For exploring large, complex data sets, nothing matches the power of interactive visualizations that let users directly manipulate data and arrange it in new ways.
But with data sets of millions or billions of data points, interactivity is hard to achieve. It’s hindered by the sheer number of computations required to return a query result, and by the latency inherent in accessing data so big it requires disk storage. AT&T researchers realized the best way to maintain interactivity was to move all data into main memory. How to do that was the challenge, and it required a new type of data structure: the nanocube. Read more.
Nanocubes for Real-Time Exploration of Spatiotemporal Datasets, a paper by Lauro Lins, James T. Klosowski, and Carlos Scheidegger, was awarded an Honorable Mention at this year’s IEEE INFOVIS Conference. The paper presents algorithms to compute and query a nanocube, a new type of data structure that aggregates spatiotemporal data and makes it possible to interactively visualize billion-point data sets. Nanocubes are the subject of the main article.
Taniya Mishra never touched a computer prior to college, but she quickly realized how much a computer can do. Now an AT&T Researcher, she uses her computer skills to build synthetic voices that embody different characters and emotions. Her aim is to both improve communication and reduce accessibility barriers for all people. In this blog post for the Huffington Post’s Girls in STEM series, she gives tips for how to get and stay interested in STEM.
One of the tricks to ensuring small cell deployment is worth the effort and expense involved is the use of new network planning models specifically designed for the new world of heterogenous networks (HetNets). With that in mind, engineers at AT&T Labs set out to identify radio-frequency propagation models that will help pinpoint optimal small cell placements.
The result of this research is AT&T's proprietary HetNet Analysis and Resource Planning (HARP) tool
Joint-Family: Adaptive BitRate Video-on-Demand Streaming over Peer-to-Peer Networks with Realistic Abandonment Patterns.
Kyung Hwang, Vijay Gopalakrishnan, Rittwik Jana, Seungjoon Lee, Vishal Misra, K. K. Ramakrishnan, Dan Rubenstein
In The News