PhD thesis, Universidad de Deusto, June 16th, 2014.
- Antonio F. Gómez-Skarmeta (University of Murcia, chair)
- George Roussos (University of London, vocal)
- Mikel Larrea (University of the Basque Country, secretary)
Ubiquitous Computing (UbiComp) envisions environments where devices interact among themselves to work seamlessly together on behalf of humans. In recent years, the emergence of the Internet of Things (IoT) concept, which opts for connecting everyday objects to the Internet, and the Mobile Computing paradigm have contributed to strengthening UbiComp. For this reason, UbiComp environments are not necessarily populated by powerful computers. On the contrary, resource constrained devices (e.g., embedded and mobile devices) are the main actors in these environments. Thus, it is important for the environment to deal with their heterogeneity, unreliability, and replaceability.
In order to cope with heterogeneity, the Semantic Web has proposed several standards and models to clearly define the terms so that they can be reused across applications boundaries. Regarding unreliability and replaceability, space-based computing (or Tuple Spaces) promotes the uncoupled coordination of the devices. Solutions based on semantic tuple spaces combine these three beneficial aspects resulting from bridging the Semantic Web and Tuple Spaces domains for UbiComp.
Most of these semantic tuple spaces consider embedded and mobile devices as mere clients in a space managed by more powerful devices. Such delegation helps to reduce the workload of devices with computing and energy limitations. However, this delegation moves the data away from where it is physically generated. This creates a conflict between providing updated data and generating unnecessary network traffic for unused information. In addition, this delegation makes constrained devices intrinsically dependent on other devices when it might not always be necessary. This dissertation explores how these constrained devices can act as fully fledged semantic knowledge providers to create a more decentralized space.
In conclusion, this dissertation presents a novel adaptation of semantic tuple space which considers the energy and computational impact on the devices. Specifically, this dissertation proposes the following contributions:
- A space model which considers the principles which have made the web flourish in the last decades, together with the uncoupling properties of space-based computing.
- An energy-aware search mechanism for autonomous constrained devices.
- An alignment of two approaches to act on the physical environment, namely a space-based indirect actuation and a web-based direct actuation.