Identification in Encrypted Wireless Networks Using Supervised Learning

In recent years, not only has the number of wireless devices significantly increased, but also their level of integration into daily life. Devices ranging from laptops and cell phones to cameras and TVs are now connected to networks. As the ability to secure these devices advances, public and private organizations are adopting and establishing both public and private wireless networks. Wireless networks ease this integration, but not without cost. The nature of this medium presents challenges. This work aims to demonstrate and codify a mechanism by which we can increase our ability to verify and validate the identity of the device through encrypted data observation. This paper focuses on device identification. Multiple supervised learning techniques were vetted and a reference implementation was constructed and executed using real traffic. Incremental learning methods were identified as the classification mechanism of choice for streaming data.