Study On Characteristics Of Nonlinear Dynamics And Applications For Semiconductor Lasers With Random Distributed Feedback

In this paper,the generation and research advances of chaos using semiconductor laser(SL)are introduced.To solve the problem of security of chaotic communication and the transmission rate of encrypted data,a new scheme of unidirectionally coupled SLs(UCSLs)with random distributed feedbadk(RDF)was proposed.Furthermore,a new structure of mutually coupled SLs(MCSLs)with random optical injection was proposed,which not only suppresses time-delay signature(TDS)well,but also enhances the chaos bandwidth.In addition,the system can generate two-channel high-speed physical random number sequences at 200 Gbits/s for each channel.Finally,a reservoir computing(RC)system based on RDF-SL was proposed and numerically studied.The detailed work and novelty points are summarized as follows:(1)A new scheme of high-quality chaos generation based on RDF-UCSLs was proposed.The scheme can achieve TDS suppression and bandwidth enhancement simultaneously.In order to show its advantages,we carried out the comparision with UCSLs based on single reflection mirror feedback(SRF).The influences of system parameters on TDS suppression [characterized by autocorrelation function(ACF)and permutation entropy(PE)] and output bandwidth was discussed.The results show that:for SRF-UCSLs,only when the injection strength is small,a better TDS suppression can be obtained;if the injection strength exceeds a certain value,the TDS of the slave SL is obviously deteriorated due to the injection locking effect.This makes bandwidth of the slave SL in SRF-UCSLs is lower generally in the low TDS region.In contrast,in RDF-UCSLs,the slave SL can suppress TDS in a wider range of injection strength,and the bandwidth as high as ?24 GHz can be achieved.(2)A new scheme of chaos generation based on random optical injection for MCSLs was proposed.The influences of system parameters(including frequency detuning,injection strength,injected current)on TDS suppression and chaotic bandwidth were studied.The results show that,for appropriate coupling strength,the enhanced BW can be obtained with larger detuning;with the increase of bias current,the parameter region with larger bandwidth(>20 GHz)increases significantly,while the optimized TDS(<0.1)parameter region does not decrease clearly.In the optimized parameter region,this system can simultaneously achieve two chaotic outputs with bandwidth enhancement(>20 GHz)and TDS suppression.(3)The feasibility of generating two channel high-speed physical random number by random injection-based MCSLs was exploited.Due to the lower TDS,higher bandwidth,and weak cross-correlation function(CCF)under certain conditions,very lower statistical bias and ACF(exceep for zero delay)of the gnenerated random bits can be acheieved.For the selected system parameters,the binary sequences generated by the two SLs have passed all NIST test standards,and the rate of each channel is200 Gbits/s(=50 GS/s×4 LSB).Because the generated random bit sequences almost completely eliminate the cross-correlation,the system has the potential to generate400 Gbits/s ultra-high speed physical random number by crossly combining the random numbers generated by the two SLs.(4)A new RC system based on nonlinear dynamics of RDF-SLwas proposed.The influences of system parameters,such as sampling period,injection current,injection strength,feedback strength,frequency detuning and scaling factor,on time-series prediction task were studied.The results show that,the optimized performance is always achieved around the edge of bifurcation points;the area with good prediction performance is mainly arrived under negative detuning condition;under optimized parameters,lower normalized mean squared error(NMSE)and better prediction performance can be achieved to perform task of time-series prediction,due to the rich node states for the proposed RDF-SL.