Abstract
In conventional quantum key distribution protocols, the secure key is normally extracted from the measurement outcomes of the system. Here, a different approach is proposed, where the secure key is extracted from the measurement bases, rather than outcomes. Compared to the original Bennett-Brassard-1984 protocol, the proposed protocol involves no hardware change but modifications in data postprocessing. We show that this protocol is more robust against detector efficiency attacks and photon-number-splitting attacks when practical detectors and photon sources are used.