Building efficient wireless infrastructures for pervasive computing environments

Pervasive computing is an emerging concept that thoroughly brings computing devices and the consequent technology into people\u27s daily life and activities. Most of these computing devices are very small, sometimes even invisible , and often embedded into the objects surrounding people. In addition, these devices usually are not isolated, but networked with each other through wireless channels so that people can easily control and access them. In the architecture of pervasive computing systems, these small and networked computing devices form a wireless infrastructure layer to support various functionalities in the upper application layer.;In practical applications, the wireless infrastructure often plays a role of data provider in a query/reply model, i.e., applications issue a query requesting certain data and the underlying wireless infrastructure is responsible for replying to the query. This dissertation has focused on the most critical issue of efficiency in designing such a wireless infrastructure. In particular, our problem resides in two domains depending on different definitions of efficiency. The first definition is time efficiency, i.e., how quickly a query can be replied. Many applications, especially real-time applications, require prompt response to a query as the consequent operations may be affected by the prior delay. The second definition is energy efficiency which is extremely important for the pervasive computing devices powered by batteries. Above all, our design goal is to reply to a query from applications quickly and with low energy cost.;This dissertation has investigated two representative wireless infrastructures, sensor networks and RFID systems, both of which can serve applications with useful information about the environments. We have comprehensively explored various important and representative problems from both algorithmic and experimental perspectives including efficient network architecture design and efficient protocols for basic queries and complicated data mining queries. The major design challenges of achieving efficiency are the massive amount of data involved in a query and the extremely limited resources and capability each small device possesses. We have proposed novel and efficient solutions with intensive evaluation. Compared to the prior work, this dissertation has identified a few important new problems and the proposed solutions significantly improve the performance in terms of time efficiency and energy efficiency. Our work also provides referrable insights and appropriate methodology to other similar problems in the research community