Improvement in performance of ternary sequence using binary step size LMS algorithm

Abstract

Pulse compression is one of the standard signal processing techniques that is widely used in various applications to achieve desired range resolution by using a matched filter and large detection range. Better performance was achieved with adaptive filtering. Least mean square algorithm is one of the stochastic gradient algorithms used in the communication system. The convergence characteristics of LMS adaptive filter is related to the autocorrelation of the input sequence. In this paper, the input sequences are considered as good ternary chaotic sequences. But the main drawback of this algorithm is its relatively slow rate of convergence. The drawback of slow convergence has overcome by using binary step size least mean square (BSSLMS) algorithm, where two variable step sizes are used in the computation of LMS. In this method, the output of the matched filter is processed through least mean square adaptive filter with binary step size LMS algorithm. These two variable step sizes are continuously changing in each iteration. By using this method superior performance in terms of peak sidelobe ratio, autocorrelation sidelobe peak is obtained and a better convergence performance is achieved as compared to LMS algorithm.