Accelerometer Based Body Area Network Sensor Authentication

Abstract

In recent years, the increasing number of wearable sensors by humans can serve for many purposes like emergency care, health care remote monitoring, personal entertainment and communication. In the case of health care applications, wearable sensors enable long-term continuous physiological monitoring without disturbing day to day activities, which is important for the treatment and management of many chronic illnesses, neurological disorders, and mental health issues. Examples include: diabetes, autism spectrum disorder (ASD), depression, drug addiction, and anxiety disorders. In general, wearable sensors can be used individually or as networked systems. In any case, sensors communication and authentication are key aspects in the deployment of these systems. A Wireless Body Area Network (WBAN) is created by wearing small sensors on the human body. To ensure authentication in on-body sensor networks, a mechanism which intuitively proves that all the communicating nodes are trusted ones is required. In order to achieve sensor authentication, in this paper a technological approach where accelerometer data gathered from sensors are used to distinguish whether or not the devices are carried on the same individual’s body, is presented. This system detects this situation by calculating the correlations between embedded accelerometers in a cellphone and accelerometer sensors. We evaluate our method over a dataset of fifteen wearers with sensors in various positions on their body.