The increasing penetration of the real world with embedded and globally networked sensors leads to the formation of the Internet of Things, offering global online access to the current state of the real world.
We argue that on top of this realtime data, a Web of Things is needed, a software infrastructure that allows the construction of applications involving sensor equipped real-world entities living in the Internet of Things. A key service for such an infrastructure is a search engine that supports lookup of real-world entities that exhibit a certain current state as perceived by sensors.
In contrast to existing Web search engines, such a real-world search engine has to support searching for rapidly changing state information generated by sensors. In this paper, we show how the existing Web infrastructure can be leveraged to support publishing of sensor and entity data.
Based on this we present a real-time search engine for the Web of Things. Essentially, our search engine called Dyser supports the search for real-world entities with a user-specified current state.
For example, Dyser could be used to search for rooms in a large building which are currently occupied, for bicycle rental stations which have currently bikes available, for currently quiet places at the waterfront, or for current traffic jams in a city. We studied the performance of Dyser on a real-world data set containing data from 385 sensors over a period of 5 months.
An important thread for future work is the distribution – where several instances of the search engine take care of certain subsets of entity and sensor pages – and parallelization of the search engine – where the search engine itself and its index are distributed over multiple computers.
The former is expected to offer a significant speedup for the execution of multiple simultaneous queries, while the latter can be expected to also speed up the execution of a single query.
To read this external content in full, dowload the paper from the author archives at the ETH's Instutte for Pervasive Computing.