Physical Layer Security for In-Body Wireless Cardiac Sensor Network

Abstract

The thesis explores the physical layer security approaches for securing an in body multi-nodal leadless cardiac pacemaker (LCP) communication system. Pacemakers are implanted medical devices, used to treat different types of cardiac arrhythmias. The widely used version of these pacemakers is implanted with intravascular leads. Due to lead related complications, the next generation of pacemaker systems are becoming wireless i.e., connecting multiple nodes wirelessly without intravascular leads. Besides the unquestionable benefits of LCPs such as less invasive surgery, there are also some concerns associated with it. The wireless nature of these devices is a significant security risk and could lead to threats like eavesdropping, data tampering, and device modification. This thesis deals with the problem of quantifying the severity of risks associated with the wireless nature of these next generation LCPs and the corresponding countermeasures by utilizing the physical layer security (PLS) techniques. To evaluate the system eavesdropping risk without PLS, we use the concept of communication link outage probability. A link is said to be in an outage if the received signal to noise (SNR) ratio falls below the threshold required for error free decoding. We compute the eavesdropper (Eve) link outage probability for evaluation of eavesdropping risk with respect to the distance around the body. Similarly, for developing the corresponding countermeasures, we explore two different approaches of PLS for securing LCP. The first approach provides a secure communication strategy via channel modeling and offers data secrecy and reliability simultaneously, without use of data encryption. The second approach provides an alternative for symmetric key generation between legitimate nodes and avoids the use of key management and distribution servers as in the case of conventional cryptographic methods. For channel modeling strategy, our hypothesis is on the availability of positive secrecy capacity in the close proximity of the human body. Secrecy capacity is the performance metric that supports secrecy and reliability at the same time and is the maximum attainable secure communication rate without leakage of information to Eve. Secrecy capacity depends on the inherent noise within the wireless channels. To implement a channel modeling approach, prior knowledge about wireless channels is required and can only be implemented when the legitimate nodes have superior channel quality over Eve on the physical layer. To evaluate the secrecy capacity, the methodology of electromagnetic simulations and experimental measurements is adopted for modeling the in-body to in-body (legitimate) and in-body to off-body (Eve) wireless channels. The results show that the positive secrecy capacity is achievable within the human personal space of 25 cm, with practical antenna realizations. Furthermore, to examine the effect of electromagnetic radiations through the human body across different angles in three dimensional space, the spatial secrecy capacity is also evaluated. The angle from which the maximum leakage of information takes place is found to the left from front, just above the heart and is termed as the “Eve sweet spot angle”. Eve’s sweet spot angle has the least secrecy capacity among all the eavesdropper spatial positions with the human heart as a reference position. The results proved our hypothesis that the human body as a lossy medium for electromagnetic propagation inherently provides high attenuation to off-body Eve link, thus offering legitimate nodes an advantage on the physical layer for implementation of channel modeling approach. For solving the issues related to key management and distribution in case of traditional cryptographic algorithms, the dissertation also explores the source modeling approach to establish symmetric keys between legitimate nodes. The source modeling approach exploit the correlated information source between legitimate nodes for key generation. We hypothesized that the electromagnetic reflections experienced due to in-body transmissions provide enough randomness to generate a symmetric key from wireless parameters like received signal strength (RSS), phase, angle of arrival, etc. Therefore, we generated a symmetric key string between the in-body nodes by utilizing the randomness in the RSS measurements. Similarly, due to the availability of inherent physiological signals, the feasibility of symmetric group key establishment across multiple nodes of the leadless pacemaker system is also analyzed. Both methods provide viable alternatives with RSS based key generation method outperforming the other with a bit mismatch rate of approximately 1%.