Promotors: prof.dr. J.J. Lukkien (TU/e) and prof.dr.ir. J.P.M.G. Linnartz (TU/e)
Co-promotor: dr.ir. P.H.F.M. Verhoeven (TU/e)
Technische Universiteit Eindhoven
Date: March 4
A Body Area Sensor Network (BASN), in most deployments, consists of sensors mounted on or implanted in the human body and a more powerful device named the body hub. Many BASNs have been developed for a broad variety of applications such as fitness, gaming, ambient assisted living, dietary advice or infant and elderly monitoring. However, there are still challenges that arise from the domain of applications, devices and wireless communication, which limit the usage of the BASNs. For example, many complete applications are currently specialized services and they are implemented based on specific devices and communication protocols. It, therefore, is difficult to reuse or adapt the BASN to new requirements; applications cannot access each other’s information and in many cases, the data acquired is not even accessible to users. Besides, the users’ acceptance of BASNs will largely be determined by the confidence that human subjects have in the trustworthiness, security and privacy of such systems. The transparency of the underlying architecture can be an important factor. For example, the human subject can transparently own and manage his BASN, and he can also understand by whom and for what purposes the data is collected. Moreover, it is vital that the system remains functioning under any changes.
This dissertation aims to explore the underlying key questions for the architecture of BASNs. We focus on system software aspects of BASNs and not on the details of the sensing technology. In particular, we focus on two key concerns of the software architecture, flexibility and trustworthiness, which are addressed in the design. In our context, the flexibility refers to capabilities of adding sensors as well as changing their configuration, dynamic loading of applications, and sharing resources and data. The trustworthiness of the system refers to system aspects including dependability, data quality, and security and privacy. We first examine related work and global concerns of the software architecture of a BASN and analyze requirements of different stakeholders through a set of use cases. We then present the architecture through a set of view models. A layered and modular architecture is proposed and architectural trade-offs are analyzed.
To address the concern of flexibility, concepts of component-based systems and downloadable software components are employed for managing changes in the system, for example, adding a new service or application. A new sensor is attached by downloading a driver indicated by a QR code. An approach of semantic interoperability is also adopted to facilitate the resources and data sharing among system components. Moreover, we employ design patterns and provide design decisions as well as concrete technical solutions. The architecture is evaluated through an actual platform implementation, which consists of multiple sensors and an Android smart phone. Several sensor drivers, services and applications are also implemented for this purpose.
To address the concern of trustworthiness, we propose a trust management model, in which trustworthiness can be evaluated and monitored at component, application and system levels with respect to quality properties like dependability, security and performance. Moreover, a trust evaluation framework through monitoring, which is based on subjective logic, is proposed to address the question of how to derive trust from observations of certain properties. The framework is further applied to evaluate the trustworthiness of sensor data such as electrocardiography and galvanic skin response.