Abstract

Large-scale sensor networks are usually widely distributed, often in inaccessible places. Making changes to such deployments in situ is undesirable (or impossible if the nodes are installed in an area hazardous to humans), thus requiring that the code installed in the nodes be rigorously tested prior to deployment. The most common method for such testing is through simulation.

Discrete event simulators, such as TOSSIM , provide “high fidelity” simulation of large sensor networks, primarily through high fidelity radio simulations at the bit level and hardware emulation at a high level of abstraction. Most discrete event simulators, by their very nature, mask race conditions, since simulated interrupts will NEVER interrupt running code. Additionally, such simulators typically require that all simulated nodes run the same application code. Since both of these problems reduce confidence in the deployed system, the focus of this work is to eliminate these problems via complete emulation of sensor networks using virtualization techniques.

This talk describes our experience in producing a version of TinyOS, XenoTiny, which can be executed as a guest domain over the Xen virtualization kernel. XenoTiny is well integrated with the TinyOS build process. Since each node runs independently in its own guest domain, race conditions are able to manifest themselves, and each node can run a node-appropriate application. The hardware emulation is performed at the lowest possible hardware abstraction layer, thus maximizing the amount of actual TinyOS code that is executed during emulation. Finally, a novel, Xen-specific radio model mechanism has been introduced, easing the introduction of different radio models for use during emulation runs.