CN102201788A - Digital noise generation method - Google Patents

Abstract

The invention discloses a digital noise generation method, which comprises the following steps of: 1, generating a group of standard noises forming Gaussian distribution, and storing the standard noises in a read only memory (ROM); 2, generating a plurality of groups of random addresses for reading the standard noises, reading a plurality of paths of standard noises from the ROM by utilizing the plurality of groups of random addresses, and storing the read standard noises into a first register; 3, superposing the plurality of paths of read standard noises; 4, calculating a multiplication coefficient according to the average power of input signals, superposed noise power and an input signal to noise ratio; and 5, multiplying the multiplication coefficient by using the superposed noises to generate required digital noises. By the digital noise generation method, the noises with relatively higher amplitude can be generated, and the requirements of a plurality of modulation ways on a bit error rate in a high signal to noise ratio can be met; and the method is high in randomness and rate and highly accurate.

Description

The digital noise production method

Technical field

The present invention relates to digital communication technology field, relate in particular to a kind of digital noise production method of under the noise circumstance system being tested that is applicable to.

Background technology

At present, in the digital communication system design process,, need under noise circumstance, test system in order to detect the quality of communication quality.When test, adopt usually on radio frequency and add noise with the method for simulation sending data, add noise with the method for numeral all is a difficult point all the time.

Security reliability analysis during digital noise can be used for communicating by letter, or the generation in dexterous noise jamming source, in a lot of electronic equipments of modern times, the noise jamming test has become inspection machine and whether has had one of important link of good noiseproof feature, therefore also more and more higher for the requirement of noise source, method commonly used is to produce noise with signal source, but the price of general signal source is very high, buy for domestic consumer just very to one's profit just to test, thereby design digital noise source voluntarily and replace signal source, have using value preferably.When the design digital noise source, use the method for more generation digital noise at present, be the thermal noise and the pseudo random number emulation of gathering electronic devices and components.The power spectrum of preceding a kind of method is uneven and be difficult to control, thereby alternative by a kind of method in back gradually; A kind of method in back mainly comprises linear congruence, linear feedback shift register etc., but the noise amplitude of its generation is less, can not satisfy the error rate requirement of multiple modulation system under the high s/n ratio.

Summary of the invention

(1) technical problem that will solve

The technical problem to be solved in the present invention is: a kind of digital noise production method is provided, and it can produce the bigger noise of amplitude, can satisfy the error rate requirement of multiple modulation system under the high s/n ratio, and has high randomness, two-forty, high precision.

(2) technical scheme

For addressing the above problem, the invention provides a kind of digital noise production method, comprise step:

S1 produces and the standard noise of one group of Gaussian Profile of storage in ROM;

S2 generates the random address that many groups are used to read standard noise, utilizes described many group random addresss to read the described standard noise of multichannel from ROM and is stored in first register;

S3 is with the multichannel standard noise stack of being read;

S4 is according to average power and the noise power after the stack and the input signal-to-noise ratio calculating times multiplying factor of input signal;

S5 multiply by a times multiplying factor with the noise after the stack, produces required digital noise.

Wherein, described step S1 comprises:

The figure that Gaussian Profile probability density function curve and reference axis are surrounded is divided into the bar shaped that several areas equate along being parallel to the coordinate y direction;

The abscissa intermediate value at each bar shaped place is rounded as the integer value of representing this bar shaped;

Store the integer value of each bar shaped representative.

Wherein, among the described step S2, generating the random address that many groups are used to read standard noise comprises: produce many group pseudo-random binary sequences by second register, and with after the every random alignment of pseudo-random binary sequence that produces as the step of the random address that reads standard noise.

Wherein, among the described step S3, the described standard noise stack of multichannel is comprised:, the multichannel standard noise that is read is superposed to the step of one road noise according to the principle that remains Gaussian process after the independent Gaussian process stack.

Wherein, described step S4 comprises: the average power of the noise after the known stack is p _n, according to the input data signal is averaged energy measuring, obtain signal energy and be designated as p _Av, again according to the signal to noise ratio snr and the p that import _AvUsing formula

$\mathrm{SNR}=10\×\log \frac{p_{\mathrm{av}}}{p_{\mathrm{sn}}}$

Calculate the noise power p of actual needs _Sn, use

Calculate a times multiplying factor.

Wherein, described method also comprises, every tentation data, the signal of input carried out power detection again, recomputates and upgrades a times multiplying factor.Adopt power detection dynamically to adjust the method for times multiplying factor, can simulate required noise more accurately.

Wherein, described second register is the different non-isometric linear feedback shift registers of multidiameter delay of many group generator polynomials.

(3) beneficial effect

The present invention at first produces and stores the standard noise of one group of Gaussian Profile, generate many group random addresss then, read many group standard noise and make it stack according to the additivity principle of Gaussian process, calculate times multiplying factor according to noise power after the average power of input signal and the stack and input signal-to-noise ratio again, multiply by a times multiplying factor with the noise after the stack at last, produce the needed number noise; The present invention can produce the bigger noise of amplitude, can satisfy the error rate requirement of multiple modulation system under the high s/n ratio, and has high randomness, two-forty, high precision.

Description of drawings

Fig. 1 is the flow chart of digital noise production method in the embodiment of the invention;

Fig. 2 is the schematic diagram that produces standard noise in the described method of the embodiment of the invention;

Fig. 3 is that numeral adds the performance evaluation figure that makes an uproar in the described method of the embodiment of the invention.

Embodiment

Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.

As shown in Figure 1, digital noise production method of the present invention comprises step:

S1 produces and the standard noise of one group of Gaussian Profile of storage in ROM;

In this step, specifically comprise: for consideration to resource and Gaussian noise characteristic, the figure that Gaussian Profile probability density function curve and reference axis are surrounded along be parallel to the coordinate y direction be divided into a plurality of, for example 2 ¹⁴The bar shaped that (promptly 16384) individual area equates;

The intermediate value of the abscissa at each bar shaped place rounds the back as the integer value of representing this bar shaped;

The integer value of each bar shaped representative is calculated by computer, is stored in 2 ¹⁴The bytes of memory device is for example on the sheet among the ROM.

Fig. 2 is for producing the schematic diagram of standard noise, for example adopts the bar shaped method that figure is divided into the schematic diagram of 8 bar-shaped zones, is that figure has been divided into 2 among the present invention ¹⁴Individual bar-shaped zone.

S2 generates the random address that many groups are used to read standard noise, utilizes described many group random addresss to read the described standard noise of multichannel from ROM and is stored in first register;

In this step, specifically comprise: by second register, the non-isometric linear feedback shift registers of multidiameter delay that for example many group generator polynomials are different produce many group pseudo-random binary sequences, and with after the every random alignment of pseudo-random binary sequence that produces as the random address that reads standard noise.

In this step, because the size of the ROM of storage standards noise is 2 ¹⁴(promptly 16384) byte is so each random address that is used to read standard noise is 14 2 system sequences.One group of random address is input among the ROM, can obtains storing value corresponding among the ROM.

Linear feedback shift register described in this step (LFSR) can produce pseudo-random binary sequence (be called for short PRBS, be called the M sequence again).If the register progression of LFSR is m, the formation sequence cycle is 2 so ^m-1.The M sequence has a lot of good characteristics, such as ' 0 ' ' 1 ' is bordering on even distribution in the sequence; Distance of swimming distribution character is preferably arranged; Another advantage of LFSR is only to have used XOR, calculates simply, and logical time delay is little.When adopting multidiameter delay LFSR to generate multiplex sequence, multichannel LFSR works simultaneously according to different generator polynomials, and the 1bit result that every road LFSR generates represents the bit of the pseudorandom integer that is produced respectively.By ' 0 ' ' 1 ' being bordering on equally distributed character and obtaining the pseudorandom integer and be bordering on equally distributed characteristic on each bit position.Multidiameter delay LFSR adopts non-isometric register progression, and then the cycle of the pseudorandom integer sequence of Chan Shenging is the least common multiple of all M sequence periods.

For example, the even distribution pseudorandom integer bit wide that produces among the present invention is 14 bits, and the progression of 14 road LFSR is followed successively by 19～32, asks for least common multiple with [] expression, and then the pseudorandom integer sequence Cycle Length of Sheng Chenging is

T＝[2 ¹⁹-1，2 ²⁰-1，...，2 ³²-1]≈1.31×10 ⁹⁰

For producing many group pseudo-random binary sequences, adopt 14 tunnel different parallel LFSR of many group generator polynomials among the present invention; For correlation between the pseudo-random binary sequence is organized in more effective elimination more, the pseudo-random binary sequence step-by-step random alignment that again 14 tunnel parallel LFSR is generated among the present invention, concrete enforcement is as follows: the invariant position of fixing No. 14 registers, the 1bit that every road register is produced composes each to 14 bit address, and assignment adopts random sequence in proper order.

Table 1

Table 2

Table 1 and table 2 generate the signal table for random address.Prbs** is corresponding * * register in the table 1, and x_re is 14 random addresss, and the numerical value in the x_re bracket is the position information that puts in order of address.Table 2 is compared with table 1, the invariant position of register, but the position information difference that puts in order of random address x_re, both be generate at random, separate.Because there is the cut position error in the standard noise of storage, thus take to produce at every turn multichannel for example 16 tunnel random noises superpose and produce the method for a road sign quasi-noise, can effectively remedy the too small problem of span that the cut position error is brought.

S3 is with the described standard noise stack of the multichannel that is read;

Remain the principle of Gaussian process after this step superposes according to independent Gaussian process, the multichannel standard noise that is read is superposed to one road noise.

S4 is according to average power and the noise power after the stack and the input signal-to-noise ratio calculating times multiplying factor of input signal;

Noise power after the known stack is p _n, according to the input data signal is averaged energy measuring, obtain signal energy and be designated as p _Av, again according to the signal to noise ratio snr and the p that import _AvUsing formula

$\mathrm{SNR}=10\×\log \frac{p_{\mathrm{av}}}{p_{\mathrm{sn}}}$

Calculate the noise power p of actual needs _Sn, use

Calculate a times multiplying factor.To I, the Q two paths of signals carries out total average power that power detection obtains two paths of signals, again according to the signal to noise ratio of importing, calculate required noise power by formula, carry out computing with the noise power after this noise power and the existing stack again, obtain the coefficient that on the existing standard noise, doubly to take advantage of.Every tentation data, 4096 point data are for example carried out power detection again to the signal of input, recomputate and upgrade a times multiplying factor.Adopt power detection dynamically to adjust the method for times multiplying factor, can simulate required noise more accurately.

S5 multiply by a times multiplying factor with the noise after the stack, produces required digital noise.

Fig. 3 adds the performance evaluation figure that makes an uproar for adopting method of the present invention to carry out numeral, the modulation system that adopts is 16QAM, dotted line is the characteristic curve of error code of 16QAM in theory, solid line is for adopting the actual error code curve that obtains of this method, as can be seen from the figure, from-3db is in the SNR scope of 16.5db, the error code curve of this method and the error of theoretical curve are no more than 0.1db.And the error rate requires to reach 10 ^-9

The present invention is according to the additivity principle of Gaussian process, and promptly separate Gaussian process stack still is a Gaussian process later, and power is the addition relation.The employing multichannel for example method of 16 tunnel random noises stack produces digital noise, can enlarge the amplitude peak of noise under the condition that does not change noise power, makes module still can produce error code when high s/n ratio, more accurate simulation white Gaussian noise.At SNR and input power p _AvUnder certain condition, the power of noise also is certain, is designated as p _SnIf the method that does not adopt multipath noise to superpose only reads with single channel, resulting standard noise is expressed as n, and its power is p _n, at this moment the noise of Chan Shenging is:

$N=n\×\sqrt{p_{\mathrm{sn}}/p_n}---\left(1\right)$

If the noise of amplitude maximum is n among the n _Max, the amplitude peak noise that then can produce is:

$N_{\max }=n_{\max }\×\sqrt{p_{\mathrm{sn}}/p_n}---\left(2\right)$

If adopt the method for 16 tunnel random noises stack, because 16 road noises that produce are independent noise, according to the principle of stacking of Gaussian process, after stack, resulting noise n _AddPower be 16 * p _n, and p at this moment _SnConstant, the noise of generation is:

$N^{\′}=n_{\mathrm{add}}\×\sqrt{p_{\mathrm{sn}}/\left(16p_n\right)}---\left(3\right)$

This moment n _AddThe noise of middle amplitude maximum is 16 * n _Max, the amplitude peak noise that then can produce is:

$N_{\max }^{\′}=16\×n_{\max }\×\sqrt{p_{\mathrm{sn}}/\left(16p_n\right)}=4\×n_{\max }\×\sqrt{p_{\mathrm{sn}}/p_n}---\left(4\right)$

For not adopting four times of amplitude peak noise under the multipath noise stack situation.

Above execution mode only is used to illustrate the present invention; and be not limitation of the present invention; the those of ordinary skill in relevant technologies field; under the situation that does not break away from the spirit and scope of the present invention; can also make various variations and modification; therefore all technical schemes that are equal to also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.

Claims (7)

1. a digital noise production method is characterized in that, comprises step:

S1 produces and the standard noise of one group of Gaussian Profile of storage in ROM;

S2 generates the random addresss that many groups are used to read standard noise, utilizes described many group random addresss to read the described standard noise of multichannel from ROM and is stored in first register;

S3 is with the multichannel standard noise stack of being read;

S4 is according to average power and the noise power after the stack and the input signal-to-noise ratio calculating times multiplying factor of input signal;

S5 multiply by a times multiplying factor with the noise after the stack, produces required digital noise.

2. digital noise production method as claimed in claim 1 is characterized in that: described step S1 comprises:

The figure that Gaussian Profile probability density function curve and reference axis are surrounded is divided into the bar shaped that several areas equate along being parallel to the coordinate y direction;

The abscissa intermediate value at each bar shaped place is rounded as the integer value of representing this bar shaped;

Store the integer value of each bar shaped representative.

3. digital noise production method as claimed in claim 1, it is characterized in that: among the described step S2, generating the random address that many groups are used to read standard noise comprises: produce many group pseudo-random binary sequences by second register, and with after the every random alignment of pseudo-random binary sequence that produces as the step of the random address that reads standard noise.

4. digital noise production method as claimed in claim 1, it is characterized in that: among the described step S3, the described standard noise stack of multichannel is comprised:, the multichannel standard noise that is read is superposed to the step of one road noise according to the principle that remains Gaussian process after the independent Gaussian process stack.

5. digital noise production method as claimed in claim 1 is characterized in that: described step S4 comprises: the average power of the noise after the known stack is p _n, according to the input data signal is averaged energy measuring, obtain signal energy and be designated as p _Av, again according to the signal to noise ratio snr and the p that import _AvUsing formula

$\mathrm{SNR}=10\×\log \frac{p_{\mathrm{av}}}{p_{\mathrm{sn}}}$

Calculate the noise power p of actual needs _Sn, use

Calculate a times multiplying factor.

6. digital noise production method as claimed in claim 1 is characterized in that: described method also comprises, every tentation data, the signal of input carried out power detection again, recomputates and upgrades a times multiplying factor.

7. digital noise production method as claimed in claim 3 is characterized in that: described second register is the different non-isometric linear feedback shift register of multidiameter delay of many group generator polynomials.

Images