Framework for high level programming of wireless sensor and actuator networks.

The vision of ubiquitous computing (UbiComp), as introduced by Mark Weiser, goes beyond simple embedding of computational capabilities in the environment. The vision foresees the way humans interact with technology in a very different way; the future of man-machine interaction will not be through clicking and typing on computers but through interacting with objects around us through speech, touch, gestures or movement, etc.; any required computation will be performed in the background, hidden from the user. Since Weiser first introduced his idea, technology has been progressing making essential bricks of such an environment available and successful research projects have been developed; however, the UbiComp vision is still far from current reality. This is partly due to the fact that technology is still too visible, and partly due to the complexity of the systems themselves but even more so due to the complexity of setting up and handling both systems and services. Most users of UbiComp applications can be expected to be technology agnostic, they don’t have technical background and their knowledge of the programming languages that are typically used to set up UbiComp applications is either very limited or does not exist. This has so far limited building smart environment applications to professional developers. At the same time, the actual users of UbiComp have typically rather individual requirements depending on their intended uses and environment; they tend to define tasks in their own way to take control over their environment. Thus there is a need to move the customisation functionality of applications to the users’ side of the process. This entails simplifying the way service are defined and customised by making discovery, operational modification and provisioning transparent to the user. This thesis presents a top down approach providing a high level UbiComp programing language that aims at empowering end-users with the ability to control their environment without needing understanding about the underlying technologies. The work defined and investigated a set of algorithms for mapping users’ service requirements to available assets in the network and proved that they outperform SOTA in resource allocation. The here defined programming language builds on this mapping and hides away the complexity of the system from its users. The usability of the language was verified using “cognitive dimensions”, formal semantics was used for verifying the language and connecting it to the mapping algorithms