| Resumo: | Biometrics are a field of study with relevant developments in the last decade. Specifically, electrocardiogram (ECG) based biometrics are now deemed a reliable source of identification. One of the major advances in this technology was the improvements in off-the-person authentication, by requiring nothing more than dry electrodes or conductive fabrics to acquire an ECG signal in a non-intrusive way through the user’s hands. However, identification still has a relatively poor performance when using large user databases. In this dissertation we suggest using ECG authentication associated with a smartphone security token in order to improve performance and decrease the time required for the recognition. We develop this technique in a user authentication scenario for a Windows login. We developed our solution using both normal Bluetooth (BT) and Bluetooth Low Energy (BLE) technologies to preserve phone battery; also, we develop apps for Windows Phone and Android, due to limitations detected. Additionally, we took advantage of the Intel Edison’s mobility features to create a more versatile environment. Results proved our solution to be possible. We executed a series of tests, through which we observed an improvement in authentication times when compared to a simple ECG identification scenario. Also, ECG performance in terms of false-negatives and false-positives is also increased.
|
|---|