Research On Time Delay Signature Estimation Method Of Chaotic System Based On Optoelectronic Reservoir Computing

With the development of optical fiber communication technology,the security performance of optical communication system is also concerned.Chaos communication is a physical layer encryption method widely used in the field of secure communication with strong anti-interference and high security.When the parameters of transmitter and receiver match,chaotic communication can be realized through chaotic synchronization.Time delay is very important in chaotic communication systems and can even be used to reconstruct chaotic systems.At present,Time Delay Signature(TDS)parameters are generally estimated by mathematical statistics,but with the increase of nonlinearity in chaotic systems,the effectiveness of the method decreases obviously,and the amount of data required is relatively large.The TDS estimation method based on machine learning has improved the recognition accuracy,but it needs to obtain some prior knowledge of chaotic systems in advance,and the algorithm structure is complex and the calculation time is long.In order to improve the effectiveness and efficiency of TDS parameter estimation method and cope with a more complex application environment,this thesis proposes and verifies the delay feature estimation method based on reservoir computing.The main content of the thesis includes:(1)This thesis proposes time delay estimation method based on reservoir computing,two-dimensional image reconstruction method of time delay in the time series of chaotic systems TDS into two-dimensional image samples,using the reservoir computing of network in terms of computational efficiency advantage of image delay sample classification recognition,successfully by means of traversing the estimate the TDS in photoelectric chaotic system parameters.Under normal experimental conditions,the recognition rate of image samples is 100%.For extremely strong nonlinear systems,it can still reach 92.2%.Experimental results demonstrate the effectiveness of the proposed method in estimating TDS parameters.(2)This thesis proposes a model-free time delay signature estimation method based on reservoir computing,directly from the light in the time series of chaotic systems to identify parameters of TDS,using the reservoir computing network model between the sequence and the delay time sequence,with mean square error as the evaluation index to explore the relationship of time delay and the TDS parameters,so as to estimate the success photoelectric TDS parameters of chaotic systems.Experimental results show that this method requires less data than traditional methods.In addition,this thesis conducted experiments on four internal parameters of reservoir computing,including feedback coefficient,signal input gain,delay and number of nodes in the reservoir,to study the influence of reservoir computing parameters on TDS estimation accuracy.