ON THE PERFORMANCE OF LINEAR ADAPTIVE FILTERS DRIVEN BY THE ERGODIC CHAOTIC LOGISTIC MAP

Abstract

Chaotic dynamical systems are increasingly considered for use in coding and transmission systems. This stems from their parameter sensitivity and spectral characteristics. The latter are relevant for channel estimation methods. In particular, the logistic map fλ = λx(1 - x) has been employed in chaotic coding and spread spectrum transmission systems. For λ = 4, the statistical properties of sequences generated by f4 are considered as ideal drive signals for channel estimation schemes. This assumption is proven in the present paper. To this end, the higher order statistical moments and the autocorrelation of time series generated by f4 are derived. It is shown that for λ = 4 the zero mean time series is uncorrelated. The adaptation performance of finite impulse response (FIR) digital adaptive filters (DAF) used for channel estimation is analyzed. It is shown that using zero mean sequences of f4 leads to the maximal possible FIR DAF performance. An optimal value for the damping parameter in the LMS scheme is derived that leads to the maximal performance and ensures stability. The analytic considerations are confirmed by simulation results.