Chaotic communication in alpha-stable noise channel using robust estimators

The thesis study investigates the behavior of non-coherent chaotic communication system under non-Gaussian noise. Due to broadband nature of chaotic signals, conventional band pass filters cannot be applied for noise reduction. The robust estimators are utilized to filter impulsive non-Gaussian noise effect on chaotic signal. These estimators are called as Median, Myriad and Meridian estimators. The non- Gaussian noise is modeled to have 𝛼 −stable distribution. The performances of these estimators are given in terms of error probability for varying characteristic exponent which tunes impulsiveness of the channel noise. The estimators are used for time varying signal by assigning a sliding window function having certain length. It is shown that even if the selection of window length directly affects the amount of noise removal from the signal, all of the robust estimators provide a certain bit error rate performance improvement. It should be noted that the linearity parameter of Myriad and medianity parameter of Meridian filter should be assigned properly to achieve satisfactory bit error rate improvement.