Building an efficient wireless sensor network system for applied indoor research.

This thesis reasons on dynamic wireless sensor networks (WSN) analyzing different models and architectures. The main goal of all the work is the development of a tool designed to fulfill the needs of real on-field research, especially applied to indoor environment such as houses and hospitals. The idea was born from my last university thesis where alongside the original project of a remote controlled surgical room arose the need to monitor the hospital's patients and surgeons. In that occasion we were forced, for many reasons, to use a sort of batch system consisting of RFID tags and sensors with memory, whose data were downloaded after many hours or days, or simply sensors directly connected to computers. Instead a set of wireless sensors have allowed a real-time interaction with the remote controlled system. My thesis first shows a background of the wireless sensor network theory, technologies and security issues and shows how the work was developed. In the first period of time a developer starter kit was chosen from many available, evaluating different properties, including costs, open source or free IDE and the possibility of modify the base library supplied. An initial project was developed using this kit, modifying the nodes to install the required sensors. Following the main requests of the team, the network was created balancing the energy consumption and the reliability. The developing work was supported by field tests on real research scenarios of increasing complexity. The initial test results on the original project, developed with the supplied library, revealed weaknesses in battery life, nodes connection stability and security thus changes were made accordingly. An heartbeat system was designed and implemented to create a fault tolerant system consisting in a couple of devices. Then the security issues were evaluated in consideration of the increasing number of attack's techniques designed for the WSNs. A cryptography key protection mechanism was implemented to protect the AES algorithm itself and the data, together with a software deletion mechanism, in front of an hardware read access, to avoid the steal of the initial pass-phrases of the program itself. Finally some consideration were made on the WSN performance and study results. In conclusion we can say that our wireless network has been used successfully, although it has shown limits on stability and data flow capacity. In most cases the WSN capacity of disappearing and to be unattended for long time were preferred then the data stream reliability because of the comfort quality perceived by study’s subjects (i.e. house's inhabitants or patients).