Detecting malicious morphological alterations of ECG signals in body sensor networks

Abstract

Body Sensor Network (BSN) -- a network of body-worn wireless health monitoring sensors -- have a tremendous potential to remove the space and time restrictions on health management. Given the importance of the data BSNs collect for improved health outcomes, securing the data from unauthorized tampering is essential. A compromised (or externally influenced) sensor in a BSN may generate erroneous patient data leading to, among other things, wrong diagnosis and treatment. In this paper, we present a novel approach to address the problem of detecting maliciously induced morphological alterations in the ECG signal (i.e., inducing changes to its shape). Our approach works by correlating the ECG signals with synchronously measured arterial blood pressure (ABP) signal measured using a distinct (and un-compromised) sensor. Initial analysis of our system demonstrates promising results, with 99.75% accuracy in detecting ECG signal morphological alterations for healthy patients with normal sinus rhythms.