UC Irvine

Body Area Sensor Networks (BANs) utilize wireless communicating sensor nodes attached to a human body for convenience, safety, and health applications.

Physiological characteristics of the body, such as the heart rate or ECG signals, are promising means to simplify the setup process and to improve security of BANs.

This thesis describes the design and implementation steps required to realize an ECG-based authentication protocol to identify sensor nodes attached to the same human body.

This thesis starts by reviewing the latest literature in the filed of Body Area Network, and looks at possible applications, challenges, and overall architecture of BANs.

It also addresses the design of a body-area sensor system, including the hardware setup, analogue and digital signal processing, and required ECG feature detection techniques.

A model-based design flow is applied, and strengths and limitations of each design step are discussed.

Real-world measured data originating from the implemented sensor system then are used to set up and parametrize a novel physiological authentication protocol for BANs.

The authentication protocol utilizes statistical properties of expected and detected deviations

to limit the number of false positive and false negative authentication attempts.

The result of the described holistic design effort is the first practical implementation of biometric authentication in BANs that reflects timing and data uncertainties in the physical and cyber parts of the system.