CN103475382B - A kind of Zero intermediate frequency Gaussian white noise adding method and device - Google Patents

Abstract

The invention discloses a kind of Zero intermediate frequency Gaussian white noise adding method and device, method mainly comprises the following steps: (1) inputs zero intermediate frequency signals and calculates the average energy of signal; (2) method for designing using matrix to transform generates additive white Gaussian noise; (3) what use search approximate algorithm to obtain amplitude factor and then produce specific signal to noise ratio adds zero intermediate frequency signals of making an uproar; (4) zero intermediate frequency signals of making an uproar will be added and be input to digital Auto Gain module, according to the output gain of the Threshold Control Method signal of setting.The present invention compared with prior art has following advantage: the consumption substantially reducing hardware logic unit; Improve the delivery efficiency of white Gaussian noise; What achieve gain controllable adds zero intermediate frequency signals of making an uproar simultaneously; The method can realize modularized design.

Description

A kind of Zero intermediate frequency Gaussian white noise adding method and device

Technical field

The present invention relates to the communications field, particularly a kind of Zero intermediate frequency Gaussian white noise adding method and device.

Background technology

Additive white Gaussian noise is that a kind of each spectrum component obedience is uniformly distributed (i.e. white noise), and the noise signal of whose amplitude obeys Gaussian Profile.Because of its additive property, whose amplitude obeys Gaussian Profile and gaining the name for the one of white noise.

In communication and signal processing system, white Gaussian noise is very common noise signal, and is more useful noise signal, especially for use in the interference free performance of inspection communication system.In radio communication or satellite communication, the Gaussian white noise channel of different signal to noise ratio and gain-variable can be needed to check the interference free performance of receiving equipment according to actual conditions.

Through finding the contrast of existing patent and document, there is following problem in existing white Gaussian noise adding method: one, white Gaussian noise generation unit adopts logical block and shift register to realize in hardware implementing, and the cycle of this method is 2 ^m-1, wherein m is the figure place of shift register, and each clock exports a bit.Suppose to require to export nbit, then need n × m trigger to realize.If cycle request is longer larger to hardware resource consumption; Two, when arranging certain signal to noise ratio and exporting, use multiplier and trigger realization can increase the consumption to hardware resource.This just causes, when batch checks the interference free performance of radio communication or satellite receiving equipment, causing high resource, inefficient consumption.

Therefore, from realizing angle, selecting that hardware implementing is simple, resource consumption is few, efficiency is high and the noise addition methods of signal to noise ratio and gain-variable can be realized and device is significantly.

Summary of the invention

The object of the invention is to overcome the shortcoming of prior art and deficiency, adopt and take the smaller nonequivalence operation generation white Gaussian noise of hardware resource, and in conjunction with searching algorithm and digital Auto Gain, realize a kind of efficient white Gaussian noise adding method.

Another object of the present invention is to, a kind of device of Zero intermediate frequency Gaussian white noise adding method is provided.

In order to reach above-mentioned first goal of the invention, the present invention by the following technical solutions:

A kind of Zero intermediate frequency Gaussian white noise adding method, comprises the steps:

(1) energy power_avg is averaging to the zero intermediate frequency signals of input: the energy power first calculating the zero intermediate frequency signals I+jQ of input, is then averaging to obtain averaged signal energy power_avg to the signal energy of a period of time;

(2) produced additive white Gaussian noise α and β of unit energy by white Gaussian noise generation module, and then obtain the multiple noise α+j β that energy is 2;

(3) energy that the averaged signal energy power_avg obtained according to given signal to noise ratio snr and step (1) and step (2) obtain is the multiple noise α+j β of 2, the signal amplitude factors A that search is corresponding and noise amplitude factor B;

Described signal to noise ratio wherein b=(2 ^cnt_shift) ²; a=y ²>1, and a<4; Cnt_shift is the bit number that multiple noise signal α+j β is moved to the left, and namely multiplication factor is 2 ^cnt_shift, then energy multiplication factor is b=(2 ^cnt_shift) ²; Y is signal amplitude multiplication factor, then energy amplifies y ²;

Described signal amplitude factors A=y, noise amplitude factor B=2 ^cnt_shift;

(4) zero intermediate frequency input signal I+jQ is multiplied by signal amplitude factors A that step (3) obtains adjusted after signal SIG; The multiple noise signal α+j β obtained by step (2) be multiplied by noise amplitude B that step (3) obtains adjusted after noise NOISE; Signal SIG and noise NOISE obtains through noise summation module adding the signal SIG_NOISE made an uproar;

(5) digital Auto Gain module DAGC automatically adjusts to the SIG_NOISE that step (4) obtains according to given thresholding R.

In step (2), the implementation method of white Gaussian noise random number is as follows:

(2-1) the uniform random number t on algorithm generation (0,1) of matrixing is used according to Tausworthe method _n;

Described uniform random number t _nwith the random sequence x of L length _nthere is following relational expression:

$t_n=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ns}+i-1}\&CenterDot;2^{-i}$

Wherein, L and s is non-zero positive integer, and L is that random number exports bit wide, and s is saltus step step-length; Obviously, random number t _nwith binary sequence T _n=X _ns=(x _ns, x _ns+1..., x _ns+L-2, x _ns+L-1) one_to_one corresponding, therefore t in the present invention _nwith T _ndo not do special differentiation;

Described random sequence x _nrecurrence Relation as follows:

$x_n=a_1{x}_{n-1}\&CirclePlus;a_2{x}_{n-2}\&CirclePlus;\&CenterDot;\&CenterDot;\&CenterDot;\&CirclePlus;a_k{x}_{n-k}$

Wherein, a ₁, a ₂a _kfor proper polynomial P (z)=z ^k-a ₁z ^k-1-...-a _kcoefficient;

(2-2) the combinedTausworthe random number u of period expansion is produced by J Tausworthe uniform random number _n;

Described combined Tausworthe random number relational expression is:

$u_n=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}(x_{\mathrm{ns}+i-\mathrm{1,1}}\&CirclePlus;x_{\mathrm{ns}+i-\mathrm{1,2}}\&CirclePlus;...\&CirclePlus;x_{\mathrm{ns}+i-1,J})\&CenterDot;2^{-i}$

Wherein, be J independently random number sequence, as J=1, u _n=t _n;

(2-3) according to the random number called after that step (2-2) two combined Tausworthe produce: u _{n, 1}and u _{n, 2}, produce gaussian random distribution number α and β thus; Relational expression is as follows:

$\mathrm{\&alpha;}=\sqrt{-2\&times;\ln \left(u_{n,1}\right)}\sin \left(2\mathrm{\&pi;}{u}_{n,2}\right)$

$\mathrm{\&beta;}=\sqrt{2-\&times;\ln \left(u_{n,1}\right)}\cos \left(2\mathrm{\&pi;}{u}_{n,2}\right)$

In step (2-1), described matrixing algorithm, can produce a binary sequence T within a clock cycle _n, thus improve execution efficiency, concrete shift step is as follows:

(2-1-1) calculate transfer matrix A, expression formula is:

$A=\left[\begin{array}{cc}{C}_{k\&times;1}& I_{(L-1)\&times;(L-1)}\\ 0_{(L-k)\&times;(L-1)}& 0_{1\&times;1}\end{array}\right],$

Wherein, C _{k × 1}=[a ₁a ₂a _k] be the coefficient vector of proper polynomial P (Z), 0 _{(L-K) × (L-1)}, 0 _{1 × 1}for null matrix, I _{(L-1) × (L-1)}for unit matrix;

Described proper polynomial gets trinomial P (Z)=z ^k-z ^q-1;

(2-1-2) the transfer matrix A using step (2-1-1) to obtain and the L position state information X of current time _n=(x _n+L-1, x _n+L-2..., x _n+1, x _n), calculate the L position state information X of subsequent time _n+1, i.e. X _n+1=X _n× A;

(2-1-3) according to relational expression and (2-1) middle random number t of step (2-1-2) _nwith binary sequence T _n=X _nsone-to-one relationship, recursion can obtain following relational expression:

T _n+1=X _(n+1)s=X _(n+1)s-1×A=(X _(n+1)s-2×A)×A=…=X _ns×A ^s=T _n×A ^s

(2-1-4) relational expression obtained according to step (2-1-3) can obtain a new random number t _n+1, its saltus step step-length is s; Random number producing method can use following matrix notation:

T _n+1=T _n×A ^s。

In step (3), described search approximate algorithm comprises the following steps:

(3-1) amplitude factor B is searched for: first make the signal amplitude factor be 1, i.e. a=y ²=1, then cnt_shift increases progressively by 0, until meet till, obtain cnt_shift and B=2 ^cnt_shift;

(3-2) search signal amplitude factor A: make a by 1 by fixing multiple x ²the mode increased progressively searches for a, until meet $\mathrm{snr}\&le;\frac{a\&times;\mathrm{power}\_\mathrm{avg}}{2b},$ Complete and search for $A=y=\sqrt{a};$

Described fixing multiple x is the multiple of amplitude factor y, and its choosing method is: assuming that the quantizing bit number of fixing multiple incremental change x is BITS, the value after taken amount is X=2 ^bITS+ 1, then when getting BITS=11, x=1.00048828125, now energy per increases progressively 0.004dB, and namely energy error is in 0.004dB;

Described choosing method, when BITS is enough large, therefore the incremental change of signal energy can be expressed as $\mathrm{power}\_\mathrm{avg}\&CenterDot;x^2=\mathrm{power}\_\mathrm{avg}+\frac{\mathrm{power}\_\mathrm{avg}}{2^{(\mathrm{BITS}-1)}}.$

Described step (5) is specially:

(5-1) energy balane carrying out signal obtains E _y(n), and by logarithmic transformation, energy value is converted into logarithmic form E _y1(n); Pass through E _y1n () compares with threshold value R's, and the subtraction carrying out logarithm obtains ε (n);

(5-2) amplitude gain regulated value G is obtained through antilogarithm computing;

(5-3) the amplitude gain G in a multiplier and a upper moment was eventually passed _n-1be multiplied to obtain the gain G of current time _n.

In order to reach another object above-mentioned, the present invention by the following technical solutions:

Based on the device of above-mentioned a kind of Zero intermediate frequency Gaussian white noise adding method, comprise zero intermediate frequency signals input module, average energy computing module, amplitude factor search module, additive white Gaussian noise generation module AWGN, noise summation module, digital Auto Gain module DAGC;

Zero intermediate frequency signals input module is used for being connected with average energy computing module and amplitude factor search module, and input signal is obtained through A/D conversion or digital intermediate frequency input by analog signal;

Described average energy computing module is connected with zero intermediate frequency signals input module and amplitude factor search module respectively;

Described amplitude factor search module is connected with zero intermediate frequency signals input module and AWGN module respectively, and the signal to noise ratio snr according to input produces signal amplitude factors A, noise amplitude factor B;

Described AWGN module is connected with amplitude factor search module, and the white Gaussian noise unit energy of generation is 1, and then obtains the multiple noise α+j β that energy is 2;

Described noise summation module is connected with the zero intermediate frequency signals I+jQ after amplitude adjusted, white Gaussian noise signal alpha+j β and DAGC module respectively;

Described DAGC module is connected with noise summation module, export certain gain and zero intermediate frequency signals containing white Gaussian noise.

The present invention has following advantage and effect relative to prior art:

(1) the method can use field programmable gate array (FPGA) to be realized by digital form, and can use as the standalone module of signal source, is convenient to module and transplants;

(2) in the generation module of additive white Gaussian noise, use the method for matrixing within a clock cycle, produce the random number of a L position, and in the hardware configuration of transition matrix, only comprising the exclusive or logic gate of s coordination, the method greatly improves the delivery efficiency of white Gaussian noise;

(3) when searching for amplitude factor, when setting increases progressively multiple, making the mode of calculating signal energy increment value be converted to addition and displacement by multiplication, decreasing the consumption of logical resource;

(4) pipelining is used to realize search approximate algorithm, the search efficiency of the increase rate factor.

Accompanying drawing explanation

Fig. 1 is the composition schematic diagram of the Zero intermediate frequency Gaussian white noise adding method of embodiments of the invention;

The white Gaussian noise α that Fig. 2 (a) produces for embodiments of the invention AWGN module;

The white Gaussian noise β that Fig. 2 (b) produces for embodiments of the invention AWGN module;

Fig. 3 is the streamline schematic diagram of embodiments of the invention search approximate algorithm.

Fig. 4 is embodiments of the invention DAGC structural representation.

Embodiment

Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited thereto.

Embodiment

As shown in Figure 1, the present embodiment comprises: zero intermediate frequency input signal I+jQ, average energy computing module, amplitude factor search module, additive white Gaussian noise generation module (AWGN), noise summation module and digital Auto Gain module (DAGC).

Described zero intermediate frequency input signal I+jQ is connected with average energy computing module and amplitude factor search module respectively, and input signal can be obtained through A/D conversion or digital intermediate frequency input by analog signal;

Described average energy computing module is connected with zero intermediate frequency signals I+jQ and amplitude factor search module respectively;

Described amplitude factor search module is connected with zero intermediate frequency signals I+jQ input signal and AWGN module respectively, and the signal to noise ratio snr according to input produces signal amplitude factors A, noise amplitude factor B;

Described AWGN module is connected with amplitude factor search module, and the white Gaussian noise unit energy of generation is 1, and then obtains the multiple noise α+j β that energy is 2.

Described noise summation module is connected with the zero intermediate frequency signals I+jQ after amplitude adjusted, white Gaussian noise signal alpha+j β and DAGC module respectively;

Described DAGC module is connected with noise summation module, export certain gain and zero intermediate frequency signals containing white Gaussian noise.

A kind of Zero intermediate frequency Gaussian white noise interpolation side of the present invention, comprises the following steps:

(1) energy power_avg is averaging to the zero intermediate frequency signals of input: the energy power first calculating the zero intermediate frequency signals I+jQ of input, is then averaging to obtain averaged signal energy power_avg to the signal energy of a period of time.

(2) produced additive white Gaussian noise α and β of unit energy by white Gaussian noise generation module, and then obtain the multiple noise α+j β that energy is 2.The implementation procedure of white Gaussian noise random number is as follows:

(2-1) the uniform random number t on algorithm generation (0,1) of matrixing is used according to Tausworthe method _n.

Described uniform random number t _nwith the random sequence x of L length _nthere is following relational expression:

$t_n=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ns}+i,1}\&CenterDot;2^{-i}$

Wherein, L and s is non-zero positive integer, and L is that random number exports bit wide, and s is saltus step step-length.Obviously, random number t _nwith binary sequence T _n=X _ns=(x _ns, x _ns+1..., x _ns+L-2, x _ns+L-1) one_to_one corresponding, therefore t in the present invention _nwith T _ndo not do special differentiation.

Described random sequence x _nrecurrence Relation as follows:

$x_n=a_1{x}_{n-1}\&CirclePlus;a_2{x}_{n-2}\&CirclePlus;\&CenterDot;\&CenterDot;\&CenterDot;\&CirclePlus;a_k{x}_{n-k}$

Wherein, a ₁, a ₂a _kfor proper polynomial P (z)=z ^k-a ₁z ^k-1-...-a _kcoefficient.

Described matrixing algorithm, can produce a binary sequence T within a clock cycle _n, thus improve execution efficiency, concrete shift step is as follows:

(2-1-1) calculate transfer matrix A, expression formula is:

$A=\left[\begin{array}{cc}{C}_{k\&times;1}& I_{(L-1)\&times;(L-1)}\\ 0_{(L-k)\&times;(L-1)}& 0_{1\&times;1}\end{array}\right]$

Wherein, C _{k × 1}=[a ₁a ₂a _k] be the coefficient vector of proper polynomial P (Z), 0 _{(L-K) × (L-1)}, 0 _{1 × 1}for null matrix, I _{(L-1) × (L-1)}for unit matrix.

Described proper polynomial gets trinomial P (Z)=z ^k-z ^q-1.

(2-1-2) the transfer matrix A using step (2-1-1) to obtain and the L position state information X of current time _n=(x _n+L-1, x _n+L-2..., x _n+1, x _n), calculate the L position state information X of subsequent time _n+1, i.e. X _n+1=X _n× A.

(2-1-3) according to relational expression and (2-1) middle random number t of step (2-1-2) _nwith binary sequence T _n=X _nsone-to-one relationship, recursion can obtain following relational expression:

T _n+1=X _(n+1)s=X _(n+1)s-1×A=(X _(n+1)s-2×A)×A=…=X _ns×A ^s=T _n×A ^s

(2-1-3) relational expression obtained according to step (2-1-3) can obtain a new random number t _n+1, its saltus step step-length is s.Random number producing method can use following matrix notation:

T _n+1=T _n×A ^s

(2-2) the combinedTausworthe random number u of period expansion is produced by J Tausworthe uniform random number _n.

Described combined Tausworthe random number relational expression is:

$u_n=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}(x_{\mathrm{ns}+i-\mathrm{1,1}}\&CirclePlus;x_{\mathrm{ns}+i-\mathrm{1,2}}\&CirclePlus;...\&CirclePlus;x_{\mathrm{ns}+i-1,J})\&CenterDot;2^{-i}$

Wherein, be J independently random number sequence, as J=1, u _n=t _n.

(2-3) according to the random number called after that step (2-2) two combined Tausworthe produce: u _{n, 1}and u _{n, 2}, produce gaussian random distribution number α and β thus.Relational expression is as follows:

$\mathrm{\&alpha;}=\sqrt{-2\&times;\ln \left(u_{n,1}\right)}\sin \left(2\mathrm{\&pi;}{u}_{n,2}\right)$

$\mathrm{\&beta;}=\sqrt{-2\&times;\ln \left(u_{n,1}\right)}\cos \left(2\mathrm{\&pi;}{u}_{n,2}\right)$

White Gaussian noise α and β that Fig. 2 (a) and Fig. 2 (b) produces for the present invention.

(3) energy that the averaged signal energy power_avg obtained according to given signal to noise ratio snr and step (1) and step (2) obtain is the multiple noise α+j β of 2, the signal amplitude factors A that search is corresponding and noise amplitude factor B.

Described signal to noise ratio wherein b=(2 ^cnt_shift) ²; a=y ²>1, and a<4; Cnt_shift is the bit number that multiple noise signal α+j β is moved to the left, and namely multiplication factor is 2 ^cnt_shift, then energy multiplication factor is b=(2 ^cnt_shift) ²; Y is signal amplitude multiplication factor, then energy amplifies y ².

Described signal amplitude factors A=y, noise amplitude factor B=2 ^cnt_shift.

Described search step is as follows:

(3-1) amplitude factor B is searched for: first make the signal amplitude factor be 1, i.e. a=y ²=1, then cnt_shift increases progressively by 0, until meet till, obtain cnt_shift and B=2 ^cnt_shift.

(3-2) search signal amplitude factor A: make a by 1 by fixing multiple x ²the mode increased progressively searches for a, until meet $\mathrm{snr}\&le;\frac{a\&times;\mathrm{power}\_\mathrm{avg}}{2b},$ Complete and search for $A=y=\sqrt{a}.$

Described fixing multiple x is the multiple of amplitude factor y, and its choosing method is: assuming that the quantizing bit number of fixing multiple incremental change x is BITS, the value after taken amount is X=2 ^bITS+ 1, then when getting BITS=11, x=1.00048828125, now energy per increases progressively 0.004dB, and namely energy error is in 0.004dB.

Described choosing method, when BITS is enough large, therefore the incremental change of signal energy can be expressed as $\mathrm{power}\_\mathrm{avg}\&CenterDot;x^2=\mathrm{power}\_\mathrm{avg}+\frac{\mathrm{power}\_\mathrm{avg}}{2^{(\mathrm{BITS}-1)}}.$ This method for expressing only uses adder and shift register just can complete multiplying, saves hardware logic resource.

As shown in Figure 3, the present invention devises a line level aggregated(particle) structure for realizing search approximate algorithm.

(4) zero intermediate frequency input signal I+jQ is multiplied by signal amplitude factors A that step (3) obtains adjusted after signal SIG; The multiple noise signal α+j β obtained by step (2) be multiplied by noise amplitude B that step (3) obtains adjusted after noise NOISE; Signal SIG and noise NOISE obtains through noise summation module adding the signal SIG_NOISE made an uproar.

(5) as shown in Figure 4, digital Auto Gain module DAGC automatically adjusts to the SIG_NOISE that step (4) obtains according to given thresholding R.Step is as follows: the energy balane first carrying out signal obtains E _y(n), and by logarithmic transformation, energy value is converted into logarithmic form E _y1(n); Pass through E _y1n () compares with threshold value R's, and the subtraction carrying out logarithm obtains ε (n); Then amplitude gain regulated value G is obtained through antilogarithm computing; Eventually passed the amplitude gain G in a multiplier and a upper moment _n-1be multiplied to obtain the gain G of current time _n.The mean effort of integrator can cut noise reduced interference.

Described step can represent by following calculating formula:

E _y(n)=I ²+Q ²

E _y1(n)=log ₂(I ²+Q ²)

$\mathrm{\&epsiv;}\left(n\right)=\frac{1}{2}[{\log }_{2}R-{\log }_2(I^2+Q^2)]$

G=2 ^ε(n)

G _n=G×G _n-1

The device of a kind of Zero intermediate frequency Gaussian white noise adding method of the present invention, uses FPGA(field programmable logic array) throughput that realizes producing random number is more than 400 times that software realization mode produces random number throughput; Search approximate algorithm realizes energy error within 0.004dB; This device increases automatic gain module, is convenient to transplant, and can be used for the interference free performance detecting Wireless Telecom Equipment or satellite radio receiver equipment in signal source system.

A kind of device realizing described Zero intermediate frequency Gaussian white noise adding method comprises:

Zero intermediate frequency input signal I+jQ, average energy computing module, amplitude factor search module, additive white Gaussian noise generation module (AWGN), noise summation module, digital Auto Gain module (DAGC).

Described zero intermediate frequency input signal I+jQ is connected with average energy computing module and amplitude factor search module respectively, and input signal can be obtained through A/D conversion or digital intermediate frequency input by analog signal;

Described average energy computing module is connected with zero intermediate frequency signals I+jQ and amplitude factor search module respectively;

Described amplitude factor search module is connected with zero intermediate frequency signals I+jQ input signal and AWGN module respectively, and the signal to noise ratio snr according to input produces signal amplitude factors A, noise amplitude factor B;

Described AWGN module is connected with amplitude factor search module, and the white Gaussian noise unit energy of generation is 1, and then obtains the multiple noise α+j β that energy is 2.

Described noise summation module is connected with the zero intermediate frequency signals I+jQ after amplitude adjusted, white Gaussian noise signal alpha+j β and DAGC module respectively;

Described DAGC module is connected with noise summation module, export certain gain and zero intermediate frequency signals containing white Gaussian noise.

Above-described embodiment is the present invention's preferably execution mode; but embodiments of the present invention are not restricted to the described embodiments; change, the modification done under other any does not deviate from Spirit Essence of the present invention and principle, substitute, combine, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.

Claims (5)

1. a Zero intermediate frequency Gaussian white noise adding method, is characterized in that, comprises the steps:

(1) energy power_avg is averaging to the zero intermediate frequency signals of input: the energy power first calculating the zero intermediate frequency signals I+jQ of input, is then averaging to obtain averaged signal energy power_avg to the signal energy of a period of time;

(2) produced additive white Gaussian noise α and β of unit energy by white Gaussian noise generation module, and then obtain the multiple noise α+j β that energy is 2;

(3) energy that the averaged signal energy power_avg obtained according to given signal to noise ratio snr and step (1) and step (2) obtain is the multiple noise α+j β of 2, the signal amplitude factors A that search is corresponding and noise amplitude factor B;

Described signal to noise ratio wherein b=(2 ^cnt_shift) ²; A=y ²>1, and a<4; Cnt_shift is the bit number that multiple noise signal α+j β is moved to the left, and namely multiplication factor is 2 ^cnt_shift, then energy multiplication factor is b=(2 ^cnt_shift) ²; Y is signal amplitude multiplication factor, then energy amplifies y ²;

Described signal amplitude factors A=y, noise amplitude factor B=2 ^cnt_shift;

The signal amplitude factors A adopting the search of search approximate algorithm corresponding and noise amplitude factor B, described search approximate algorithm comprises the following steps:

(3-1) amplitude factor B is searched for: first make the signal amplitude factor be 1, i.e. a=y ²=1, then cnt_shift increases progressively by 0, until meet till, obtain cnt_shift and B=2 ^cnt_shift;

(3-2) search signal amplitude factor A: make α by 1 by fixing multiple x ²the mode increased progressively searches for α, until meet $\mathrm{snr}\&le;\frac{a\&times;\mathrm{power}\_\mathrm{avg}}{2b},$ Complete and search for $A=y=\sqrt{a};$

Described fixing multiple x is the multiple of amplitude factor y, and its choosing method is: assuming that the quantizing bit number of fixing multiple incremental change x is BITS, the value after taken amount is X=2 ^bITS+ 1, then when getting BITS=11, x=1.00048828125, now energy per increases progressively 0.004dB, and namely energy error is in 0.004dB;

Described choosing method, when BITS is enough large, therefore the incremental change of signal energy can be expressed as $\mathrm{power}\_\mathrm{avg}\&CenterDot;x^2=\mathrm{power}\_\mathrm{avg}+\frac{\mathrm{power}\_\mathrm{avg}}{2^{(\mathrm{BITS}-1)}};$

(4) zero intermediate frequency input signal I+jQ is multiplied by signal amplitude factors A that step (3) obtains adjusted after signal SIG; The multiple noise signal α+j β obtained by step (2) be multiplied by noise amplitude B that step (3) obtains adjusted after noise NOISE; Signal SIG and noise NOISE obtains through noise summation module adding the signal SIG_NOISE made an uproar;

(5) digital Auto Gain module DAGC automatically adjusts to the SIG_NOISE that step (4) obtains according to given thresholding R.

2. a kind of Zero intermediate frequency Gaussian white noise adding method according to claim 1, is characterized in that, step (2) is specially:

(2-1) the uniform random number t on algorithm generation (0,1) of matrixing is used according to Tausworthe method _n;

Described uniform random number t _nwith the random sequence x of L length _nthere is following relational expression:

$t_n=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ns}+i-1}\&CenterDot;2^{-i},$

Wherein, L and s is non-zero positive integer, and L is that random number exports bit wide, and s is saltus step step-length; Obviously, random number t _nwith binary sequence T _n=X _ns=(x _ns, x _ns+1..., x _ns+L-2, x _ns+L-1) one_to_one corresponding, therefore t in the present invention _nwith T _ndo not do special differentiation;

Described random sequence x _nrecurrence Relation as follows:

$x_n=a_1{x}_{n-1}\&CirclePlus;a_2{x}_{n-2}\&CirclePlus;...\&CirclePlus;a_k{x}_{n-k}$

Wherein, a ₁, a ₂a _kfor proper polynomial P (z)=z ^k-a ₁z ^k-1-...-a _kcoefficient;

(2-2) the combinedTausworthe random number u of period expansion is produced by J Tausworthe uniform random number _n;

Described combined Tausworthe random number relational expression is:

$u_n=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}(x_{\mathrm{ns}+i-\mathrm{1,1}}\&CirclePlus;x_{\mathrm{ns}+i-\mathrm{1,2}}\&CirclePlus;...\&CirclePlus;x_{\mathrm{ns}+i-1,J})\&CenterDot;2^{-i}$

Wherein, be J independently random number sequence, as J=1, u _n=t _n;

(2-3) according to the random number called after that step (2-2) two combined Tausworthe produce: u _{n, 1}and u _{n, 2}, produce additive white Gaussian noise α and β thus; Relational expression is as follows:

$\mathrm{\&alpha;}=\sqrt{-2\&times;\ln \left(u_{n,1}\right)}\sin \left(2\mathrm{\&pi;}{u}_{n,2}\right)$

$\mathrm{\&beta;}=\sqrt{-2\&times;\ln \left(u_{n,1}\right)}\cos \left(2\mathrm{\&pi;}{u}_{n,2}\right).$

3. a kind of Zero intermediate frequency Gaussian white noise adding method according to claim 2, is characterized in that, in step (2-1), described matrixing algorithm, can produce a binary sequence T within a clock cycle _n, thus improve execution efficiency, concrete shift step is as follows:

(2-1-1) calculate transfer matrix A, expression formula is:

$A=\left[\begin{array}{cc}{C}_{k\&times;1}& I_{(L-1)\&times;(L-1)}\\ 0_{(L-k)\&times;(L-1)}& 0_{1\&times;1}\end{array}\right],$

Wherein, C _{k × 1}=[a ₁a ₂a _k] be the coefficient vector of proper polynomial P (Z), 0 _{(L-K) × (L-1)}, 0 _{1 × 1}for null matrix, I _{(L-1) × (L-1)}for unit matrix;

Described proper polynomial gets trinomial P (Z)=z ^k-z ^q-1;

(2-1-2) the transfer matrix A using step (2-1-1) to obtain and the L position state information X of current time _n=(x _n+L-1, x _n+L-2..., x _n+1, x _n), calculate the L position state information X of subsequent time _n+1, i.e. X _n+1=X _n× A;

(2-1-3) according to relational expression and (2-1) middle random number t of step (2-1-2) _nwith binary sequence T _n=X _nsone-to-one relationship, recursion can obtain following relational expression:

T _n+1＝X _(n+1)s＝X _(n+1)s-1×A＝(X _(n+1)s-2×A)×A＝…＝X _ns×A ^s＝T _n×A ^s

(2-1-4) relational expression obtained according to step (2-1-3) can obtain a new random number t _n+1, its saltus step step-length is s; Random number producing method can use following matrix notation:

T _n+1＝T _n×A ^s。

4. a kind of Zero intermediate frequency Gaussian white noise adding method according to claim 1, is characterized in that, described step (5) is specially:

(5-1) energy balane carrying out signal obtains E _y(n), and by logarithmic transformation, energy value is converted into logarithmic form E _y1(n); Pass through E _y1n () compares with threshold value R's, and the subtraction carrying out logarithm obtains ε (n);

(5-2) amplitude gain regulated value G is obtained through antilogarithm computing;

(5-3) the amplitude gain G in a multiplier and a upper moment was eventually passed _n-1be multiplied to obtain the gain G of current time _n.

5. a kind of device of Zero intermediate frequency Gaussian white noise adding method according to any one of claim 1-4, it is characterized in that, comprise zero intermediate frequency signals input module, average energy computing module, amplitude factor search module, additive white Gaussian noise generation module AWGN, noise summation module, digital Auto Gain module DAGC;

Zero intermediate frequency signals input module is used for being connected with average energy computing module and amplitude factor search module, and input signal is obtained through A/D conversion or digital intermediate frequency input by analog signal;

Described average energy computing module is connected with zero intermediate frequency signals input module and amplitude factor search module respectively;

Described amplitude factor search module is connected with zero intermediate frequency signals input module and AWGN module respectively, and the signal to noise ratio snr according to input produces signal amplitude factors A, noise amplitude factor B;

Described AWGN module is connected with amplitude factor search module, and the white Gaussian noise unit energy of generation is 1, and then obtains the multiple noise α+j β that energy is 2;

Described noise summation module with the zero intermediate frequency signals I+jQ after amplitude adjusted, energy be respectively 2 multiple noise α+j β and DAGC module be connected;

Described DAGC module is connected with noise summation module, export certain gain and zero intermediate frequency signals containing white Gaussian noise.