Next year is the 20th anniversary of Mark Weiser's ubicomp vision landmark Scientific American paper on ubiquitous computing: "The Computer for the 21st Century". While there has been significant progress towards his vision most research has focused on the development of small-scale pervasive systems, tested by a handful of users, interacting with a limited number of devices. In order to advance the field and make technology truly pervasive, the research community needs to address the issue of scale. Future large-scale pervasive systems need to operate over different spatial and temporal scales, encompass a large number of diverse devices (e.g., mobile phones, tablets, wearables, embedded wireless sensors) that enable a spectrum of applications, deal with very large amounts of data distributed over a diverse set of networking platforms and devices, and support 100s of millions of users.

The trajectory from small to massive scale pervasive computing systems is underway. Recently, smart environments, body area sensor networks, and smartphones with embedded sensors are enabling the delivery of a wide variety of applications from predicting traffic jams and modeling human activities, to social interactions, behavioral and mobility patterns, to community health tracking, public safety and large-scale environmental sensing. These recent developments are being driven by the availability of embedded sensors; the ease at which researchers and developers can distribute new applications to millions of users; and the emergence of the mobile computing cloud.

The goal of this NSF sponsored workshop is to discuss the challenges for scaling our future pervasive applications, algorithms, models, data and systems. The problem of scaling pervasive systems is multi-disciplinary in nature, including challenges in HCI, machine learning, data mining, mobile systems, wireless and sensor networks, smart environments, security and privacy, signal processing, control theory, information theory, game theory, optimization techniques, psychology and social networking.

The workshop program includes four keynote talks by Gaetano Borriello (University of Washington), Andy Hopper (Cambridge University), David Culler (University of California at Berkeley) and Shwetak Patel (University of Washington); NSF perspective by Keith Marzullo (Division Director of CNS Division, NSF); a number of panels and breakout sessions related to scaling ubicomp; and a plenary discussion on 20 years after Mark Weiser's ubicomp vision moderated by Mahadev Satyanarayanan (Carnegie Mellon University) and Roy Want (Intel).

The white papers by attendees, slides from presentations and video of plenary sessions are now online. The final NSF report on Pervasive Computing at Scale will be available later in the year.


Final NSF report on the workshop's findings, October 2012