Statistical inference attack against PHY-layer key extraction and countermeasures

The establishment of secure secret keys ahead transmissions is one the key issues in field information security. security traditional cryptographic mechanisms seriously challenged by computing-intensive attacks, with fast growth high-performance computing. As an alternative, considerable efforts have been made to develop physical (PHY) layer measures recent years, such as link-signature-based (LSB) extraction techniques. Those believed secure, based on fundamental assumption that wireless signals received at two locations are uncorrelated when separated more than half a wavelength. However, this does not hold some circumstances under latest observations, rendering LSB vulnerable attacks. To address problem, formal theoretical analysis channel correlations both real indoor and outdoor environments provided paper. Moreover, paper proposes empirical statistical inference attacks (SIA) against extraction, whereby adversary infers signature target link. Consequently, extracted from has recovered observing surrounding links. In contrast prior literature assumes link-correlation models for inference, our study make any link correlation. Instead, we employ machine learning (ML) methods empirically measured signatures. We further propose countermeasure SIAs, called forward-backward cooperative protocol helpers (FBCH). FBCH, (other trusted nodes) introduced provide randomness extraction. Our experimental results shown proposed still quite effective even without making assumptions Furthermore, effectiveness FBCH validated experiment results.