBSERVATION value to upgrade, thereby increased substantially the precision of observable envelope estimated value according to characteristics of signals.

Above-mentioned is to be example with observable envelope estimated value, detailed description utilizes quadratic function to fit the method for estimated value, if for observable Energy Estimation value, correspondingly, secondary weighted envelope detected module 203 of feedforward among Fig. 2 and the secondary weighted envelope detected module 206 of feedback replace with the secondary weighted energy detection module of feedforward respectively and the secondary weighted energy detection module of feedback gets final product, it utilizes the realization principle of the mode that fits of mode that quadratic function fits and may observe envelope estimated value identical, only need will | z (n+x) | replace with | z (n) | ², at this repeated description no longer.

Specify the implementation of auto-adaptive controling signal computing module below.This module adopts the algorithm of least mean-square error (LMS, Least Mean Square) to calculate, and its main thought is by minimizing a cost function, the renewal that draws the weight vector of being asked.Its specific implementation process is as follows:

Make v _F(n) represent the feedfoward control gain signal that n is ordered, order

The weighting column vector of expression M rank sef-adapting filter, then

As the formula (9):

$\stackrel{\→}{w}\left(n\right)=[w_1\left(n\right),w_2\left(n\right),...,w_M\left(n\right)]---\left(9\right)$

Make v _B(n) represent the feedback control gain signal that n is ordered,

Represent n point and former continuous N-1 thereof the M that the feedback control gain signal is formed a dimension FEEDBACK CONTROL vector,

Form as the formula (10):

${\stackrel{\→}{v}}_B=[v_B\left(n\right),v_B(n-1),...,v_B(n-M+1)]---\left(10\right)$

The initial weight vector of negate feedback ride gain signal It is as the formula (11):

$\stackrel{\→}{w}\left(0\right)=[\frac{1}{N},\frac{1}{N},...\frac{1}{N}]---\left(11\right)$

According to above-mentioned feedback initial weight vector, and feedfoward control gain signal v _F(n) and feedback control gain signal v _BWeight vector when (n), the n of employing LMS algorithm computation feedback control gain signal goes on foot iteration

It is as the formula (12):

$\stackrel{\→}{w}\left(n\right),\stackrel{\→}{w}\left(n\right)=\stackrel{\→}{w}(n-1)+2\mathrm{\μ}{\stackrel{\→}{v}}_B\left(n\right)(v_F\left(n\right)-\stackrel{\→}{w}(n-1){{\stackrel{\→}{v}}_B}^T\left(n\right))---\left(12\right)$

Wherein, the parameter μ in the formula (12) is for upgrading the factor, and its value has determined the convergence rate of update algorithm,

Weight vector when being n-1 step iteration;

The feedback control gain signal that expression utilizes the weight coefficient in n-1 step to obtain; The control signal of the output that the weight coefficient that expression utilizes n-1 to go on foot obtains and the error signal between the desired value, this error signal are exactly the target of adjusting, and promptly need minimized cost function;

Gradient direction among the expression LMS;

Coefficient update gradient among the expression LMS.

The FEEDBACK CONTROL vector of order according to the iteration weight vector and the n in n step, total gain control signal that must be to the end, its as the formula (13):

$v_c\left(n\right)=\stackrel{\→}{w}\left(n\right){{\stackrel{\→}{v}}_B}^T\left(n\right)---\left(13\right)$

Contrast formula (13) and formula (2) as can be known, the auto-adaptive controling signal computing module in the present embodiment is to be desired value with feedforward ride gain signal, adaptive the feedback control gain signal is weighted, and obtains total gain control signal v _C(n).Thereby obtained total more accurately gain control signal, improved systematic function.

In order to further specify effect of the present invention, below the simulation result of the present invention and prior art is done one relatively.

Fig. 3 is step response waveform figure, and it is mainly used in the acquisition performance of checking AGC.Wherein Fig. 3 a is the step response waveform figure of prior art scheme, and Fig. 3 b is the step response waveform figure of technical solution of the present invention.As can be known, step of the present invention is rung the time should be shorter than the step response time of prior art behind comparison diagram 3a and Fig. 3 b, and response wave shape has entered stable state faster, and momentary fluctuation is smoothly a lot of more than prior art.

Fig. 4 is the tracking characteristics to time varying signal.Wherein, Fig. 4 a is the envelope diagram of input time varying signal, and Fig. 4 b is the tracking oscillogram to time varying signal of prior art output, and Fig. 4 c is for using the tracking oscillogram to time varying signal of the present invention's output.Comparison diagram 4b and Fig. 4 c as can be known, adopt the present invention after, the waveform of output signal is steady more a lot of than prior art.

Figure 5 shows that quadriphase PSK (QPSK) demodulation planisphere.It is mainly used in the influence of description to Signal-to-Noise.Wherein, Fig. 5 a is the QPSK demodulation planisphere of prior art, and Fig. 5 b is for using QPSK demodulation planisphere of the present invention.Comparison diagram 5a and Fig. 5 b as can be known, the QPSK restituted signal in the QPSK demodulation planisphere of the present invention is more concentrated, therefore, the signal to noise ratio of QPSK restituted signal of the present invention is higher.

As seen comprehensive above-mentioned simulation result is used the present invention and is all being made moderate progress than prior art aspect the signal to noise ratio of capture time, tracking performance and signal.

The above only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (25)

1, a kind of method that realizes automatic gain control is characterized in that this method may further comprise the steps:

A, use the may observe estimated value that mode that quadratic function fits is calculated input signal and output signal respectively, this may observe estimated value is obtained the feedfoward control gain signal v of input signal respectively after the filtering _F(n) and the feedback control gain signal v of output signal _B(n);

B, according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v _C(n);

C, according to the described total gain control signal v of step b _C(n) obtain gain amplifier signal G (v _C(t)), afterwards, with this gain amplifier signal G (v _C(t)) with the delay input signal z (t-T that handles through delay ₀) multiply each other, obtain output signal y (t) through automatic gain control.

2, method according to claim 1 is characterized in that, described may observe estimated value is observable envelope estimated value, or observable Energy Estimation value.

3, method according to claim 2 is characterized in that, the method that the mode that the described application quadratic function of step a fits is calculated observable envelope estimated value may further comprise the steps:

A1, order treat that timing statistics length is 2N+1, and wherein N is a natural number, and quadratic function is y=a ₀+ a ₁X+a ₂x ², and make independent variable x=i-n, and i ∈ [n-N, n+N], x ∈ [N, N], the INSTANTANEOUS OBSERVATION value y in i ∈ [n-N, the n+N] time _xFor: y _x=| z (i)=| z (n+x) |, and make the vector of independent variable x be: $\stackrel{\→}{x}=[-N,-N+1,\·\·\·,N-1,N],$ INSTANTANEOUS OBSERVATION value y then _xVector be: $\stackrel{\→}{y}=\left[\right|z(n-N)|,\·\·\·,|z\left(n\right)|,\·\·\·,|z(n+N)\left|\right];$

A2, according to step a1 obtains following equation group with 2N+1 to INSTANTANEOUS OBSERVATION value substitution quadratic function: ${\stackrel{\→}{y}}^T=A\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=A\stackrel{\→}{a};$

Wherein, $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]$ Be the vector of quadratic function coefficient, $A=\left[\begin{array}{ccc}1& {\left[\stackrel{\→}{x}\right]}^T& {\left[{\stackrel{\→}{x}}^2\right]}^T\end{array}\right]=\left[\begin{array}{ccc}1& -N& N^2\\ 1& -(N-1)& {(N-1)}^2\\ 1& -(N-2)& {(N-2)}^2\\ .& .& .\\ .& .& .\\ .& .& .\\ 1& N-2& {(N-2)}^2\\ 1& N-1& {(N-1)}^2\\ 1& N& N^2\end{array}\right]$ Matrix for independent variable x;

The least square solution of a3, the described equation group of calculation procedure a2 obtains $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=W{\stackrel{\→}{y}}^T;$

Wherein, W is the generalized inverse of A, $W={\left(A^{T}A\right)}^{-1}{A}^T=\left[\begin{array}{c}{\stackrel{\→}{w}}_1\\ {\stackrel{\→}{w}}_2\\ {\stackrel{\→}{w}}_3\end{array}\right];$

A4, according to the described least square solution of step a3, obtain a ₀Estimated value: $a_0={\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T;$

A5, to a ₀Estimated value carry out the weight coefficient normalizing, obtain the observable envelope estimated value v of n constantly _D(n), $v_D\left(n\right)=\frac{a_0}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{{\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,i}\left|z(n+x)\right|}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}.$

4, method according to claim 1 is characterized in that, step b is described according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v _C(n) method is: with feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) carry out simple addition.

5, method according to claim 1 is characterized in that, step b is described according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v _C(n) method is: with feedforward ride gain signal is desired value, by adaptively the feedback control gain signal being weighted, obtains total gain control signal.

6, method according to claim 5 is characterized in that, described is desired value with feedforward ride gain signal, by adaptively the feedback control gain signal being weighted, obtains total gain control signal v _C(n) method may further comprise the steps:

B1, order $\stackrel{\→}{w}\left(n\right)=[w_1\left(n\right),w_2\left(n\right),...,w_M\left(n\right)],$ Wherein,

The weighting column vector of expression M rank sef-adapting filter; Order ${\stackrel{\→}{v}}_B\left(n\right)=[v_B\left(n\right),v_B(n-1),\·\·\·,v_B(n-M+1)],$ Wherein, Represent n point and former continuous N-1 thereof the M that the feedback control gain signal is formed a dimension FEEDBACK CONTROL vector;

B2, negate feedback initial weight vector $\stackrel{\→}{w}\left(0\right)=[\frac{1}{N},\frac{1}{N},\·\·\·\frac{1}{N}],$ According to this feedback initial weight vector, and the described feedfoward control gain signal of step a v _F(n) and feedback control gain signal v _BWeight vector when (n) calculating n step feedback iteration

$\stackrel{\→}{w}\left(n\right)=\stackrel{\→}{w}(n-1)+2\mathrm{\μ}{\stackrel{\→}{v}}_B\left(n\right)(v_F\left(n\right)-\stackrel{\→}{w}(n-1){{\stackrel{\→}{v}}_B}^T\left(n\right)),$

Wherein, μ is for upgrading the factor, and its value has determined the convergence rate of update algorithm,

Weight vector when being n-1 step feedback iteration;

B3, according to the FEEDBACK CONTROL vector that the iteration weight vector and the n in the described n of step b2 step are ordered, calculate total gain control signal v adaptively _C(n), $v_c\left(n\right)=\stackrel{\→}{w}\left(n\right){\stackrel{\→}{v_B}}^T\left(n\right).$

7, method according to claim 1 is characterized in that,

The mode that the described application quadratic function of step a fits is calculated respectively before the may observe estimated value of input signal and output signal, and this method further comprises: input signal and output signal are carried out analog-to-digital conversion process respectively;

The gain control signal v that the described calculating of step b is total _C(n) afterwards, further comprise: this total gain control signal v _C(n) carry out digital-to-analogue conversion and handle, and then execution in step c.

8, a kind of method that realizes automatic gain control is characterized in that this method may further comprise the steps:

A, obtain the feedfoward control gain signal v of input signal respectively _F(n) and the feedback control gain signal v of output signal _B(n);

B, according to feedfoward control gain signal v _F(n) and feedback control gain signal v _B(n) calculate total gain control signal v adaptively _C(n);

C, according to the described total gain control signal v of step B _C(n) obtain gain amplifier signal G (v _C(t)), afterwards, with this gain amplifier signal G (v _C(t)) with the delay input signal z (t-T that handles through delay ₀) multiply each other, obtain output signal y (t) through automatic gain control.

9, method according to claim 8 is characterized in that, described may observe estimated value is observable envelope estimated value, or observable Energy Estimation value.

10, method according to claim 9 is characterized in that, the described feedfoward control gain signal v that obtains input signal of steps A _F(n) and the feedback control gain signal v of output signal _E(n) method may further comprise the steps:

A1, make the independent variable x=i-n of linear function, and i ∈ [n-N, n+N], x ∈ [N, N], wherein N is a natural number, then with moment n be benchmark constantly, in the time [n-N, n+N], calculate the observable envelope estimated value v of input signal and output signal respectively _D(n), $v_D\left(n\right)=\frac{1}{(2N+1)}\underset{i=N}{\overset{N}{\mathrm{\Σ}}}\left|z(n+x)\right|,$

Wherein, | z (n+x) | the INSTANTANEOUS OBSERVATION value during for (n+x);

Obtain the feedfoward control gain signal v of input signal after A2, the may observe envelope estimated value difference filtering with steps A 1 described input and output signal _F(n) and the feedback control gain signal v of output signal _B(n).

11, method according to claim 9 is characterized in that, the described calculating input signal of steps A feedfoward control gain signal v _F(n) and the feedback control gain signal v of output signal _B(n) method is: use mode that quadratic function fits and calculate the considerable envelope of input signal and output signal respectively and survey estimated value, this may observe estimated value is obtained the feedfoward control gain signal v of input signal respectively after the filtering _F(n) and the feedback control gain signal v of output signal _B(n).

12, method according to claim 11 is characterized in that, the method that the mode that described application quadratic function fits is calculated observable envelope estimated value may further comprise the steps:

A01, order treat that timing statistics length is 2N+1, and wherein N is a natural number, and quadratic function is y=a ₀+ a ₁X+a ₂x ², and make independent variable x=i-n, and i ∈ [n-N, n+N], x ∈ [N, N], the INSTANTANEOUS OBSERVATION value y in i ∈ [n-N, the n+N] time _xFor: y _x=| z (i) |=| z (n+x) |, and make the vector of independent variable x be: $\stackrel{\→}{x}=[-N,-N+1,\·\·\·,N-1,N],$ INSTANTANEOUS OBSERVATION value y then _xVector be: $\stackrel{\→}{y}=\left[\right|z(n-N)|,\·\·\·,|z\left(n\right)|,\·\·\·,|z(n+N)\left|\right];$

A02, according to steps A 01 obtains following equation group with 2N+1 to INSTANTANEOUS OBSERVATION value substitution quadratic function: ${\stackrel{\→}{y}}^T=A\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=A\stackrel{\→}{a};$

Wherein, $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]$ Be the vector of quadratic function coefficient, $A=\left[\begin{array}{ccc}1& {\left[\stackrel{\→}{x}\right]}^T& {\left[{\stackrel{\→}{x}}^2\right]}^T\end{array}\right]=\left[\begin{array}{ccc}1& -N& N^2\\ 1& -(N-1)& {(N-1)}^2\\ 1& -(N-2)& {(N-2)}^2\\ .& .& .\\ .& .& .\\ .& .& .\\ 1& N-2& {(N-2)}^2\\ 1& N-1& {(N-1)}^2\\ 1& N& N^2\end{array}\right]$ Matrix for independent variable x;

The least square solution of A03, the described equation group of calculation procedure A02: $\stackrel{\→}{a}=\left[\begin{array}{c}{a}_0\\ a_1\\ a_2\end{array}\right]=W{\stackrel{\→}{y}}^T;$

Wherein, W is the generalized inverse of A, $W={\left(A^{T}A\right)}^{-1}{A}^T=\left[\begin{array}{c}{\stackrel{\→}{w}}_1\\ {\stackrel{\→}{w}}_2\\ {\stackrel{\→}{w}}_3\end{array}\right];$

A04, according to steps A 03 described least square solution, obtain a ₀Estimated value: $a_0={\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T;$

A05, to a ₀Estimated value carry out the weight coefficient normalizing, obtain the observable envelope estimated value v of n constantly _D(n), $v_D\left(n\right)=\frac{a_0}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{{\stackrel{\→}{w}}_1{\stackrel{\→}{y}}^T}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}=\frac{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,i}\left|z(n+x)\right|}{\underset{x=-N}{\overset{N}{\mathrm{\Σ}}}{w}_{1,x}}.$

According to claim 10 or 11 described methods, it is characterized in that 13, step B is described to calculate total gain control signal v adaptively _C(n) method is: with feedforward ride gain signal is desired value, by adaptively the feedback control gain signal being weighted, obtains total gain control signal.

14, method according to claim 13 is characterized in that, described is desired value with feedforward ride gain signal, by adaptively the feedback control gain signal being weighted, obtains total gain control signal v _C(n) method may further comprise the steps:

B1, order $\stackrel{\→}{w}\left(n\right)=[w_1\left(n\right),w_2\left(n\right),...,w_M\left(n\right)],$ Wherein,

The weighting column vector of expression M rank sef-adapting filter; Order ${\stackrel{\→}{v}}_B\left(n\right)=[v_B\left(n\right),v_B(n-1),\·\·\·,v_B(n-M+1)],$ Wherein,

Represent n point and former continuous N-1 thereof the M that the feedback control gain signal is formed a dimension FEEDBACK CONTROL vector;

B2, negate feedback initial weight vector $\stackrel{\→}{w}\left(0\right)=[\frac{1}{N},\frac{1}{N},\·\·\·\frac{1}{N}],$ According to this feedback initial weight vector, and the described feedfoward control gain signal of step a v _F(n) and feedback control gain signal v _BWeight vector when (n) calculating n step feedback iteration

$\stackrel{\→}{w}\left(n\right)=\stackrel{\→}{w}(n-1)+2\mathrm{\μ}{\stackrel{\→}{v}}_B\left(n\right)(v_F\left(n\right)-\stackrel{\→}{w}(n-1){{\stackrel{\→}{v}}_B}^T\left(n\right)),$

Wherein, μ is for upgrading the factor, and its value has determined the convergence rate of update algorithm,

Weight vector when being n-1 step feedback iteration;

B3, according to the FEEDBACK CONTROL vector that the iteration weight vector and the n in the described n of step B2 step are ordered, calculate total gain control signal v adaptively _C(n), $v_c\left(n\right)=\stackrel{\→}{w}\left(n\right){\stackrel{\→}{v_B}}^T\left(n\right).$

15, according to claim 10 or 11 described methods, it is characterized in that,

Before the described may observe estimated value of obtaining input signal and output signal, this method further comprises: input signal and output signal are carried out analog-to-digital conversion process respectively;

The gain control signal v that the described calculating of step B is total _C(n) afterwards, further comprise: this total gain control signal v _C(n) carry out digital-to-analogue conversion and handle, and then execution in step C.

16, a kind of device of realizing automatic gain control, comprise the time delay module, feedforward filter, feedback filter, the error signal calculation module, control signal computing module and controllable gain amplifier, wherein, the control signal computing module is according to the feedfoward control gain signal from feedforward filter filtering, with feedback control gain signal from feedback filter, calculate total gain control signal, and this total gain control signal sent into controllable gain amplifier, calculate the gain amplifier signal by controllable gain amplifier, and with this gain amplifier signal with multiply each other through the delay input signal of time delay module, acquisition is through the output signal of automatic gain control, it is characterized in that, this device also comprises secondary weighted detection module of feedforward and the secondary weighted detection module of feedback, wherein

The secondary weighted detection module of described feedforward carries out the quadratic function weighted to input signal, and the may observe estimated value that is obtained is sent into feedforward filter;

The secondary weighted detection module of described feedback, output signal is carried out the quadratic function weighted, the may observe estimated value that is obtained is sent into the error signal calculation module, after calculating error amount between this may observe estimated value and the desired value by the error signal calculation module, this error amount is sent into feedback filter.

17, device according to claim 16 is characterized in that, described control signal computing module is the adder that is used for feedfoward control gain signal and feedback control gain signal plus.

18, device according to claim 16, it is characterized in that, described control signal computing module is desired value for what adopt least-mean-square error algorithm with feedforward ride gain signal, adaptively the auto-adaptive controling signal computing module that the feedback control gain signal is weighted.

19, device according to claim 16 is characterized in that, described secondary weighted detection module is a quadratic function weighting envelope detected module; Perhaps, described secondary weighted detection module is secondary weighted energy detection module.

According to the arbitrary described device of claim 16 to 19, it is characterized in that 20, this device also comprises feedforward analog-to-digital conversion module, feedback modulus modular converter and D/A converter module, wherein,

Described feedforward analog-to-digital conversion module carries out input signal to send into the secondary weighted detection module of feedforward after the analog-to-digital conversion, described feedback modulus modular converter carries out input signal to send into the secondary weighted detection module of feedback after the analog-to-digital conversion, and described D/A converter module will be carried out sending into controllable gain amplifier after the digital-to-analogue conversion from the signal of control calculated signals module.

21, a kind of device of realizing automatic gain control, comprise the time delay module, feedforward filter, feedback filter, the error signal calculation module, the feed forward detection module, feedback detection module, the error signal calculation module, and controllable gain amplifier, wherein, the feed forward detection module is sent into feedforward filter after input signal is converted to the may observe estimated value, feedback detection module is sent into the error signal calculation module after output signal is converted to the may observe estimated value, after calculating error amount between this feedback may observe estimated value and the desired value by the error signal calculation module, this error amount is sent into feedback filter, it is characterized in that, this device also comprises the auto-adaptive controling signal computing module

Described auto-adaptive controling signal computing module is according to the feedfoward control gain signal from feedforward filter, with feedback control gain signal from feedback filter, adopt least-mean-square error algorithm to calculate total gain control signal, and this total gain control signal sent into controllable gain amplifier, calculate the gain amplifier signal by controllable gain amplifier, and with this gain amplifier signal with multiply each other through the delay input signal of time delay module, obtain output signal through automatic gain control.

22, device according to claim 21, it is characterized in that, described auto-adaptive controling signal computing module is desired value for what adopt least-mean-square error algorithm with feedforward ride gain signal, adaptively the auto-adaptive controling signal computing module that the feedback control gain signal is weighted.

23, device according to claim 22 is characterized in that, this device also comprises feedforward analog-to-digital conversion module, feedback modulus modular converter and D/A converter module, wherein,

Described feedforward analog-to-digital conversion module carries out input signal to send into the feed forward detection module after the analog-to-digital conversion, described feedback modulus modular converter carries out output signal to send into feedback detection module after the analog-to-digital conversion, and described D/A converter module will carry out from the signal of auto-adaptive controling signal computing module sending into controllable gain amplifier after the digital-to-analogue conversion.

According to claim 21 or 23 described devices, it is characterized in that 24, described feed forward detection module is feedforward envelope detected module or feedforward energy detection module, described feedback detection module is feedback envelope detection module or feedback energy detection module.

25, device according to claim 24 is characterized in that, described feedforward envelope detected module fits the secondary weighted envelope detected module of feedforward of computing for the execution quadratic function, or carries out the feedforward envelope detected module of linear weighted function computing; Described feedforward energy detection module fits the secondary weighted energy detection module of feedforward of computing for the execution quadratic function, or carries out the feedforward energy detection module of linear weighted function computing;

Described feedback envelope detection module fits the secondary weighted envelope detected module of feedback of computing for the execution quadratic function, or carries out the feedback envelope weight detection module of linear weighted function computing; Described feedback energy detection module fits the secondary weighted energy detection module of feedback of computing for the execution quadratic function, or carries out the feedback energy detection module of linear weighted function computing.

Images