UC Davis

Cyber-physical systems continue to be increasingly important to global society because of automation, miniaturization of electronics, and the Internet of Things. More smartphones and tablets are being manufactured and discarded or recycled every year. This research presents a design and characterization of a system that gives retired mobile devices further use within cyber-physical systems beyond their typical end-of-life. The contributions include an experimental characterization of wireless communication protocols for embedded systems involving wide-area networks using query vehicles, a design of an app-based distributed computing architecture named Cluster made from retired smartphones and tablets, implementation of four distributed computing tasks, and a characterization of the performance, energy consumption, and network utilization of Cluster and each of the four compute tasks. This research analyzes the potential and performance of retired, but still functional, mobile devices to be used as nodes in a distributed computer for applications such as image pre-processing, dependent parallel calculations, server-heavy coordinated computing efforts, and a distributed data store based on a key-value data store implemented on a "Fast Array of Wimpy Nodes". The results of this research demonstrate that Cluster is a viable distributed computer with some expected reliability concerns from reusing retired computing resources, comparable compute performance, lower power, greater physical space efficiency, and lower monetary costs than most other alternative distributed computing architectures.