JP2001075782A - Semiconductor physical random-number generating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a real physical random number at a necessary generating speed and epoch-making low cost by amplifying thermal noise by an amplifier and extracting it as a pulse signal, measuring a time interval between this random pulse signal and the pulse signal which is generated successively to the pulse signal, and supplying a measured value as the physical random number. SOLUTION: Thermal noise of a diode 2 is previously amplified by a preamplifier 3 and converted into an analog pulse signal. The pulse signal amplified by the preamplifier 3 is further amplified by a main amplifier 4 and converted into a voltage signal of several volts which can easily be detected. The voltage signal from the main amplifier 4 is supplied to a wave height discriminator 5 to select the pulse signal with a necessary wave height value, which is shaped by a waveform shaper 6 into the rectangular pulse signal. A time interval between the pulse signal which is outputted successively from the waveform shaper 6 and the pulse signal which is generated next is measured by a time measurer 7 by using clock pulses from a timer pulse generator 8.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a physical random number generation device for amplifying thermal noise of a semiconductor and extracting it as a pulse signal to generate a true random number. Alternatively, the present invention relates to a physical random number generation device which is used in a game machine to make fraud impossible.

[0002]

2. Description of the Related Art As a conventional physical random number generation system, Japanese Patent Application Nos. 9-32439 and 9-3638 by the present inventor, which are made by detecting a random event of nuclear decay.
No. 9 and Japanese Patent Application Hei.

A physical random number generator utilizing the decay of a radiation source can easily generate a true physical random number. However, since a radiation source is used, the manufacturing process is complicated and the cost is high. There were drawbacks such as the above regulations and difficulty in gaining public understanding of the use of radioactive materials.
A physical random number generator that generates light by detecting light from a light emitting diode (LED) with a photodiode can generate true physical random numbers at a relatively low cost. Technology development was needed.

[0004]

SUMMARY OF THE INVENTION In the present application, an existing inexpensive electronic component and an existing simple technical configuration are used to generate a true physical random number at a required generation speed at an epoch-making low cost. To any field that requires a physical random number.

[0005]

The first aspect of the present invention utilizes the fact that among various types of noise generated by a semiconductor, thermal noise based on thermal motion of a semiconductor carrier is a random phenomenon. The thermal noise is amplified by an amplifier and taken out as a random pulse signal, the time interval between the random pulse signal and the next successively generated pulse signal is measured, and this measured value is supplied as a physical random number.

According to a second aspect of the present invention, in measuring a time interval between random pulse signals generated by amplifying thermal noise of a semiconductor, a peak value of the random pulse signal to be measured is determined by peak height discrimination. A true random number can be freely generated at a required random number generation speed by selecting a threshold value using a device.

In the third aspect, as a source of thermal noise, thermal noise of all kinds of semiconductor elements such as a silicon photodiode can be used, but the most inexpensive circuit diode is used as a source of thermal noise.

According to a fourth aspect of the present invention, semiconductors and other circuit systems, which are sources of thermal noise, are integrated and sealed in various types of IC cards or IC chips having information processing circuits. To provide computer communication encryption,
It can be used for personal authentication in mail-order sales and transactions involving various IC cards, as a probability generator for gaming machines, and the like.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Noise generated from a semiconductor includes Johnson noise based on thermal motion of carriers, so-called 1 / f noise generated in relation to the surface state of the semiconductor, and localized levels in the semiconductor. Gr noise resulting from the generation and recombination of the carrier is known. The present invention takes advantage of the fact that Johnson noise is the largest of these noises, Johnson noise is based on thermal motion, and thermal motion is random.

An embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 shows the overall configuration of a semiconductor physical random number generator.
Shown in The semiconductor physical random number generation device 1 includes a silicon diode 2, a preamplifier 3, a main amplifier 4, a wave height discriminator 5, a waveform shaper 6, a time measuring device 7, a clock pulse generator 8, a random number generator 9, a random number storage device. Is built in the controller 10.

Since the silicon diode 2 which is a source of the thermal noise may be any kind of semiconductor, a commercially available inexpensive circuit diode is used.

The diode and all the circuit systems are sized to be integrated and sealed in an IC card or IC chip.

The thermal noise of the diode 2 is amplified by the preamplifier 3 and converted into an analog pulse signal. The pulse signal amplified by the preamplifier 3 is converted by the main amplifier 4 into a voltage signal of several volts which is further amplified and easily detected. The voltage signal from the main amplifier 4 selects a pulse signal of a required peak value by a peak value discriminator 5 and shapes it into a rectangular pulse signal by a waveform shaper 5.

FIG. 2 shows a commercially available inexpensive circuit diode 2.
Is a peak value distribution 20 of a pulse signal obtained by amplifying the thermal noise of FIG. As shown in FIG. 2, the pulse signal generated from the thermal noise and the noise of the electronic circuit system are sufficiently separated.

In FIG. 2, the required number of pulses per second can be freely generated by changing the threshold of the wave height discriminator 5 as indicated by the arrow A or B. For example, when a threshold is set for the arrow A in the peak value distribution of FIG.

A time interval t between a pulse signal continuously output from the waveform shaper 6 and a pulse signal generated next time.
Is measured using the clock pulse from the clock pulse generator 8 and the time measuring device 7. The time interval t is
It is represented by the number of clock pulses. FIG. 3 shows an experimental result of the frequency distribution of the time interval t when the number of generated pulse signals is 500 per second.

The distribution of the experimental values indicated by black circles in FIG.
As shown by the solid line in FIG. 3, the exponential distribution exp (-t / T
o). Here, To is an average value of the time interval t, and To = (1 / n) = (1/500) = 2
Milliseconds. The average value is indicated by an arrow in FIG. The exponential distribution of the frequency distribution of the time interval of the pulse signal generated by amplifying the thermal noise of the diode indicates that the thermal noise is a random phenomenon that follows the Poisson distribution. It guarantees the validity of using measured values for random numbers.

Assuming that the clock frequency of the clock pulse generator 8 for measuring the time interval is F hertz, the time unit of the measurement is (1 / F) seconds. That is, the time interval t is set to (1
/ F). In the measurement of the time interval t, for example, when the number of measurements exceeds 256 using an 8-bit (256) counter, that is, the time interval t> (25
In the case of 6 / F), 256 measurements are repeated and are represented by the remaining number of measurements. For example, in the case where the time interval t is 2,000, 2,000 = 256 × 7 + 208, so the measured number, that is, the random number is 208.

FIG. 4 shows a frequency distribution 30 of the number of occurrences of pulse signals when the threshold value is fixed to the arrow B in the peak value distribution of FIG. FIG. 4 shows a frequency distribution when the number of pulse signals generated in 10 seconds is measured and this measurement is repeated 2000 times.

The average value (the number of peak counts) in FIG.
0,450 (1,045 per second). The deviation value σ in this case is σ = √ (10,450) = 102. σ
A Gaussian distribution having a value of 102 is shown by a broken curve in FIG. The fact that the thermal noise occurrence frequency distribution of a semiconductor is a Gaussian distribution guarantees that the thermal noise of the semiconductor is a random phenomenon and that the thermal noise can be used as a random number.

The time interval t is set to a clock frequency of 11.25M.
It is measured with a (mega) hertz clock and this measurement is used to calculate
FIG. 5 shows the frequency distribution of 8-bit random numbers when bit random numbers are generated. In FIG. 5, the clock frequency is set to 11.25.
By fixing the frequency to M hertz and changing the threshold value of the pulse height discriminator 5, the frequency of generation of the pulse signal is changed from 160 bits per second (20 counts per second) to 8,000 bits per second (1,000 counts per second). .

As shown in FIG. 5, even if the frequency of pulse generation is greatly changed, the distribution of random number generation frequency is uniform.

A small semiconductor physical random number generator capable of uniformly generating true random numbers at an arbitrary required generation rate is mounted on an IC card, or mounted on a circuit board of a computer or other equipment. It is used for encryption processing of computer communication, personal authentication in mail-order sales, and commercial transactions involving various IC cards. In addition, it can be used as a probability generator for gaming machines in which impropriety is impossible.

[Brief description of the drawings]

FIG. 1 is a block circuit diagram showing an overall configuration of a semiconductor physical random number generation device according to the present invention.

FIG. 2 is an experimental graph of a peak value distribution diagram of a random pulse generated by amplifying thermal noise of a diode in the embodiment.

FIG. 3 is an experimental graph showing that a frequency distribution of a time interval between a random pulse signal generated by amplifying thermal noise of a diode and a random pulse signal generated next has an exponential distribution in the embodiment.

FIG. 4 is an experimental graph showing that the frequency distribution of the number of pulses per 10 seconds of the random pulse signal generated by amplifying the thermal noise of the diode in the embodiment becomes a Gaussian distribution.

FIG. 5 shows a case where the clock frequency for measuring the time interval of the random pulse signal in the embodiment is fixed at 11.25 MHz, even if the random number generation rate per second n is changed from 160 bits to 8000 bits. 9 is an experimental graph showing that true random numbers can be generated at various generation speeds.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 semiconductor physical random number generator 2 diode 3 preamplifier 4 main amplifier 5 wave height discriminator 6 waveform shaper 7 time measuring device 8 clock pulse generator 9 random number generator 10 random number storage and controller 20 wave height analyzer 30 pulse number measurement vessel

Claims (4)

[Claims] 1. Among various types of noise generated by a semiconductor, the thermal noise is a random phenomenon based on the thermal motion of a carrier of the semiconductor. A semiconductor physical random number generating apparatus which measures a time interval between the random pulse signal and a pulse signal generated next successively and supplies the measured value as a physical random number. . 2. In the measurement of a time interval between the random pulse signal and a signal generated next successively, a peak value of a random pulse signal to be measured is selected as a threshold value by a pulse height discriminator. 2. The semiconductor physical random number generation device according to claim 1, wherein the required random number generation speed can be selected freely. 3. The type of semiconductor that is a source of the thermal noise is as follows.
2. The semiconductor physical random number generator according to claim 1, wherein the thermal noise of all the semiconductor elements can be used, but the cheapest semiconductor such as a circuit diode can be used as a source of the thermal noise. 4. The semiconductor and other circuit systems are integrated, encapsulated in various IC cards or IC chips provided with an information processing circuit, and the physical random numbers are supplied to the information processing circuit to perform computer communication. 2. The semiconductor physical random number generation device according to claim 1, wherein the semiconductor physical random number generation device can be used for encryption processing, authentication of a person in a mail order or a transaction involving various IC cards, a probability generator of a game machine, and the like.