The major challenge to accurate Time of Arrival (TOA) based indoor localization systems is ranging in non-line of sight (NLOS) environments. The high probability of Direct Path (DP) blockage causes large errors that cannot be mitigated by increasing system bandwidth alone. In this paper we propose a novel ultra wideband (UWB) multi-band (MB)-orthogonal frequency division multiplexing (OFDM) ranging technique that identifies and mitigates blockage in indoor environments through frequency diversity. Based on empirical UWB measurement data collected in a typical office environment we first analyze the correlation between the occurrences of DP blockage to the TOA estimation variations across different system sub-bands. Then we propose a hypothesis test to identify the occurrence of blockage. Frequency diversity is then used to minimize the estimation error in the absence and presence of DP blockage. Results show that employing MB-OFDM diversity outperforms single sub-band and large bandwidth estimations.