Dissertation > Excellent graduate degree dissertation topics show

Research on Non-cooperative Chaotic Signal Processing Techniques

Author: XiongBo
Tutor: GanLu
School: University of Electronic Science and Technology
Course: Signal and Information Processing
Keywords: phase trajectories recover chaos synchronization parameter estimation CD3S broken
CLC: TN911.7
Type: Master's thesis
Year: 2013
Downloads: 11
Quote: 0
Read: Download Dissertation

Abstract

Nonlinear sciences especially chaos science have got extensive uses in signalprocessing and communication, for its irregularity in time-domain and noise-likecharacteristic in frequency-domain. Different from signal processing under cooperativecommunication environment, an interpreter need to extract or recover signal blindly orhalf-blindly and non-cooperative-based chaos signal processing is proved to bedifficulty. In this thesis, three points that will be discussed in detail include: the blindrecover of phase trajectory for one-dimensional chaotic systems; synchronization basedparameter estimation of chaotic systems; the cryptanalysis of CD3S communicationsystems. These research points listed as follows:1To avoid the deficiencies of different kinds of denoising algorithms and thinkabout specific situations when clean phase trajectory or dynamic equations is all thatneeded, this paper proposed the Gaussian mixture model based algorithm to recover thephase trajectory of one-dimensional chaotic systems, with nothing is known previouslyand low signal-noise-ratio. The new algorithm models the attractor trajectory as atwo-dimensional random vector at first, and then the probability distribution of thevector will be approximated by Gaussian mixture models, each Gaussian kernel will beupdated by measurements under the criterion of maximization the expectation oflikelihood function. After a rough trajectory was estimated by previous steps, a threesection of polynomial will be applied to smooth the trajectory; better denoising effectswill be obtained at last.2Introduce two synchronization-based parameter estimation methods: the leastsquare based and synchronization stability theory based methods, which presentlyexisted into accelerated prediction synchronization model and retarded delaysynchronization model. Updating equations of parameters to be estimated is deducedwith these new synchronization models, converged parameters are obtained combinedwith synchronized systems, this paper also analyzed the estimation performance andconvergence rate.3Two CD3S broken algorithms were put forward, which include DM-UKF method and IMM-UKF method. The previous algorithm applies a generalizedsynchronization system, and different filtering models are built corresponding todifferent information symbol values, both filters work simultaneously to fit theintercepted CD3S signal, and the best matched filter of that time point will be decidedby fitting errors, thus recover information symbols indirectly. For IMM-UKF brokenmethod, the CD3S signal was modeled as the output of two alternatively operatedsubsystems, drive equations of subsystems are decided by corresponding symbol values.The same as DM-UKF, fitting errors are used for determining the best matchedsubsystem, and thus extract symbol values. In both methods, an adaptiveerror-controlling-factor is introduced to reduce the impact of noise and fitting process.Compared with DM-UKF algorithm, IMM-UKF method not only preserves theadvantage of decreased influence of noise, multipath fading and low spreading factor,but can also calculate the spreading factor of CD3S signals based on model probability.

Related Dissertations

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Communication theory > Signal processing
© 2012 www.DissertationTopic.Net Mobile