Noise-aided SNR amplification by parallel arrays of sensors with saturation

We consider sensor devices with saturation in their response, in charge of the transmission of a sinusoidal signal buried in Gaussian white noise, with a performance assessed by an input-output gain in the signal-to-noise ratio. We show that such saturating devices can always be tuned to achieve a signal-to-noise ratio gain larger than unity. When replicated to form parallel arrays, further improvement of the gain can be obtained with independent noise injected on the sensors. This provides smart arrays of simple nonlinear sensors capable of acting as noise-aided amplifiers, and where the highest gains in signal-to-noise ratio are always obtained at a nonzero level of the added noises. (c) 2005 Elsevier B.V. All rights reserved.