Multiresolution wavelet denoising for ultra-wideband time-of-arrival estimation with regularized least squares

Improving accuracy in wireless localization and ranging is a challenging task which often demands an increase in the signal-to-noise ratio (SNR). Impulsive ultra-wideband (UWB) technology is a promising signaling alternative that is capable of high-resolution ranging with minimal cost on SNR. Unfortunately, typical UWB time-of-arrival (ToA) estimators are complicated and perform poorly in the low SNR environment. In this paper, we propose a regularized least squares (RLS) approach with wavelet denoising to improve the estimator accuracy at low SNR. Our algorithm estimates the ToA as a by-product of the RLS channel estimator based on a thresholding technique, which is simple and can enable fast, on-the-fly, accurate ToA estimation applicable to real-time application. In addition to devising a threshold selection framework based on the constant false alarm (CFA) criterion, we demonstrate the robustness of our algorithm first by computer simulation, then applying it to a realistic situation of range estimation via the UWB impulse radio. In both cases, our algorithm is shown to supersede both high-resolution algorithms and energy detector in ToA estimation and computational complexity when the sampling rate is available. Furthermore, the CFA criterion is shown to improve the estimator performance as compared to the fixed threshold selection strategy.