Securing wireless implantable medical devices using electrocardiogram signals

Implantable Medical Devices (IMDs), such as pacemakers and cardiac defibrillators, can perform a variety of health monitoring and therapeutic functions. A wireless module has become an intrinsic part of many modern IMDs for parameter configuration and medical data transmission. However, such wireless modules can be manipulated to compromise a patient's safety or privacy by eavesdropping or by sending unauthorized commands. A unique challenge in this scenario is that doctors who are not pre-authorized need to have access to the IMDs in an emergency situation. In this thesis, we study the use of electrocardiogram (ECG) signals for securing the IMDs. Blood circulation system in the body is regarded as an inborn secure channel to transmit ECG signals to the IMD and to its external programmer simultaneously. Measurements extracted from the ECG signal, e.g., inter-pulse intervals (IPIs) and random binary sequences (BSes), are used for security purposes. In an emergency situation, doctors can gain access to the IMDs by measuring the patient's real-time ECG signal. However, adversaries cannot achieve this as long as they do not have any physical contact with the patient. We design two ECG-based key distribution schemes based on a fuzzy commitment primitive and a fuzzy vault primitive, respectively. Using the schemes, doctors can obtain a symmetric key by measuring the patient's real-time ECG signal. We also compare these two schemes from different perspectives and discuss their advantages and disadvantages. In order to provide information-theoretically unbreakable encryption for the IMDs, we design an ECG based Data Encryption (EDE) scheme. This scheme combines two well-known techniques of classic One-Time Pads (OTPs) and error correcting codes. Meanwhile, in order to improve the effciency of the BS generation, we develop an ECG Multiple Fiducial-points based Binary Sequence Generation (MFBSG) algorithm. Existing methods solely rely on ECG IPIs to produce BSes and hence introduce unacceptable levels of latency. On the other hand, the proposed algorithm uses five distinct feature values from one heartbeat cycle. By doing this, the time required to generate a BS is reduced, and we achieve the key design goal of low-latency. In conclusion, this thesis explores the use of ECG signals for securing the IMDs. The proposed ECG-based schemes can solve the unique challenge prevalent in this environment.