Section: New Results

Simulation of wireless sensor networks

In wireless multi-hop networks (i.e., ad hoc, sensor and mesh networks), there is a growing need for the performance evaluation of protocols and distributed applications. Three main methodologies are generally adopted to investigate the performance and the behavior of networking protocols: theoretical analysis, experimentation and simulation. Due to the high complexity of wireless communications, analytical studies are often based on unrealistic assumptions and inaccurate physical layer, e.g., the so-called Unit Disk Graph model. The experimentation approach may provide valuable insight into the performance and the behavior of wireless systems, however, setting-up testbeds is a tedious and expensive task. At the end, simulations are generally considered as the most convenient methodology to explore the behavior of protocols and distributed applications. Nonetheless, the complexity of the physical phenomena constituting the radio medium (PHY) introduces a tradeoff between accuracy and computational cost in wireless network simulation. In [6] , we have focused on the physical layer modeling issue and we have investigated its impact on the accuracy and the complexity of wireless network simulations. The question we have raise is what is the real cost of PHY simulation accuracy ? We have deliberately kept aside optimizations and scalability of the node and protocol aspects of simulations which have been the subject of other studies. We have used a flexible and modular simulation framework, WSNet, to evaluate the PHY tradeoff and better understand the impact of the PHY layer on the obtained simulation results.