The Prediction And Simulation Of Frequency Hopping Sequences Based On Bayesian Networks

With the rapid development of the communication technology, the traditionalsingle frequency communications can’t satisfied the requirement of the moderntelecommunication． Frequency Hopping(FH) communication is widely used in thefields of military and civilian communication, because of its better ability inanti-interference and net-working．At the same time it is pose grim challenge to thecounter measures of communication technology． The FH sequences which are used tocontrol the carrier frequency changes have an important role in FH communicationsystem． When taken communication jamming, if we can predict the regularity ofchanging for the FH sequences effectively, a lot of interference of bandwidth and powerwill be saved． Thus, for the prediction of FH sequences, A Bayesian network modelapplied to FH sequences prediction is proposed． The main contributions included inthe dissertation are summarized as follows．First, According to the chaotic characteristics in the low dimensional space such ascomplex pseudo random and nonlinearity, we analyze the necessity and feasibility ofthe reconstruction of phase space, and discuss the choice of optimal embeddingdimension and time delay in detail． The experiment shows that the reconstruction ofphase space can expand the original kinetics characteristics of FH sequences, and thephase space reconstruction system is differentiable homeomorphism with raw system,besides the prediction of FH sequences in high dimension phase-space is equivalent tolow dimension phase-space．Second, A Bayesian network model apply to FH sequences single-step prediction isproposed based on the phase-space reconstruction． We regard entire phase-space as apriori data information to learn Bayesian networks． Then we have the prediction offrequency hopping using the Bayesian network inference algorithm． Through theexperiment of the FH sequences for the four chaotic systems, the experiment resultsshow that A Bayesian network model method can provide prediction effectively．Final, we analysis the principle of the FH sequences multi-step prediction andstudy the best model of the data quantity to improve the real-time of the predictionmodel． Experiment result indicate that the accuracy slightly different between themulti-step prediction and single-step prediction to a certain step size． So, The Bayesian network prediction model is better than the other model in the FH sequences multi-stepprediction．...