BroAuth: evaluating different levels of visual feedback for 3D gesture-based authentication

Using digital gadgets we authenticate ourselves regularly. Usually authentication relies on standard PIN or password but novel input hardware facilitates new authentication techniques. In this work we present an authentication mechanism based on body movements captured by a depth sensor. This idea is motivated by the cultural body movements used as welcoming gestures, especially by gang members (secret handshakes). Our authentication technique 'BroAuth' lets the user interact with a virtual partner to perform password input. This is done through touching target zones on the own body and on the body of a virtual partner. In this paper we focus on evaluating usability and security of onscreen feedback for such a system. Three different types of feedback were tested during the input process: Text-only (1D), abstract user representation (2D) and a virtual avatar (live 3D). The most detailed but most insecure 3D feedback performed much worse than the abstract input modalities. Input times and user opinions show that an abstract 2D representation is the best tradeoff between usability and security for such a system.