A matching performance study on IPI-based entity identifiers for body sensor network security

To address the critical problem of symmetric key distribution in the emerging wireless body sensor network, a family of novel biometrics methods has been recently proposed to achieve automatic key distribution. As the core of such biometrics methods, the quality of entity identifiers generated from bio-signals determines the maximal performance that can be achieved in such systems. This work evaluates the matching performance of inter-pulse interval based entity identifier, which is the most efficient one among the existing proposals, using the experimental data collected from 85 clinical subjects. The results of statistical analysis show that the error distribution of entity identifiers from genuine communicating pairs has a certain pattern, which should be taken into consideration while designing the consequent fuzzy identification scheme.