Fast simulation methods to predict wireless sensor network performance
Miloš Blagojević, Majid Nabi, Teun Hendriks, Twan Basten, Marc C. W. Geilen · 2009
With the increasing capabilities of Wireless Sensor Networks (WSN), complexity and expectation of the WSN applications increase as well. In order to make design-space exploration possible, it is necessary to have fast models that provide adequate insight in system behavior. In this paper, we propose a highly abstracted, hierarchical, system-level modeling method for WSN. Based on the model properties, fast simulation techniques can be applied. First, an abstract discrete event simulation based on a Probabilistic Graph Model (PGM) is introduced. Then, a fast Monte Carlo simulation approach is proposed for speeding up the simulation process. This approach combines Stochastic-Variable Graph Models (SVGM), providing a high level of abstraction, with shortest path calculations. As a case study, a temperature mapping application in a gossip-based WSN is used, showing a good accuracy of the model predictions.