A compensation method to improve the performance of IPI-based entity recognition system in body sensor networks

Security of wireless body sensor networks (BSNs) with telemedicine applications remains a crucial issue. A family of novel biometrics schemes has been recently proposed for node recognition and cryptographic key distribution without any pre-deployment in BSNs, where dynamic entity identifiers (EIs) generated from physiological signals captured by individual sensor nodes are used for nodes to recognize each other. As the recognition performance of EIs determines the maximal performance that can be achieved in such biometric systems, several kinds of EI generation schemes have been proposed. The inter-pulse intervals based EI generation scheme is more promising for such applications in actual scenarios because of its acceptable recognition performance. However, it was found that such generated EIs by true pairs of nodes, i.e. two nodes from the same BSN, have some error pattern which could be considered while doing node recognition or key distribution for an improved success rate. To address the problem, this work proposes an error-correcting code based compensation method which can be easily combined together with the key distribution process to achieve an improved recognition performance. Results of statistical analysis with experimental data collected from 14 subjects show that the bit difference between EIs from true pairs of nodes can be effectively reduced with the proposed method.