Ultra-wideband (UWB) is one of the fledging paradigms that the next generation of wireless personal area network (WPAN) is developing towards. With its appealing dimensions, UWB communication comes with a set of challenges of which channel estimation is prominent. In this article, we explore the problem of channel estimation in ECMA-368-based multiband orthogonal frequency division multiplexing (MB-OFDM) UWB system affected by unknown narrowband interference (NBI). We analytically show that the unknown NBI causes larger variance of a channel estimator. Thus, MB-OFDM UWB systems are supposed to experience significant performance degradation. To reduce this effect, we follow a new preamble-based channel estimation approach, where the interference power on each subcarrier is treated as a nuisance parameter which is averaged out from the corresponding likelihood function. The later is then optimized in an iterative manner according to the quasi-Newton algorithm. Furthermore, we address highly-accurate channel estimation in a time-variant channel and propose an iterative interpolation method. Simulation results indicate that our proposed approaches provide more precise estimation performances than those offered by conventional channel estimation methods.