In this paper a novel idea of estimating the 3D geometry and building material of indoor environments is proposed. The idea is based on assumption that the received radio signal is distorted due to interaction with the surrounding objects and thus includes the signature of the radio environment. We describe the methodology how to apply radio environment signatures for characterization of geometry and building material of indoor environments applying machine learning tools using ray tracing simulations and ultra-wide band communication technology. The idea is based on the assumption that the emerging broadband wireless systems applies multiple antenna and is able to estimate the channel impulse response for each transmit-receive antenna pair.