On time-of-arrival estimation in dense ultra-wideband channels

We derive a pilot-assisted maximum likelihood (ML) timing estimator for ultra-wideband radio systems for the realistic scenario in which the received pulse shape and the channel are unknown. We compare the performance of the ML timing estimator to the Cramer-Rao bound and to sub-optimal methods such as the dirty-template method and a delay-correlate timing estimator for transmitted reference systems, both of which do not require knowledge of the channel and received pulse shape. The performance results indicate that, at medium-to-high signal-to-noise ratios, the NIL estimator significantly outperforms the sub-optimal schemes, especially in non-line-of-sight channels.