Improving 802.11 fingerprinting of similar devices by cooperative fingerprinting

Abstract

Fingerprinting 802.11 devices has been proposed to identify devices in order to mitigate IEEE 802.11 weaknesses. However, important limitations prevent any real deployment. On the first hand, fingerprinting has a low accuracy when the devices have similar hardware and software. On the second hand, attackers may forge signatures to impersonate devices. We propose Diversity, a cooperative fingerprinting approach that improves accuracy of existing fingerprinting methods while relying only on off-the-shelf hardware. Diversity improves fingerprinting up to the reliable individual identification of identical 802.11 devices. This approach modifies the signature of devices by modifying slightly their traffic attributes. We evaluate Diversity with both a simulation and an implementation, achieving a false positive rate of 0% with a dataset including identical devices. Finally, we complement Diversity by mechanisms for detecting attackers that try to forge signatures.