Quantum Random

8/3/2019 Quantum Random

1/24

Design and Implementation of a QuantumDesign and Implementation of a Quantum

True Random Number GeneratorTrue Random Number Generator


2/24

What is True Randomness?What is True Randomness?

Must be unpredictable.Must be unpredictable.

For a given set of binary data 'i', the i+1For a given set of binary data 'i', the i+1 thth

bit can only bebit can only be

predicted with 50% accuracy.predicted with 50% accuracy.

Must be unbiased and non algorithmic. Computers can't doMust be unbiased and non algorithmic. Computers can't do

this (and neither can you!).this (and neither can you!).

Useful for cryptography, science, and games (gambling andUseful for cryptography, science, and games (gambling and

drinking).drinking).


3/24

Types of Random Number GeneratorsTypes of Random Number Generators

Pseudorandom (PRNG): Uses an algorithm and a 'secret'Pseudorandom (PRNG): Uses an algorithm and a 'secret'

initial sequence. This is what your computer does.initial sequence. This is what your computer does.

True Random (TRNG): Samples a physical system of highTrue Random (TRNG): Samples a physical system of high

entropy.entropy.

Both are easy to design wrong and they fail silently!Both are easy to design wrong and they fail silently!


4/24

Types of TRNGTypes of TRNG

Non Quantum: Samples a complex system of highNon Quantum: Samples a complex system of high

entropy (lavalamps, time between keypresses).entropy (lavalamps, time between keypresses).

Higher bandwidth, easier to construct, numbers areHigher bandwidth, easier to construct, numbers are

not produced by any obvious algorithm.not produced by any obvious algorithm.

Is complexity as good as randomness?Is complexity as good as randomness?


5/24

Types of TRNGTypes of TRNG

Quantum TRNG (QTRNG): Samples a simple system ofQuantum TRNG (QTRNG): Samples a simple system of

high entropy (behavior of single photons or particles).high entropy (behavior of single photons or particles).

Low bandwidth, difficult to sample quantum levelLow bandwidth, difficult to sample quantum level

phenomenae.phenomenae.

However, the output 'should' be truly random!However, the output 'should' be truly random!


6/24

Mistakes to AvoidMistakes to Avoid

Do not use more than oneDo not use more than one

entropy source or detector.entropy source or detector.

The author of this paperThe author of this paper

11

suggests a good method too.suggests a good method too.


7/24

Mistakes to AvoidMistakes to Avoid

Do not use a counter + CPUDo not use a counter + CPU

interrupts. The number ofinterrupts. The number of

events within a given time isevents within a given time is

not a random distribution, itnot a random distribution, itis a Poisson distributionis a Poisson distribution2

2..

Do not try to calculateDo not try to calculate

'expected' time between'expected' time betweenevents, either.events, either.


8/24

Our DesignOur Design


9/24


A PIN photodiode and opamps are used as a solid stateA PIN photodiode and opamps are used as a solid state

particle detector that operates at low voltages. It wasparticle detector that operates at low voltages. It was

enclosed in a Faraday cage.enclosed in a Faraday cage.


10/24

Flex PCBFlex PCB

Boards were printed on flex PCB for its good rapidBoards were printed on flex PCB for its good rapid

prototyping characteristics.prototyping characteristics.


11/24


Pulse shaping is done by a Schmitt trigger hex inverter.Pulse shaping is done by a Schmitt trigger hex inverter.


12/24


13/24


Sampling and parallelSampling and parallel

output are done by anoutput are done by an

ATtiny261 MCU @ATtiny261 MCU @

8Mhz.8Mhz.


14/24



15/24

DemonstrationDemonstration

Hopefully no magic blue smokeHopefully no magic blue smoke


16/24

Basic Analysis of OutputBasic Analysis of Output

Check for bias by creating simple frequency charts.Check for bias by creating simple frequency charts.

X means of Y random values should approach the binomialX means of Y random values should approach the binomial

distribution for large X and Ydistribution for large X and Y33..

Keep in mind that 'proving' randomness is impossibleKeep in mind that 'proving' randomness is impossible

through hypothesis testing.through hypothesis testing.


17/24


18/24


19/24

Advanced AnalysisAdvanced Analysis

NIST Random Number Generation Technical WorkingNIST Random Number Generation Technical Working

Group Statistical Test Suite 2.0Group Statistical Test Suite 2.044

100 megabits of data were used for these tests, with default100 megabits of data were used for these tests, with default

options and a=0.01options and a=0.01

The last of 188 tests (linear complexity) did not seem to runThe last of 188 tests (linear complexity) did not seem to run

correctly.correctly.


20/24

Advanced AnalysisAdvanced Analysis

Given a block of random data divided into X bitstreams andGiven a block of random data divided into X bitstreams and

subjected to Y tests at some threshold alpha value forsubjected to Y tests at some threshold alpha value for

faliure Z, you expect true random data to fail X*Y*Z tests.faliure Z, you expect true random data to fail X*Y*Z tests.

Minimum 1/alpha bitstreams required (our alpha=0.01)Minimum 1/alpha bitstreams required (our alpha=0.01)

We expect 187 faliures, we observe 205.We expect 187 faliures, we observe 205.

Faliures do not cluster on any specific test.Faliures do not cluster on any specific test.


21/24

Future Technology!Future Technology!

Single photon QTRNGs using single photon emitters andSingle photon QTRNGs using single photon emitters and

photomultiplier tubes (Ebay!)photomultiplier tubes (Ebay!)

High bandwidth, zero bias! No isotopes! Requires highHigh bandwidth, zero bias! No isotopes! Requires high

voltage and vacuum tubes (seriously).voltage and vacuum tubes (seriously).


22/24

AcknowledgementsAcknowledgements

Foulab for a space to work, and friends to workFoulab for a space to work, and friends to work

with.with.

TRNG driver and data management scriptsTRNG driver and data management scriptswritten by fxwritten by fx


23/24

Want to make one yourself?Want to make one yourself?

If all goes well, boards will be available in oneIf all goes well, boards will be available in one

month atmonth at www.legionheavyindustries.comwww.legionheavyindustries.com
http://www.legionheavyindustries.com/http://www.legionheavyindustries.com/http://www.legionheavyindustries.com/


24/24

ReferencesReferences

1:1: http://isi.cbs.nl/iamamember/CD2/pdf/545.PDFhttp://isi.cbs.nl/iamamember/CD2/pdf/545.PDF

2:2: http://en.wikipedia.org/wiki/Poisson_processhttp://en.wikipedia.org/wiki/Poisson_process

3:3: http://en.wikipedia.org/wiki/Binomial_distributionhttp://en.wikipedia.org/wiki/Binomial_distribution

4:4: http://csrc.nist.gov/groups/ST/toolkit/rng/index.htmlhttp://csrc.nist.gov/groups/ST/toolkit/rng/index.html

For more information:For more information:

http://www.national.com/onlineseminar/2004/photodiode/PhotodiodeAmplifers.pdfhttp://www.national.com/onlineseminar/2004/photodiode/PhotodiodeAmplifers.pdf

http://jp.hamamatsu.com/resources/products/ssd/pdf/s1223_series_kpin1050e01.pdfhttp://jp.hamamatsu.com/resources/products/ssd/pdf/s1223_series_kpin1050e01.pdf

http://www.fourmilab.ch/hotbits/how3.htmlhttp://www.fourmilab.ch/hotbits/how3.html
http://isi.cbs.nl/iamamember/CD2/pdf/545.PDFhttp://isi.cbs.nl/iamamember/CD2/pdf/545.PDFhttp://en.wikipedia.org/wiki/Poisson_processhttp://en.wikipedia.org/wiki/Poisson_processhttp://en.wikipedia.org/wiki/Binomial_distributionhttp://en.wikipedia.org/wiki/Binomial_distributionhttp://csrc.nist.gov/groups/ST/toolkit/rng/index.htmlhttp://csrc.nist.gov/groups/ST/toolkit/rng/index.htmlhttp://www.national.com/onlineseminar/2004/photodiode/PhotodiodeAmplifers.pdfhttp://www.national.com/onlineseminar/2004/photodiode/PhotodiodeAmplifers.pdfhttp://jp.hamamatsu.com/resources/products/ssd/pdf/s1223_series_kpin1050e01.pdfhttp://jp.hamamatsu.com/resources/products/ssd/pdf/s1223_series_kpin1050e01.pdfhttp://www.fourmilab.ch/hotbits/how3.htmlhttp://www.fourmilab.ch/hotbits/how3.htmlhttp://www.fourmilab.ch/hotbits/how3.htmlhttp://jp.hamamatsu.com/resources/products/ssd/pdf/s1223_series_kpin1050e01.pdfhttp://www.national.com/onlineseminar/2004/photodiode/PhotodiodeAmplifers.pdfhttp://csrc.nist.gov/groups/ST/toolkit/rng/index.htmlhttp://en.wikipedia.org/wiki/Binomial_distributionhttp://en.wikipedia.org/wiki/Poisson_processhttp://isi.cbs.nl/iamamember/CD2/pdf/545.PDF

Documents

Quantum Random