Building Capable, Energy-Efficient, Flexible Visualization and Sensing Clusters from Commodity Tablets
Conference website: datasys.cs.iit.edu/events/GCASR13
We explore the application of clusters of commodity tablet devices to problems spanning a “trilogy” of concerns: visualization, sensing, and computation. We conjecture that such clusters provide a low-cost, energy-efficient, flexible, and ultimately effective platform to tackle a wide range of problems within this trilogy. This is a work in progress, and we now elaborate our position and give a preliminary status report.
A wide range of Android tablet devices are available in terms of price and capabilities. “You get what you pay for” w.r.t. display resolution, sensors, and chipset---corresponding to the trilogy. $200 gets one a 1280x800-pixel touch display, quad-core CPU with GPU, and various input sensors (camera, accelerometer, barometer, etc.) When arranged in a suitable geometry, such as a rectangle, such devices form an innovative whole that is more powerful than the sum of its parts, with CPUs and sensors in addition to a much higher combined display resolution than that of conventional HD monitors.
Scholarly merit: The tablet cluster serves as a testbed for exploring a wide range of research questions in both technical and application domains. Numerous applications in education and research are imaginable, such as environmental and security monitoring, calculating and visualizing the energy footprint of an organization, facial recognition involving multiple cameras, exploring molecules in three dimensions, visualizing relationships among versions of a text, teaching the color space to art students, mapping audiovisual content from sequential to concurrent playback, etc.
Project status: We have made preliminary progress along multiple fronts. Specifically, we have developed an Android/Java prototype app for clustering multiple devices and drawing to the resulting combined virtual display on top of the Skeenzone middleware for distributed mobile applications (http://code.google.com/p/skeenzone). We have also developed a an Android/Scala prototype app for navigating RESTful web services as data sources.