Research On Theory And Application Of Weak Signal Detection Based On Stochastic Resonance Mechanism

Ubiquitous noise in signal processing has extremely bad effects on the extraction and detection of weak signals.How to eliminate noise interference has always been one of the hottest research topics in weak signal detection.However,most of the researches attempt to analyze the statistical properties of weak signals and noise based on information theory,electronics and physics methods and construct filters to extract weak signals.Nevertheless,when the signal and the noise band are aliased,the filtering method based on the noise-canceling idea will surely suffer from two aspects: firstly,it is difficult to detect the weak signal with low signal-to-noise ratio(SNR);secondly,the detection will inevitably cause the signal damage or the information loss.Stochastic resonance(SR)can overcome the shortcomings of traditional noise filtering methods with its unique noise utilization characteristics.It turns people's inherent idea from noise-damaged system performance into noise-enhanced signal transmission,which is very suitable for detecting weak signals under strong background noise.Therefore,the study of weak signal detection based on SR has important practical significance.According to bistable potential well model and threshold sensing model,this paper extends from a single SR system to a parallel array SR system and an adaptive array SR system.This paper investigates the extraction and detection of periodic signals,aperiodic binary pulse amplitude modulation(BPAM)signals,image signals and discrete digital signals in low SNR environments.The main research contents and contributions are as follows:1.The influence of delay feedback and potential function asymmetry on the bistable SR system driven by weak periodic signal and additive Gaussian white noise is studied.The system output SNR expression is derived under this condition.It is shown that the output SNR of the system depends not only on the asymmetry of the potential structure but also on the delay feedback.The anomalous characteristic that the asymmetric SR is better than the symmetric SR depends on the magnitude of the feedback strength.In addition,under the same size of delay,asymmetry of the potential well width can counteract the strong background noise more effectively than the depth asymmetry of the well.It also shows that the asymmetric bistable SR system with well width has better robustness to noise.2.The parallel array bistable SR communication system(P-BSR-CS)is studied,and an optimal and robust system parameter assignment scheme is proposed to improve the detection performance of weak BPAM input signal.The parallel array bistable SR module structure design makes the optimal combination of the system parameters within a cluster of optimal parameters selected to improve the robustness of system parameter selection,meanwhile,the design achieves superior system performance with weak BPAM signal transmission at very low SNR environments.Theoretical analysis and simulation results show that the proposed parameter allocation scheme is superior to other existing schemes in the performance of bit error rate(BER)and channel capacity(CC),which can guarantee the signal transmission requirement at very low SNR environments.3.Aiming at the noise interference in the transmission of two-dimensional image signals,a low peak signal-to-noise ratio(PSNR)grayscale image restoration enhancement method based on optimal adaptive array bistable SR is proposed.In this method,the Hilbert scan is used to reduce the dimension process of the grayscale image and convert the two-dimensional image signal into a one-dimensional BPAM signal.Then,we use the adaptive array bistable SR module to enhance the low SNR BPAM signal and transform the enhanced signal into a two-dimensional grayscale image signal to obtain the restored grayscale image.Simulation results show that the method proposed in this paper is better than the classical image restoration methods on the grayscale level and PSNR of the restored image,i.e.,mean filter,Wiener filter and median filter.Larger array number brings better image restoration.4.Inspired by the presence of multiplicative noise in sensing systems such as biological synapses can enhance weak stimulus transmission,the detection performance of binary discrete digital signal in the threshold sensing system is studied under the combined effect of background additive noise and external multiplicative noise.First,the background additive noise usually obeys Gaussian distribution,aiming at the four different external multiplicative noise scenes of Gaussian,uniform,Laplace and Cauchy,we utilize the maximum posteriori(MAP)criteria to study the detection problems of discrete digital signal and derive the system output error detection probability(EDP)expression.The results show that the EDP curves appear nonmonotonic with the increase of the external multiplicative noise intensity;the external multiplicative noise is more robust to the detection performance of the threshold sensing system than the background additive noise.Since the detection performance of a single threshold sensing system cannot reach the ideal state at very low SNR,we extend the research of noise-enhanced weak discrete signal detection to the array threshold sensing network.It is found that with the increase of the number of system array,the performance of binary discrete signal detection is also improving;when the number of array is large enough,the performance of the system tends to be saturated.It is very important to choose the appropriate array number.This study contributes to the design of sensor systems and to a better understanding of the biological mechanisms of synaptic transmission.