On the performance of SVD estimation in Saleh-Valenzuela channel for UWB system

Abstract

This paper describes the performance of singular value decomposition (SVD) estimation techniques for multiband orthogonal frequency-division multiplexing (OFDM) of ultra-wideband (UWB) system in Saleh-Valenzuela channel modelling. The performance of SVD estimation is compared against LS and MMSE methods corresponding to zero-padded of a UWB system. The crucial limitation in UWB system is strictly low power transmission leads to many challenges in designing the system. However, most of the conventional channel estimation in frequency domain requires either pre-storing a large matrix or performing real-time matrix inversion, which should be avoided for UWB low complexity implementation. The intuition of this paper is to analyse further the complexity of SVD techniques against least squares (LS) and minimum mean square error (MMSE) estimation method. Consequently, this work implies that the SVD method gives an improvement of 3-5 dB if compared to the LS method while a significant complexity reduction of 57.8% as compared to MMSE methods.