CN1818853A - Device and method for producing true random codes - Google Patents

Abstract

An equipment of producing really stochastic code, it comprises many generators of the multi-threshold stochastic code and the data processors, the noise generators connect with every generators of the multi-threshold stochastic code respectively that consist of many units of the generators of the multi-threshold stochastic code, every generators of the multi-threshold stochastic code connect with the data processors. The way to produce the really stochastic code: get the list of the really stochastic code discriminate the stochastic level signal from the noise generators by the generator of multi-threshold stochastic code, then in every cycle, get one digit in each list of the stochastic code that produce by the units of the generators of multi-threshold stochastic code, incorporate them to a binary really stochastic code and output it by the data processor. The equipment is getting the output of the really stochastic code by the generator of multi-threshold stochastic code that improves the output rate of the stochastic code and the randomicity. The work of the equipment is stable, the output rate can be extended, the randomicity can improve, and these accord the needs of the quanta secrecy of the really stochastic code.

Description

A kind of method that produces the device of real random code and produce real random code

Technical field

The present invention relates to light quantum private communication technology field, be specifically related to a kind of method that produces the device of real random code and produce real random code.

Background technology

The effect of real random code generator is to produce the code book that never repeats, and guarantees can cracking anything but of one-time pad.The quantum channel of quantum secret communication is not to be used for transmitting ciphertext or plaintext, but is used for setting up and the transmission code book.Finish whole quantum communications process, need to adopt two communication channels, one is quantum channel, is used for the transmission of quantum key; Article one, be classical channel, be used for the checking of key and the transmission of ciphertext.Quantum channel can not be eavesdropped by its physical principle.And be not can eavesdrop anything but by two processes that classical channel carries out.During key authentication, the result of announcement is disclosed, and in order to prevent that the listener-in from according to the part of announcing, inferring unpub part, all quantum keys must be incoherent.In actual quantum key distribution system, receiving-transmitting sides Alice and Bob need use random code to prepare at random and with the quantum state of machine testing single photon.The quantum cryptography sequence extracts from the random code sequence, thus require random code everybody add up independently.Simultaneously, verified on the mathematics, the one-time pad system is an overall safety, the password that can not break.The needed size of key of one-time pad system equals the quantity of corresponding plaintext, and only uses once.The being perfectly safe property of quantum secret communication system is to determine that by the being perfectly safe property of quantum channel and classical channel is common its theoretical foundation is respectively that quantum can not be cloned principle and one-time pad principle.Guarantee the being perfectly safe property of classical channel, must adopt the real random code generator to produce everybody and add up independently real random code,, thereby guarantee being perfectly safe of classical channel of quantum secret communication for the encryption system of one-time pad provides the code book that never repeats.

1) method that produces random code mainly contains two kinds of mathematical method and physical methods

(1) mathematical method

Random-number generating method commonly used on the mathematics has: the RAND table produces pseudo random number, the linear congruence generator produces pseudo-random number sequence, feedback shift register generation pseudo-random number sequence, pseudorandom number generator ANSIX9.17 or the like.These are only determined by algorithm and seed with the pseudo-random number sequence that mathematical method produces, in case after algorithm and seed were determined, the value of each in the sequence determined that all therefore, information entropy is zero, can not be used for the quantum secure system.

(2) physical method

The method of generation random code physically is that random signal is passed through transformation system, obtains the scale-of-two random code.The research work of real random code generator mainly concentrates in the research work of stochastic source at present, for transformation system, generally all adopts the scheme of comparer examination.

2) common random signal source

(1) chooses the natural noise of real world

There is abundant chance phenomenon in nature, can utilize the randomness of various noise signals itself to obtain real random code.

(2) quantum random occurrence

The resonance fluorescence radiation that the ion that is held in captivity produces, its photon off time is a stochastic distribution, photon randomness by optical beam-splitter or the like.

(3) random occurrence that produces around computing machine

The portion copy of each keystroke, mouse command, certain time of one day of sector number are sought content of the delay of each disk operating, actual mouse position, scanning of a display line number, actual displayed image or the like.

3) hardware of true random number produces progress both domestic and external

G.B.Agnew proposes two Metal-Insulator-Semi-Conductor capacitors (MISC) very near-earth are put together, and random order is the function of the difference of their quantities of electric charge between two; M.Gude has made a randomizer of gathering random order according to physical phenomenon such as radioactivity decay; Manfield Richter has developed the randomizer of a based semiconductor diode thermonoise; Nanjing University also succeeds in developing a kind of randomizer based on chaos principle, chaos is summarized as the uncertainty by the caused solution of equation of nonlinear effect of determinacy equation (the finger-type formula is determined), exemplary has the Lorenz differential equation and map Logist, they can utilize the very little deviation of initial value to cause the very large deviation of separating and produce after producing chaos effect, endless on the cyclical theory of random series, this method can be used for synchronous communication, the course of work be chaos → chaos synchronously → chaos; The Chinese Academy of Sciences's developed of physics is based on the light quantum stochastic source of single photon by the randomness of beam splitter; East China Normal University has developed the stochastic source based on the polarized light quantum; The broadband white noise that usefulness Erbium-Doped Fiber Amplifier (EDFA) such as Lepley and pin detector produce is done the input of high speed d type flip flop, has produced the non-return-to-zero real random code of 2.5Gb/s and 5Gb/s.

4) comparison of the whole bag of tricks of generation real random code

The real random code generator is divided into tradition and quantum two big classes by its physical accidental source of adopting.Traditional random code generator adopts noise signal as stochastic source, by simple gate limit discriminator output random code, though realize easily, speed is fast, the influence that is subjected to the various statistical properties of noise to be difficult to control and keep factors such as stablizing, the randomness of the random code of generation is relatively poor.And the quantum stochastic source is the completely random source of never repeating, though it had been both irregular can not repeat yet, but difficulty is big in data acquisition, the equipment complexity, the quantum stochastic source reaches the requirement of quantum secure far away on stability and generation speed simultaneously, and be subjected to the influence of factors such as interference noise, the randomness of the actual random code that is produced can not show a candle to the theoretical value ideal.For example, the quantum secure system test adopts single photon to make stochastic source usually, though randomness is better, but this mode is subjected to the restriction of stability and speed, also rest on laboratory stage, its related coefficient is unsatisfactory, even greater than 0.05, promptly surpassed the incoherent upper limit, can think that the random code of output is correlated with.

Quantum secret communication will reach the practical stage, and the random code that just needs generation is casual good, working stability, and the random code generator that energy consumption is little, and above-mentioned random code generator is difficult to reach simultaneously these requirements.

Summary of the invention

The object of the present invention is to provide a kind of device that produces real random code, this device is screened the output that obtains real random code by the multi-threshold random code generator, to improve random code bit rate output and randomness.

Another object of the present invention is to provide a kind of and produce the method for real random code with said apparatus, this method can produce not to be needed to adjust, working stability, and bit rate output can be expanded, and randomness can improve, and meets the real random code of quantum secret communication needs.

The device of generation real random code of the present invention comprises noise generator, several multi-threshold random code generators, data processor, noise generator links to each other respectively with each multi-threshold random code generator, form several random code generator unit, each multi-threshold random code generator links to each other with data processor.

Described multi-threshold random code generator is an analog-digital converter, and analog-digital converter is exported with lowest order; Or comprising comparator array and demoder, noise generator, comparator array, demoder, data processor connect successively; Comparator array is the uses that are connected in parallel of several comparers, and noise generator links to each other respectively with each comparer, and comparator array is exported corresponding Transistor-Transistor Logic level, through obtaining binary real random code sequence after the decoder decode.

Produce the method for real random code with said apparatus:

The level signal at random that noise generator is produced obtains binary real random code sequence by the examination of multi-threshold random code generator, then in each cycle, binary real random code sequence that several random code generator unit produce is respectively got one, merge into the real random code sequence output of a binary sequence by data processor.

The level signal at random that described noise generator produces is through the examination of comparator array, and comparator array is exported corresponding Transistor-Transistor Logic level, through obtaining binary real random code sequence after the decoder decode.

The level signal at random that described noise generator produces also can be through the examination of analog-digital converter, as output, obtains binary real random code sequence with the lowest order of analog-digital converter.

Described data processor is parallel-to-serial converter or data collecting card.

Characteristics of the present invention are that transformation system is an analog-digital converter in the transformation system of multi-threshold random code generator and the output characteristics expansion, or comparator array and demoder are connected to form.

The present invention is divided into two groups to the noise signal digitizing according to its parity then, the corresponding random code " 1 " (or " 0 ") of odd number group, the corresponding random code " 0 " (or " 1 ") of even number set.Its principle is specific as follows:

If the noise voltage V of input drops on interval [V _Min, V _Max] in, and its probability distribution density is p (V).With a plurality of threshold voltages this interval is divided into 2 ⁿIndividual minizone, and from bottom to up respectively to each interval numbering (from 1 to 2 ⁿ).Then each interval width is

$D_v=\frac{V_{\max }-V_{\min }}{2^n}$

I interval is [V _Min+ (i-1) D _v, V _Min+ iD _v], 0＜i≤2 wherein ⁿWe stipulate: when noise dropped in i the interval, when being odd number as if i, random code generator was output as " 1 " (or " 0 "), when i is even number, and output " 0 " (or " 1 ").So promptly the random noise of input is digitized as random code.

The multi-threshold examination is meant the threshold voltage of comparer in the comparator array is separately positioned on the above-mentioned threshold voltage that like this, the output of comparator array and input noise voltage have relation one to one.When noise drops in i the interval, there be i comparer to be output as high level, have 2 ⁿ-i comparer is output as low level.

Demoder is meant and can decodes to the output of comparer, makes that when noise dropped in i the interval, when being odd number as if i, random code generator was output as " 1 " (or " 0 "), and when i was even number, output " 0 " (or " 1 ") promptly got real random code output.

Analog-digital converter is the noise signal digitizing, is divided into two groups according to its parity then, the corresponding random code " 1 " (or " 0 ") of odd number group, the corresponding random code " 0 " (or " 1 ") of even number set.On behalf of noise signal, the lowest order of analog-digital converter be in odd number still is the even number interval, has so realized the function of comparator array and demoder with analog-digital converter.

Screen by multi-threshold and to make the information entropy autostable of the random code exported.Computing formula by information entropy:

$H\left(x\right)=-\underset{i=1}{\overset{n}{\mathrm{\Σ}}}p\left(x_i\right){\log }_{2}p\left(x_i\right)$

P (x in the formula _i) be the probability of " 0 " and " 1 " in the random code.When " 0 " and " 1 " etc. generally occurred, it is maximum that the entropy of random code reaches, and maximum uncertainty is arranged.The multi-threshold mode of screening can make information entropy keep bigger value, and it is analyzed as follows:

The noise voltage of input drops on i interval probability

$P\left(i\right)={\∫}_{V_{\min }+i{D}_v}^{V_{\min }+(i-1)D_v}p\left(V\right)\mathrm{dV}$

When n is enough big, i.e. D _vEnough hour,

p[V _min+(i-1)D _v]≈p(V _min+iD _v)

Then

P(i)≈D _vp(V _min+iD _v)

So,

P(i)≈P(i+1)

From the above mentioned, establishing input noise voltage, to drop on i be that the probability in the interval of odd number is P _O, the probability that drops on the even number interval is P _C, then

P _O=P _CAnd when element in the sample space of stochastic variable etc. generally occurs, the information entropy maximum.Like this, when n is enough big, just can obtain comprising the random code of maximum fault information.For the interval division of sixteen bit, need not any compensation or adjust circuit, the entropy of a random number can reach more than 0.99.

Screen the normalized autocorrelation coefficient of the random code that can reduce to export by multi-threshold.The related coefficient of random code is by the common decision of ssystem transfer function of the randomness in physical accidental source and transformation system.

Traditional simple gate limit discriminator circuit utilizes threshold friendship incident to produce random code output, the average time that positive threshold is handed in the unit interval

$\stackrel{\‾}{N}={\∫}_0^{\∞}\stackrel{\·}{x}{p}_2(x_0,\stackrel{\·}{x})d\stackrel{\·}{x}$

For Stationary Gauss Random process

$\stackrel{\‾}{N}=\frac{{\mathrm{\σ}}_{\stackrel{\·}{x}}}{2\mathrm{\π\π}}\exp [-\frac{{(x_0-m)}^2}{2{\mathrm{\σ}}^2}]$

For Stationary Gauss Random process, the average time that positive threshold is handed in the unit interval equals the average time that negative threshold is handed over, so the average time that total threshold of unit interval is handed over is 2 N.

For a stationary stochastic process time interval τ in office=| t _t-t ₁| in, related coefficient only is the function in the time interval with irrelevant constantly.

The related function of the random code of output is

$R_X(\mathrm{\τ\τ}=E[X\left(t_1\right)X\left(t_2\right)]$

$=1\·P_1$

$=\frac{1}{2}{e}^{-2\stackrel{\‾}{N}\mathrm{\τ}}[1+\frac{{\left(2\stackrel{\‾}{N}\mathrm{\τ}\right)}^2}{2!}+\frac{{\left(2\stackrel{\‾}{N}\mathrm{\τ}\right)}^4}{4!}..................]$

$=\frac{1}{4}+\frac{1}{4}{e}^{-4\stackrel{\‾}{N}\mathrm{\τ}}$

Its normalized autocorrelation functions is

r _x(τ)＝e ^{-4 Nτ}

N is the number of times that positive threshold is handed in the unit interval in the formula, for stationary Gaussian process, works as threshold value

V _a=m _xThe time, N=f is arranged ₀

In the formula, f ₀Average frequency for white Gaussian noise.So have

r _x(ττ＝exp[-4f ₀τ]

By following formula as can be known, the normalized autocorrelation functions of the random code of output is the period tau of random code and the average frequency f of white Gaussian noise ₀Function.

In the middle of engineering is used, when τ arrives to a certain degree greatly, if r _x(τ) very little, then can be similar to and think and do not had any association between X (t) and the X (t+ τ), statistics is independent.Here introduce τ correlation time ₀As τ＞τ ₀The time, then can think uncorrelated between X (t) and the X (t+ τ).τ ₀Be defined as

|r _x(τ ₀)|≤0.05

From the above analysis, when simple gate limit random code generator sample time greater than τ ₀, promptly

f ₀τ 〉=0.749 o'clock can think that between everybody of random code sequence of output be incoherent, promptly is output as real random code.But τ ₀When excessive, limited the generation speed of random code, need solve by the bandwidth that increases the random noise signal source, this just has higher requirement to the random noise signal source.

The multi-threshold discriminator circuit has a plurality of threshold voltages, and interior white Gaussian noise of unit interval drops on i zone and the number of times of the positive threshold friendship of generation on i+1 threshold voltage is

$\stackrel{\‾}{N_i}={\∫}_{D_v}^{(i+1)D_v}\mathrm{dV}{\∫}_0^{\frac{D_v}{\mathrm{\Δt}}}p(V,\stackrel{\·}{V})d\stackrel{\·}{V}$

As break over voltage D _vWhen not being very little, approximate has (being resolution)

$\stackrel{\‾}{N_i}=\frac{B}{2\mathrm{\π}}\exp -\frac{{[(i+1)D_v-m_V]}^2}{2{\mathrm{\σ}}^2}$

The average time that total positive threshold is handed in unit interval is

$\stackrel{\‾}{N_m}=\underset{i}{\mathrm{\Σ}}\stackrel{\‾}{N_i}$

Its normalized autocorrelation functions is

$r_x(\mathrm{\τ\τ}=\exp [-4\stackrel{\‾}{N_m}\mathrm{\τ}]$

Because

$\stackrel{\‾}{N_m}>>\stackrel{\‾}{N}$

So when adopting the multi-threshold comparer to screen, the normalized autocorrelation functions of the random code of output is prescribed a time limit much larger than the employing simple gate with the speed of period tau decay.So when identical sample frequency, the related coefficient of the random code that the multi-threshold discriminator circuit produces is less.Correlation time τ ₀Also corresponding having significantly reduced.

Screen the code check of the random code that improves output by multi-threshold.The maximum output speed of random code depends on τ correlation time of the random code of output ₀Voltage for multi-threshold is screened mode, as the average frequency f of white Gaussian noise ₀When remaining unchanged, because

Big in limited time a lot of than simple gate, thus correlation time τ ₀Reduce, Ma sampled point can become intensive immediately.

By merging the output of a plurality of random code generators, improve the code check of the random code of output.Because n random code generator produces n random code sequence, sequentially respectively get a random code sequence to final output from this n random code sequence at every turn, so circulation, thus n random code sequence merged into a random code sequence.Like this, the code check of output be single random code generator bit rate output n doubly.

By merging the output of a plurality of random code generators, improve the randomness of the random code of output.Because n random code generator produces n random code sequence, sequentially respectively get a random code sequence to final output from this n random code sequence at every turn, so circulation, thus n random code sequence merged into a random code sequence.Like this, under the condition of same bit rate output, the code check of single random code generator is reduced to original n/one times, sampling time interval τ increase for original n doubly, normalized correlation coefficient reduces greatly, randomness is improved.

The present invention compared with prior art has following advantage that beneficial effect is arranged:

(1) random code of the present invention output is a real random code, and the information entropy of the random code of output can autostable, and lower to the bandwidth requirement of noise source;

(2) normalized correlation coefficient of the random code of the present invention's output can reduce greatly by quantity that increases threshold voltage and the number that increases the random code generator unit that merges output, approaches zero;

(3) the present invention is by increasing the number of the random code generator unit that merges output, and the random code of output can reach great code check;

(4) working stability of the present invention is cheap, and favorable expandability is convenient to integrated.

Description of drawings

Fig. 1 is the structural representation of the device of generation real random code of the present invention;

Fig. 2 is the structural representation of the random code generator unit of apparatus of the present invention;

Fig. 3 is another structural representation of the random code generator unit of apparatus of the present invention.

Embodiment

In order to understand the present invention better, the present invention is done to describe further below in conjunction with accompanying drawing.

As shown in Figure 1, the device of generation real random code of the present invention comprises noise generator, several multi-threshold random code generators, data processor, noise generator links to each other respectively with each multi-threshold random code generator, form several random code generator unit, each multi-threshold random code generator links to each other with data processor.

As shown in Figure 2, the multi-threshold random code generator is an analog-digital converter, and analog-digital converter is exported with lowest order; As shown in Figure 3, the multi-threshold random code generator comprises comparator array and demoder, and noise generator, comparator array, demoder, data processor connect successively; Comparator array is the uses that are connected in parallel of several comparers, and noise generator links to each other respectively with each comparer, and comparator array is exported corresponding Transistor-Transistor Logic level, through obtaining binary real random code sequence after the decoder decode.

Data processor among Fig. 1 can be parallel-to-serial converter, also can be data collecting card.

Embodiment 1

Random noise source is selected the broadband white noise source for use, and model is HP346, and transformation system adopts comparator array, and the model that comparator array is made up of eight comparers is max961, and demoder adopts 741s148, output real random code sequence, and its code check is 1Mbit/s.

Embodiment 2

Random noise source is selected the broadband white noise source for use, and model is HP346; Transformation system adopts analogue-digital converter, and its model is max150; Parallel-to-serial converter is 16: 1 converters, and model is max3891; The white noise signal of HP346 output is screened through the multi-threshold of max150, output real random code sequence, and its code check is 1Mbit/s.The common random code sequence that produces of 16 HP346 and max150 is added to 16 parallel input ports of max3891 respectively, and at the serial output terminal mouth of max3891, obtaining code check is the random code output of 16Mbit/s.

Embodiment 3

Random noise source is selected the broadband white noise source for use, and model is HP346; Transformation system adopts analogue-digital converter, and its model is max150, and by data collecting card input random code sequence, the model of data collecting card is AD7812L.The white noise signal of HP346 output is screened through the multi-threshold of max150, output real random code sequence, and its code check is 1Mbit/s.The common random code sequence that produces of 8 HP346 and max150 is added to 8 digital input ports of AD7812 respectively, sequentially respectively get a random code sequence to final output from these 8 random code sequences at every turn, so circulation, thus 8 random code sequences are merged into a random code sequence.At the digital output port of AD7812L, obtaining code check is the random code output of 8Mbit/s.

Claims (10)

1. device that produces real random code, comprise noise generator, it is characterized in that also comprising several multi-threshold random code generators, data processor, noise generator links to each other respectively with each multi-threshold random code generator, form several random code generator unit, each multi-threshold random code generator links to each other with data processor.

2. the device of generation real random code according to claim 1 is characterized in that described multi-threshold random code generator is an analog-digital converter, with the lowest order of analog-digital converter as output.

3. the device of generation real random code according to claim 1 is characterized in that described multi-threshold random code generator comprises comparator array and demoder, and noise generator, comparator array, demoder, data processor connect successively.

4. the device of generation real random code according to claim 1 is characterized in that described data processor is parallel-to-serial converter or data collecting card.

5. the device of generation real random code according to claim 3 is characterized in that described comparator array is that several comparers are connected in parallel, and noise generator links to each other respectively with each comparer.

6. produce the method for real random code with the described device of claim 1, it is characterized in that process is as follows:

After screening by the multi-threshold random code generator, the level signal at random that noise generator is produced obtains binary real random code sequence, then in each cycle, binary real random code sequence that several random code generator unit produce is respectively got one, merge into a binary real random code sequence output by data processor.

7. the method for generation real random code according to claim 6, it is characterized in that the examination of the process of the level signal at random comparator array that described noise generator produces, comparator array is exported corresponding Transistor-Transistor Logic level, through obtaining binary real random code sequence after the decoder decode.

8. the method for generation real random code according to claim 6, it is characterized in that the examination of the process of the level signal at random analog-digital converter that described noise generator produces, as output, obtain binary real random code sequence with the lowest order of analog-digital converter.

9. the method for generation real random code according to claim 6 is characterized in that the noise signal digitizing, is divided into two groups according to its parity then, the corresponding random code " 1 " of odd number group, the corresponding random code " 0 " of even number set; The perhaps corresponding random code " 0 " of odd number group, the corresponding random code " 1 " of even number set.

10. the method for generation real random code according to claim 6 is characterized in that described data processor is parallel-to-serial converter or data collecting card.

Images