Gaussian Decoupling Based Stochastic Resonance for Weak Optical Signal Detection

Weak signal detection methods can improve the SNR by recognizing the effective signals submerged in background noise. However, the existing stochastic resonance (SR) method is confined exclusively to the optimization of potential function parameters. Actually, the bit-error rate (BER) can be optimized by eliminating the correlation among SR output points. In this work, we propose a Gaussian decoupling based SR method to improve the detection performance of weak optical signals. We first perform the SR method on the received weak optical signal and approximate the probability distribution of the SR output points as a multi-dimensional Gaussian distribution. Then we apply a linear transformation based on eigenvalue decomposition of the Gaussian covariance matrix to decouple the correlation among SR output points. Numerical results show that the BER of Gaussian decoupling based SR method can be one order of magnitude lower than that of the ordinary SR method even when the SNR of input signal is −8 dB.