Wireless device identification from a phase noise prospective

As wireless devices become increasingly pervasive and essential, they are becoming both a target for attacks and the very weapon with which such an attack can be carried out. Wireless networks have to face new kinds of intrusion that had not been considered previously because they are linked to the open nature of wireless networks. In particular, device identity management and intrusion detection are two of the most significant challenges in any network security solution but they are paramount for any wireless local area networks (WLANs) because of the inherent non-exclusivity of the transmission medium. The physical layer of 802.11-based wireless communication does not offer security guarantee because any electromagnetic signal transmitted can be monitored, captured, and analyzed by any sufficiently motivated and equipped adversary within the 802.11 device's transmission range. What is required is a form of identification that is nonmalleable (cannot be spoofed easily). For this reason we have decided to focus on physical characteristics of the network interface card (NIC) to distinguish between different wireless users because it can provide an additional layer of security. The unique properties of the wireless medium are extremely useful to get an additional set of information that can be used to extend and enhance traditional security mechanisms. This approach is commonly referred to as radio frequency fingerprinting (RFF), i.e., determining specific characteristics (fingerprint) of a network device component. More precisely, our main goal is to prove the feasibility of exploiting phase noise in oscillators for fingerprinting design and overcome existing limitations of conventional approaches. The intuition behind our design is that the autonomous nature of oscillators among noisy physical systems makes them unique in their response to perturbations and none of the previous work has ever tried to take advantage of thi