|March 7, 2001|
|Noise can bring quiet surprise||Page One|
|By Kimberly Patch|
when background noise interferes with, say, a dinner conversation, the
only way to clearly hear a guest who won't speak up is to decrease the
Strictly speaking, noise is anything that introduces disorder, or random fluctuations into any type of signal, including audio, an electric current, or a temperature reading.
A pair of researchers has found mathematical evidence that in some cases it is possible to reduce noise in the output of a system by adding more noise.
The researchers looked at systems like electrical current that have intrinsic noise that makes the current fluctuate. "You want to have a given intensity for the current, but... in some places it is higher or lower than the average value [because] in some places the noise is high and in others [it] is low," said Jose Vilar, a postdoctoral researcher at Princeton.
By introducing more noise into to the system, the researchers were able to change the properties of the system so that the average value for the property they cared about -- the current -- held more steady. "Usually when you add noise you mess up everything but sometimes this doesn't happen. You can add noise to the system [without] adding noise to the output of the system," Vilar said.
The key to the phenomenon is that noise affects several parameters of a system. The trick is to change the system so the fluctuations are reduced in the particular signal, or output you care about -- for instance the current -- while fluctuations in other parameters may increase. "If you have a given current and if you add noise you can keep the same static current and the same average [value] but with less noise because you can keep the average properties of the system in one place and the noise properties [of] the system in other places," Vilar said.
The idea could eventually prove useful in very small electronic systems, where noise from the environment can have a large affect. It may also prove useful for tuning particular systems like microscopes when noise is a problem. "If you're trying to design something... it is not always possible to get rid of all the noise," said Vilar. In these cases it may be useful to know that "if you leave some noise in the system... the system can be less noisy than if you get rid of all the noise you can," he said.
Vilar noticed the mathematical principal when he was working on something else. "I just saw it. It's not intuitive, it's just based on mathematics," he said.
In essence, the principal shows that noise is not always bad, and it's effects are difficult to predict, said Vilar.
Because the concept is counterintuitive, it has been overlooked, but it may come into play in several contexts, said Vilar. For example, it is possible that biological systems may take advantage of this phenomenon to reduce noise within cellular networks, he said.
The research is intriguing and may eventually be promising at the nanoscale, according to Anthony Teolis, Vice President of Defense Systems at AIMS, Inc., and a contractor at the Naval Research Lab. "I see the research as a very first idea that may lead to viable noise suppression in the far future," he said.
The research points the way toward more research, Teolis said. The research dealt only with stationery systems that are in a certain state and are then perturbed in a way that makes them change continuously into another state. "The method needs to be extended to be applicable to dynamical systems before applications to nano-scales can be contemplated," he said. Dynamical systems are more complicated because they involve attributes that take place over time, like momentum.
In addition, more research is needed to determine how the phenomenon works in terms of the signal-to-noise ratio, Teolis said. Especially in electronics systems, the important attribute of a signal is not necessarily how strong it is, but how strong it is in proportion to the particular noise in a system that can obscure the signal.
Vilar's research colleague was Miguel Rubi of the University of Barcelona. They published the research in the February 5, 2001 issue of Physical Review Letters. The research was funded by the Spanish National Science Foundation (DGICYT).
Letters About TRN Feedback
Find out about TRN Research Update, our service for Web sites.