Fulfilling the Promise of Mobile Device Crowds (PerCrowd 2020 Keynote Paper)

Mobile devices today possess an incredible amount of processing power and storage. They are increasingly ubiquitous and tend to be present in clusters or crowds. These crowds can be made up of a small number of devices when, for example, one considers the devices an individual may carry on their person or may have in their home. Alternatively, the device crowds can be made up of a large number of devices as, for example, one might find in a classroom, coffee shop or university campus. Within each crowd the devices can communicate with each other as well as with entities outside the crowd. It has been observed that one can leverage these device crowds to provide meaningful compute capability for applications from within and outside the crowd. In this talk I will provide a taxonomy for device crowds that will help emphasize the diversity of contexts in which they may operate. I will then summarize our work on a number of device crowd-based systems, such as Serendipity and FemtoCloud. This work aims to understand the processing capability that can be provided with mobile device crowds in various settings. Finally I will discuss some of the challenges that remain to be addressed for device crowds to fulfil their promise.