Time Delay Estimation from Multiband Radio Channel Samples in Nonuniform Noise

TL;DR

This paper introduces a high-resolution time-delay estimation method for multipath radio channels using multiband sampling with nonuniform noise, employing weighted subspace fitting and nonlinear least squares to outperform existing techniques.

Contribution

It proposes a novel algorithm for time-delay estimation from multiband samples with nonuniform noise, utilizing Hankel matrices and separable nonlinear least squares.

Findings

Algorithm nearly attains the Cramer Rao Lower Bound.

Outperforms multiresolution TOA, MI-MUSIC, and ESPRIT methods.

Supports high resolution from arbitrary number of bands.

Abstract

The multipath radio channel is considered to have a non-bandlimited channel impulse response. Therefore, it is challenging to achieve high resolution time-delay (TD) estimation of multipath components (MPCs) from bandlimited observations of communication signals. It this paper, we consider the problem of multiband channel sampling and TD estimation of MPCs. We assume that the nonideal multibranch receiver is used for multiband sampling, where the noise is nonuniform across the receiver branches. The resulting data model of Hankel matrices formed from acquired samples has multiple shift-invariance structures, and we propose an algorithm for TD estimation using weighted subspace fitting. The subspace fitting is formulated as a separable nonlinear least squares (NLS) problem, and it is solved using a variable projection method. The proposed algorithm supports high resolution TD estimation from an arbitrary number of bands, and it allows for nonuniform noise across the bands. Numerical simulations show that the algorithm almost attains the Cramer Rao Lower Bound, and it outperforms previously proposed methods such as multiresolution TOA, MI-MUSIC, and ESPRIT.