Authors: I.Y. Lee, X. Liu, B. Kosko and C. Zhou
Affilation: University of Southern California, United States
Pages: 264 - 267
Keywords: carbon nanotubes, transistor, noise,
Noise can help signal detection at the nano-level. Experiments on a single-walled carbon nanotube transistor confirm that a threshold is sufficient for a nondynamical system to exhibit stochastic resonance: a judicious amount of noise can help a threshold-like nanotube transistor detect subthreshold signals while large amounts of noise wiped out the signals. The nanotube produced this stochastic-resonance effect using three types of synchronized discrete-time white noise and two performance measures: mutual information and input-output correlation. The experiments added Gaussian, uniform, and impulsive (Cauchy) noise. The electrical noise corrupted a random digital (Bernoulli) voltage sequence that acted as the subthreshold input for the nanotube transistor. The noisy signal stimulated the transistor's gate and produced a sequence of random output (Bernoulli) current in the nanotube. Shannon's mutual information and simple correlation measured the nanotube system's performance gain by comparing the input and output sequences. Both measures made no assumptions on the nanotube to ensure that the nanotube system generated the threshold stochastic resonance. This nanotube SR-effect was robust: it persisted even when infinite-variance Cauchy noise corrupted the signal stream. Such noise-enhanced signal processing at the nano-level promises applications to signal detection in wideband communication systems and biological and artificial neural networks.