Single-stable Stochastic Resonance System Applied To Digital Signal Detection

This thesis investigates the reception and detection of binary codes with single stable stochastic resonance (SR) system. As a bistable SR system is transformed to a single-stable one, the SR theory is simplified and more adaptive to numerical signal processing. During the base-band numerical signals' communication, e.g. PCM signals are transmitted directly at a long distance by retransmission, if the noise in channel is Gassian white noise (GWN), the receiver with matched filters is optimal. For the reception of binary code, a matched filter is equivalent to a integrator which can be substituted by sampling and summing, so the more the sampling number is, the higher the detection performance is. When a single-stable nonlinear dynamic system is connected to its front, we discover that this new system reduces the detection performance's susceptivity to the sampling number and makes a matched filter of fewer sampling number approach an ideal one (the most sampling number during simulations) in a sense of the detection performance. In addition, we discover that the way sampling in part can improve the detection performance further. In order that the single-stable system can receive signals under the optimal condition, we are concerned how to tune the system's response speed. After solving the FPK equation of the single-stable system and then analyzing the correlation between samples, we can obtain the expression of the detection performance. The expression includes two variables, i.e. GWN's density and the system's response speed, so we can find the optimal response speed after getting GWN's density by spectrum analyzing. With the help of the spectrum analyzer and the table of GWN's density versus the system's response speed, we can receive the signals synchronously, which is more effective than a self-adaptive receiver.Though present samplers' frequency can be high enough to make matched filters ideal, when receiving multi-nary codes or multi-channel signals, the reception efficiency will decrease remarkably with the sampling number increasing greatly. If a single-stable system is connected to the front of a matched filter, on the basis of keeping the detection performance almost unchanged, we can improve the reception efficiency through reducing the sampling frequency.Practical channels are not always Gaussian. For such ones, optimal receivers are locally optimal detectors (LOD) which are only adaptive to weak signals. If the received signals pass single-stable system first, strong signals can be reduced, so we can try comparing single-stable plus LOD system with the conventional non-Gaussian receiver to the extent of strong signals in future. The reception and detection of colored noise channels is also a subject worth being investigated.