Wireless sensor networks are widely used to continuously collect data from the environment. Because of energy constraints on battery-powered nodes, it is critical to minimize communication. Suppression has been proposed as a way to reduce communication by using predictive models to suppress reporting of predictable data. However, in the presence of communication failures, missing data are difficult to interpret because these could have been either suppressed or lost in transmission. There is no existing solution for handling failures for general, spatiotemporal suppression that uses cascading. While cascading further reduces communication, it makes failure handling difficult, because nodes can act on incomplete or incorrect information and in turn affect other nodes. We propose a cascaded suppression framework that exploits both temporal and spatial data correlation to reduce communication, and applies coding theory and Bayesian inference to recover missing data resulted from suppression and communication failures. Experiment results show that cascaded suppression significantly reduces communication cost and improves missing data recovery compared to existing approaches.

Big array analytics is becoming indispensable in answering important scientific and business questions. Most analysis tasks consist of multiple steps, each making one or multiple passes over the arrays to be analyzed and generating intermediate results. In the big data setting, I/O optimization is a key to efficient analytics. In this paper, we develop a framework and techniques for capturing a broad range of analysis tasks expressible in nested-loop forms, representing them in a declarative way, and optimizing their I/O by identifying sharing opportunities. Experiment results show that our optimizer is capable of finding execution plans that exploit nontrivial I/O sharing opportunities with significant savings.

Statistical analysis of massive data is becoming indispensable to science, commerce, and society today. Such analysis requires efficient, flexible storage support and special optimization techniques. In this demo, we present RIOT (R with I/O Transparency), a system that extends R, a popular computing environment for statistical data analysis. RIOT makes R programs I/O-efficient in a way transparent to users. It features a flexible array storage manager and an optimization engine suitable for statistical and numerical operations. RIOT also seamlessly integrates with external database systems, offering additional opportunities for processing data that reside in databases by blurring the boundary between database and host-language processing. This demo will show how statistical computation can be effectively and efficiently handled by RIOT.