CSR-CSI: Collaborative Research: Dynamic Sensor/Computation Network for Wildfire Management

  • Douglas, Craig (PI)

Grants and Contracts Details

Description

This project will enhance research in cross systems integration (CSI) of a dynamic, intelligent sensor network and a critical high performance computing application: wildland fire modeling. Many types of dynamic sensors are used in this project, from stationary sensors on the ground to sensors in satellites or attached to airplanes flying overhead. The use of the sensors changes with little warning and it cannot be predicted at compile time. This causes changes in the models used. A dynamic execution environment is essential for acquiring the data, running a multi-model application, and pushing the results to the right places at the right times securely and robustly while still being simple enough for non-computer experts to operate the complex system while on a mountainside during stressful conditions. Data are proccsscd and assigned error probability distributions. The data is moving across large geographic arears in a seamless manner from the application's viewpoint. The investigators have integrated end to end runtime tools for data movement and local storage. The important decision when data are stored and in what format are handled dynamically. Notifying running computations of new data and delivering it in the correct format for the application is critical. Data processing is done using small Java applets that run on any platform that supports it, including some smartphones, PDAs, and all traditional computers. The sensors used are cyberphysical systems (CPS) and will be repprogrammed in the field often without prior knowledge. The proposed sensors' design replaces ones whose basic designs last changed in the 1960's. They will be deployed simultaneously in large numbers and be controlled by one or more computers with widespread, secure data sharing. Recompiling at execution time isessential and can take advantage of the decreasing cost of memory and storage (e.g.: flash memory instead of hard drives) in new sensor designs to save large amounts of internal information that a compiler produces but does not normally pass on to the run tiIne system. Extending battery life is crucial and radio broadcasting of complete codes drains batteries quickly. Thus, dynamically recompiling sensor codes and only broadcasting new code fragments to the sensors saves radio time and battery life. New mathematical methods will be developed to change running simulation of a highly nonlinear system with coherent features in response to data and can be extended to control sensors as well. The project will employ virtual Grid technology in the computational environment. Building on existing large research projects will allow the investigators to quickly take advantage of abilities to find computational resources and shorter queues in a highly portable manner.
StatusFinished
Effective start/end date9/1/078/31/09

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.