The Roaming Edge and its Applications

Connected sensors, e.g., cameras, LiDARs, air sensors, installed in a mobile platform (e.g., a terrestrial or aerial vehicle, or special in-person devices) can provide broad views of wide-area environments quickly and efficiently. If many vehicles incorporate such sensing systems, they together can be composed into a unique crowd-sourced platform and can gather fine-grained, diverse, and noisy information at city-scales. However, these sensors can generate large amounts of data and such data is hard to aggregate in a central server. We explore the design of a "roaming edge" - the notion that generalpurpose computing be installed in these mobile platforms, which connect over wireless paths to the static infrastructure and to the static edge nodes, to support a broad range of applications. In particular, a roaming edge node allows different sensors and data sources in-range of a mobile platform to connect to it, and supports data processing for necessary local analytics, considering both efficiency and privacy. The roaming edge, of course, does not operate in isolation and we describe a three-tier architecture that integrates it with a static edge and cloudhosted services. This paper also outlines several applications that can leverage opportunities provided by the roaming edge, and focus, briefly, on one - a real-time video query application with public safety implications.