Decentralized environmental simulation and feedback in robust geosensor networks

Environmental simulation and modeling has become a basic tool for scientific exploration of complex natural systems. Geosensor networks are an emerging technology for spatiotemporal monitoring of natural systems. This paper investigates the potential for embedding environmental simulations in geosensor networks, with the aim of increasing system robustness to both sensor and model errors. The approach taken is to construct a formal, process-oriented model of a decentralized geosensor network capable of executing a spatially distributed environmental model. This formal model is used as the basis for implementing a computer simulation of a geosensor network tasked with monitoring a simple hydrological system. Three experiments are then conducted to investigate the ability of this system to tolerate different types of sensor and model errors. The results indicate that embedded environmental models can indeed contribute to improved system robustness, by detecting, distinguishing between, and even correcting for certain types of error.