IRIS: Internet-scale Resource-Intensive Sensor Services

The proliferation and affordability of smart sensors such as webcams, microphones etc., has created opportunities for exciting new classes of distributed services. A key stumbling block to mining these rich information sources is the lack of a common, scalable networked infrastructure for collecting, filtering, and combining the video feeds, extracting the useful information, and enabling distributed queries. In this demo, we demonstrate the design and an early prototype of such an infrastructure, called IRIS (Internetscale Resource-Intensive Sensor services). IRIS is a potentially global network of smart sensor nodes, with webcams or other sensors, and organizing nodes that provide the means to query recent and historical sensor-based data. IRIS exploits the fact that high-volume sensor feeds are typically attached to devices with significant computing power and storage, and running a standard operating system. Aggressive filtering, smart query routing, and semantic caching are used to dramatically reduce network bandwidth utilization and improve query response times, as we will demonstrate. The service that we demonstrate here is that of a parking space finder. This service utilizes webcams that monitor parking spaces to answer queries such as the availability of parking spaces near a user’s destination.