Automatic Generation of WSN Simulations: From Callas Applications to VisualSense Models

Wireless sensor networks are hard to program, deploy, and test. This paper addresses a key problem of testing such networks using a simulation: model definition. Our approach is to generate simulation models directly from WSN applications, in particular, VisualSense simulator models from applications written in Callas, a programming language for WSN. For that purpose, we create a model generator that is parameterizable by network and sensor templates, and by an extensible set of simulation parameters. Our experiments show encouraging results on simulating large scale networks, as we are able to handle WSN with as many as 5000 nodes.