Curracurrong.

Describe the computation.
Distribute the work.

Curracurrong represents a query as a graph of stream operators linked by communication channels. It connects a high-level programming model with the practical tasks of placing and executing operations across a network.

The original Java framework combines a query-processing server with a sensor-node runtime for administration, communication, and scheduling. Its implementation used SunSPOT sensor nodes, an extensible operator library, data-rate annotations, and query visualisation.

A simplified illustration of a stream graph. The operator arrangement is conceptual, rather than a measured experiment or a reproduced paper figure.

Language, placement, and timing.

• Sensor networks and operator placement — express applications as stream graphs and map operators to devices with energy consumption in mind.
• Cloud execution — apply continuous processing and automated operator placement within cloud-hosted clusters.
• Query timing — calculate end-to-end delays using dependencies between data tokens and execution schedules.

The associated papers were coauthored with Bernhard Scholz, Alan Fekete, Andrew E. Santosa, Saeed Attar, and Akon Dey across the different studies. Each publication lists its full author group.

The original project page acknowledges ARC Discovery Project DP1096445, “Computing with nearly-consistent data”, naming Alan Fekete and Bernhard Scholz.