Identity-Based Attack Detection In Mobile Wireless Networks

Identity-based attacks (IBAs) are one of the most serious threats to wireless networks. Recently, received signal strength (RSS) based detection mechanisms were proposed to detect IBAs in static networks. Although mobility is an inherent property of wireless networks, limited work has addressed IBA detection in mobile scenarios. In this paper, we propose a novel RSS based technique, Reciprocal Channel Variation-based Identification (RCVI), to detect IBAs in mobile wireless networks. RCVI takes advantage of the location decorrelation, randomness, and reciprocity of the wireless fading channel to decide if all packets come from a single sender or more. If the packets are only coming from the genuine sender, the RSS variations reported by the sender should be correlated with the receiver's observations. Otherwise, the correlation should be degraded, then an attack can be flagged. We evaluate RCVI through theoretical analysis, and validate it through experiments using off-the-shelf 802.11 devices under different attacking patterns in real indoor and outdoor mobile scenarios. We show that RCVI can detect IBAs with a high probability even when the attacker is half a meter away from the genuine user.