Short Term Prediction And Time Delayed Signature Extraction Of Chaotic Laser Based On Photonic Reservoir Computing

Chaotic laser have been extensively used in secure communications,high-resolution lidar and high-speed physical random key generation.External-cavity optical feedback semiconductor lasers are the first candidate as chaotic laser source,owing to its simple structure and the rich dynamic characteristics.However,with the in-depth exploration of the optical feedback semiconductor laser,there are two prominent problems in the chaotic laser: the one is not completely randomness,the methods of machine learning can be applied to the short-term prediction of chaotic laser;the other is the existence of the time-delay signature,which implies periodicity.The leakage of time-delay signature reduces the key space of chaotic secure communication system.Therefore,the short-term prediction and time-delay signature extraction of chaotic laser can be used as the security test standards for evaluating chaotic laser applications.In view of the above ideas,we studied the short-term prediction and time-delay signature extraction of chaotic laser in external cavity feedback semiconductor laser based on the photonic reservoir computing.At the same time,we showed the influence of the structure parameters of photonic reservoir computing on the output results.The specific work and results are briefly described as follows:(1)We proposed a method of short-term prediction of chaotic laser based on photonic reservoir computing.Firstly,the semiconductor laser with external cavity feedback is simulated.Through adjusting the system parameters,we can obtain the chaotic laser sequence to training and testing the prediction system.Secondly,the photonic reservoir computing is simulated as prediction system.By selecting the system parameters suitably,the photonic reservoir computing can effectively predict the dynamic trajectory of chaotic laser about 2 ns.Moreover,we also investigated the influence of critical parameters on prediction results,include the type of the mask,the quantity of the virtual nodes,the length of the training data,the input gain,the feedback strength,the injection strength,the ridge parameter and the leakage rate.(2)We proposed a method to extract time-delay signature of chaotic laser based on photonic reservoir computing.Firstly,the semiconductor laser with external cavity feedback is simulated,and the chaotic laser with the different feedback intensities are selected to establish the training sets and the test sets.By choosing the system parameters reasonably,the photonic reservoir computing can basically achieve the accuracy rate of more than 99%.Even for the chaotic laser with weak time-delay signature,the accuracy rate is over 91%.For the parameters of the photonic reservoir computing system,the influence of the number of virtual nodes on the recognition accuracy is mainly studied.It is worth pointing out that the photonic reservoir computing used in this article is a novel artificial neural networks,which consist of three parts: an input layer,a reservoir and an output layer.Only the output weights need to be trained and the others connection weights are generated randomly.The core idea of the photonic reservoir computing is using a single nonlinear node with a delayed feedback loop structure as the middle layer of traditional neural networks.Compared with traditional reservoir computing,it has the advantages of the simple configuration and suitable for hardware implementation.Semiconductor lasers with delayed optical feedback from an external mirror represent ideal candidates to implement photonic reservoir computing.Realizing information processing utilizing semiconductor lasers could lead to a paradigm shift in the field of photonic information processing,departing from traditional approaches toward novel concepts of machine learning.