Semiconductor Lasers Generate Better Random Numbers

December 16, 2008 By Laura Mgrdichian, feature

( -- Random numbers -- numbers without any pattern -- are vital to many applications, such as computer simulations, statistics, and cryptography. There are many ways to generate them using unpredictable physical processes, including electric-signal noise and radioactive decay, but these methods cannot produce the quantities of numbers needed to keep up with the high data-processing rates of today's computers.

A group of scientists seems to have discovered a way around this problem. They have found that the physical chaos present in semiconductor lasers -- laser light produced using a semiconductor as the medium -- can produce good-quality random number sequences at very high rates.

The scientists, from Takushoku University, Saitama University, and NTT Corporation, all in Japan, achieved random number rates of up to 1.7 gigabits per second (Gbps), which is about 10 times higher than the second-best rate, produced using a physical phenomenon. They report this result in the December issue of Nature Photonics.

"We have shown that the performance of random number generators can be greatly improved by using chaotic laser devices," said the paper's corresponding author Atsushi Uchida, a researcher at Saitama University, to "The rate we obtained is faster than that of any previously reported devices for generating random numbers using physical sources."

Fields and applications that could benefit from their work are numerous, including computational models to solve problems in nuclear medicine, computer graphic design, and finance. Random numbers are also important to internet security.

Generating random numbers using physical sources -- which can be as simple as coin-flipping and tossing dice -- are preferred over other methods, such as computer generation, because they yield nearly ideal random numbers: those that are unpredictable, unreproducible, and statistically unbiased.

Lasers, Uchida and his colleagues have shown, can be excellent physical sources if they are chaotic. This is achieved, in this case, by reflecting part of the laser light back into the laser using an external reflector. This induces chaos, causing the light intensity to oscillate wildly. As a result, the light's electromagnetic signals are highly complex and cover a wide frequency range.

The researchers used a pair of semiconductor lasers in their experimental setup. Each laser is connected to a photodetector, a device that senses and measures light, and each photodetector is connected to an analog-to-digital converter (ADC), which samples the physical light signals and outputs digital numbers. In this case, the specific ADCs convert the signals into random binary numbers suitable for computing and other high-speed data manipulation.

The group achieved a bit rate of 1.7 Gbps, although future work may center on devising laser schemes that can achieving rates as high as 10 Gbps.

Citation: Atsushi Uchida, Kazuya Amano, Masaki Inoue, Kunihito Hirano, Sunao Naito, Hiroyuki Someya, Isao Oowada, Takayuki Kurashige, Masaru Shiki, Shigeru Yoshimori, Kazuyuki Yoshimura and Peter Davis Nature Photonics vol, 2, no. 12, pp. 728-732 (2008); DOI:10.1038/nphoton.2008.227

Copyright 2007
All rights reserved. This material may not be published, broadcast, rewritten or redistributed in whole or part without the express written permission of

Explore further: Superinsulators to become scientists' quark playgrounds

Related Stories

Superinsulators to become scientists' quark playgrounds

January 31, 2019

Scientists widely accept the existence of quarks, the fundamental particles that make up protons and neutrons. But information about them is still elusive, since  their interaction is so strong that their direct detection ...

Graphene crinkles can be used as 'molecular zippers'

January 31, 2019

A decade ago, scientists noticed something very strange happening when buckyballs—soccer ball shaped carbon molecules—were dumped onto a certain type of multilayer graphene, a flat carbon nanomaterial. Rather than rolling ...

Recommended for you

A quantum magnet with a topological twist

February 22, 2019

Taking their name from an intricate Japanese basket pattern, kagome magnets are thought to have electronic properties that could be valuable for future quantum devices and applications. Theories predict that some electrons ...

Sculpting stable structures in pure liquids

February 21, 2019

Oscillating flow and light pulses can be used to create reconfigurable architecture in liquid crystals. Materials scientists can carefully engineer concerted microfluidic flows and localized optothermal fields to achieve ...


Adjust slider to filter visible comments by rank

Display comments: newest first

2.8 / 5 (5) Dec 16, 2008
Is anything in this world truly random?
3 / 5 (8) Dec 16, 2008
Of course. Quantum fluctuations are truly random to the best our current experimentation and theories can determine. Laser light is the epitomy of this, therefore this laser method of generating random numbers is perfect.
4.1 / 5 (9) Dec 16, 2008
Well these quantum fluctuations might seem random to us on this scale, however taking into consideration a whole universe of influence who can really say it is truly random.
3.3 / 5 (4) Dec 16, 2008
It is possible to produce numbers with deterministic processes (computational) that pass statistical tests for randomness. That indicates strongly that it might be the case that randomness is a problem of knowledge rather than an inherent property of natural events.
4.3 / 5 (9) Dec 16, 2008
It is not possible for any system of laws to produce a truly "random" outcome.

We can see this intuitively by simply picking two infinite sequences and adding them together.

No matter what you do, there is no term which is "random".

A person who does not know what the parent sequences are may believe the daughter sequence is "random" if it so happens he cannot figure it out, but he is only drawing this conclusion from a position of ignorance.

True randomness could only exist under one of the following conditions:

1) Total Chaos


2) An all powerful being outside the universe who is in no way subject to any of the laws of the universe intervenes for no reason that has anything to do with the existing laws of the universe. Even this might not even be considered random.


3) Human beings are somehow indowed with true free will by such an all powerful being, thus giving rise to "events" that are not caused by the starting conditions.


Otherwise, every event that has ever happened in the history of the universe is a direct, inescapable result of the starting conditions.
not rated yet Dec 16, 2008
can one random thing really be more random than another random thing?
not rated yet Dec 16, 2008
The decimals of pi are truly random.
Problem is you may need a random generator to find these decimals with for example a montecarlo process.
5 / 5 (1) Dec 17, 2008
Just because randomness is an artifact of 'physical laws' or truths about the universe, this does not mean that it is not random, it means that randomness is an artifact of those laws.
Chaos and order can both exist in the same universe, one which has predictable processes and unpredictable ones.
unpredictable in this context does not mean that we cannot predict them, it means they cannot be predicted, they are not ordered, these processes exist and are manifest.
There is no need to grasp after them being somehow 'ordered in a hidden way', any more than there is a need to grasp after ordered processes being chaotic fundamentally. While it may be the case that something may have both ordered and chaotic properties that does not make them the same thing, nor does it mean that the order is chaotic or that the randomness is ordered.

4.3 / 5 (3) Dec 17, 2008
Two unrelated comments:

1. It is a popular misconception that the vast majority of modern scientists embrace the Copenhagen interpretation, which, as it is casually described, "refutes the idea of hidden variables". The Copenhagen interpretation is actually NO interpretation. Its an assumption that the question, "Why are we not able to predict the outcome of a single QM experiment (except statistically)" is simply not to be asked. In essence, its the arrogant statement, "We can't predict the outcomes because they are unpredictable". Talk about arrogance! The idea of hidden variables is alive and well among scientists. Its just that they all realize we do not yet have the understanding to even try to attempt to answer the question, so we can only "shut up and calculate" for the moment.

2. There is no lack of fast random number generators, as this article implies. You can make a fast generator simply by using more parallel slow number generators. A noisy reverse-biased bipolar transistor can generate, in practice, several hundred million numbers per second. Use several of those transistors in parallel and you can supply numbers faster than any modern supercomputer.

5 / 5 (3) Dec 17, 2008
I stated very clearly that "to the best our current experimentation and theories can determine". No one knows *FOR SURE* whether there are hidden variables, but experimentation and theory seem to point to a realization that we can never predict them, because, after all, if we could do that, we'd have the ability to violate causation by sending classical information backwards in time. This forum is too small to explain how, but anyone that understands the delayed choice experiment and John Cramer's "reverse causality" experiment based on his transactional model of QM will understand this. Instinvively, I presume there ARE hidden variables and it disturbs me that both experimentation and logical theory seems to indicate the contrary. I like Einstein's statement, "God does not play dice with the universe". But, it appears both Einstein and my own instincts may be wrong.
not rated yet Dec 17, 2008
not rated yet Dec 17, 2008
To ShadowRam and Yes. Not onli "pi", all transcendental and irrational numbers have nonperiodic digits. But these numbers are most numbers. I.e. at least mathematical world abounds with random things. May be the physical world - also.
not rated yet Dec 18, 2008
The consideration that observably random processes are really ordered in a 'hidden' way is not so far removed from the logic of ID and such.
The fact that on occasion some scientists have asserted the randomness of something that has later been found to be ordered simply shows the danger in coming to illogical conclusions, it does not support the idea that an apparantly random process may be actually ordered.
Wemight as well say:
"I know it displays the properties and behaviour of evolution, but really it is design"
"O know it looks like an alloy of silver and gold but really it is strawberry jelly"
Damn, I swore I would not post in physorg threads ever since that 'plane on a conveyor' thing.
*wanders off muttering and shaking his head*

not rated yet Dec 18, 2008
How can we measure the degree of random? I mean how could we know that a random system is better than another?
Is it just by calculating the frequency of each generated number?
5 / 5 (1) Dec 18, 2008
@ Assaad33
That is an indicator, and a good one commonly used, but also a truly random number generator could come up with a series of 00000's arbitrarily long, they would just be randomly occuring zeros.
Getting back to the original point of the article it is important that the numbers produced are not an artifact of any ordered process, but are genuinely not ordered.
Then they may be used in encryption, for instance, without contributing to any weakness.

4 / 5 (1) Dec 18, 2008
@ Assaad33,
Yes, randomness can be measured via probability distributions, which strangely is an order.

It is also defined physically as entropy in thermodynamics (energy != work),... although if one thinks of a gas in a state of maximum entropy, it would appear perfectly ordered (!) in the sense of not having a arbitrary subjective useful order.
not rated yet Dec 20, 2008
On the comments hinting at a rigorous definition for "random":

In the context of 'random number generator' we simply want a certain distribution (although admittedly ~measuring~ that distribution can be problematic).

But in the context of philosophy, the definition of 'random' is more ambiguous. Compressed data can look very random and have a flat distribution (or even Gaussian, if certain coding is chosen) but unzip that file and you might have a Shakespeare sonnet. So philosophically speaking, the definition of 'random' must include the idea of usefulness (most all would agree a Shakespeare sonnet contains more order than the list of the first ten-thousand word from a dictionary).

not rated yet Dec 22, 2008
"Well these quantum fluctuations might seem random to us on this scale, however taking into consideration a whole universe of influence who can really say it is truly random."

True randomness is a fundamental part of quantum mechanics. It is possible to construct so called hidden variable theories that do not require the statistical nature of quantum mechanics, but they are non-local and even more hard to swallow than quantum mechanics. (local hidden variable theories have been definitively ruled out by Bell test experiments).
not rated yet Dec 22, 2008
How can we measure the degree of random? I mean how could we know that a random system is better than another?

Equidistributed and uncorrelated on all timescales and in all dimensions(e.g. if you take triplets of numbers from the old Linear Congruential Generators for pseudorandom numbers you find that all points lie on repeating equidistant planes).

With pseudorandom numbers, the only thing a computer program can generate with deterministic mathematical operations, you get a predictable sequence of numbers that satisfies many but not necessarily all statistical tests for randomness. If you know the internal state and functioning of the generator you can predict each value it spits out.

The abillity to generate the same sequence again by starting from the same seed value can be highly desirable in something like a game(e.g. if you want to be able to generate the same random maze or same random critter given only a single integer), somewhat useful in a monte-carlo simulation(mostly for debugging purposes) and a fatal flaw in cryptography.
not rated yet Dec 22, 2008
Soylent, your right that local hidden variables are a silly idea, but I strongly suspect non-local hidden variables are not so silly. People ignore Mach's principle because its old and hasn't helped us in anything. Yet ignoring it doesn't make it go away. By definition, an object that exists, interacts with us. For example, it would be silly for me to say something exists that is not interacting with us. Those interactions, (which are the mediators of all non-local concepts, like conservation of energy), might also be those hidden variables.
not rated yet Dec 22, 2008
I always wondered if a usb stick with random numbers on it or with a document containing book content on it would have different weight.

I guess the stick with random numbers has less weight.

I fear randomness is the highest order nature can get. We disorder randomness.
not rated yet Dec 22, 2008
We disorder randomness and then call it order. Are we crazy or what?
not rated yet Dec 23, 2008
Just some clarification on the idea of the mass of information, in physics: "Information" means simply the numbers, themselves--regardless of their degree of compression. So a random sequence contains the same minimum mass as a very orderly sequence. In other words, in this context the idea of data compression is not considered. 500 zeroes have the same minimum mass as 500 random numbers, and data-compressing those zeroes reduces the minimum mass accordingly. One thing that might help us realize that data compression is rather arbitrary in this context is this: Any data compression scheme includes prior knowledge of the scheme. For example, I might design a data compression scheme where the number 37 appearing in the stream represents Shakespeare's "The Tempest", but no other sequences are coded. This makes clear that data compression is more a contrived process. Even the so-called ideal coding theorems (best possible data compression) are arbitrary because they require some degree of prior knowledge of the scheme--which will always vary in efficiency and vary with type of data--since type of data MUST be an issue for any non-random data. That is, there is no non-random data that can't be 'categorized' into a type of data.
not rated yet Dec 23, 2008
Just some clarification on the idea of the mass of information, in physics:

So you know everything about the origin of gravity and mass. I am impressed.
not rated yet Dec 24, 2008
How can Pi be random? Its a measurement, not a built up number.
not rated yet Dec 24, 2008
"How can Pi be random? Its a measurement, not a built up number."

Depends on which definition of "random" is being used. See my Dec 20 post above on the two common definitions.
5 / 5 (1) Dec 24, 2008
Not pi itself is random.
The digits of pi are random.
like in 3.1415926535 etc. The sequence 31415926535 etc each digit is a random number in this precise sequence up to infinity.
not rated yet Dec 26, 2008
Is it possible that some of these comments are truly random? Seriously, probability theory both defines and sets forth tests (admittedly fruitful area for new theses...) of randomness. The definition is formal, like "justice" defined as due process. Both are disappointingly unsatisfactory for cocktail party discussions, but completely unambiguous and fully defined in their respective contexts.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.