AnUnscented-Transfor m-based Filtering Algorithm ForNoisyContaminated Chaotic Signals
Jiuchao Feng · 2006
VI' Abstract- Combining themodelling technique forsignal with chaos lX Yn theUnscented Transform (UT),a newfiltering algorithm is grator + realized. Itindicates by computersimulation thatthisnew Fig. 1.Blockdiagram showing filtering noisy chaotic signal. algorithm caneffectively reducenoiseon chaotic signal no materhowparameter ofchaosgenerator varies withtime.In comparison withtheEKF algorithm, this algorithm hasabetter filtering performance inthecaseoflowsignal tonoise (SNR), and foobseve noitime estbyausin fitering tenique hasthesimilar performance inthecase ofhighSNR.Inaddition, toobtain anoptmalestmation oftheoriginal dynamical anapplication inblind channel equalization isalso found. system(7)-(10). Mostoften usedtechnique forfiltering noisy chaotic signal istheExtended KalmanFilter (EKF)algorithm, I.INTRODUCTION whichmakesuseofthefirst orderoftheTaylor series Theproblem ofnoisy corrupted chaotic signals arises in expansion ofthenonlinear transform function usedtotransfer manyapplications. Forexample, whenmeasurements aretakenGaussian randomvariables. Because ofthisapproximation, fromachaotic physical process being, themeasuring devicethealgorithm usually causestheproblems attwoaspects: the introduces error intherecorded signal. Alternatively, theactualhigher complexity incomputation duetothecomputation for chaotic phenomenon maybeimmersed inanoisy environment, theJacabian matrix; andthelower estimation precision (11). asmightbethecaseifoneseeks todetect a lowpower Accordingly, itisthepurpose ofthis papertoimplement a chaotic signal (possibly usedforcommunications) thathas new filtering methodforhighly noisy contaminated chaotic beentransmitted overanoisy anddistorted channel. Inboth data. cases, wehavetoattempt topurify ordetect achaotic signal