No abstract

Abstract-The formidable growth of WSN research has opened challenging issues about their performance evaluation. Despite the steady increase in mathematical analysis and experimental deployments, most of the community has chosen simulation for their study. Although it seems straightforward, this approach becomes a quite delicate matter. Complexity is caused by several issues. First, the large number of nodes heavily impacts simulation performance and scalability. Second, credible results demand an accurate characterization of the sensor radio channel. New aspects, inherent in WSN, must be included in simulators, e.g. a physical environment and an energy model, leading to different degrees of accuracy versus performance. Moreover, many necessary models are in the continuous-time domain (e.g. heat transmission, battery discharge), being complex to integrate into discrete event network simulators. These issues result in an exponential growth of the overall network state information. Through this survey we review these problems both quantitatively and qualitatively while depicting a common suitable simulation model. We also briefly describe the most significant simulation frameworks available.

No abstract

No abstract

Abstract:Simulations are currently an essential tool to develop and test wireless sensor networks (WSNs) protocols and to analyze future WSNs applications performance. Researchers often simulate their proposals rather than deploying high-cost test-beds or develop complex mathematical analysis. However, simulation results rely on physical layer assumptions, which are not usually accurate enough to capture the real behavior of a WSN. Such an issue can lead to mistaken or questionable results. Besides, most of the envisioned applications for WSNs consider the nodes to be at the ground level. However, there is a lack of radio propagation characterization and validation by measurements with nodes at ground level for actual sensor hardware. In this paper, we propose to use a low-computational cost, two slope, log-normal pathloss near ground outdoor channel model at 868 MHz in WSN simulations. The model is validated by extensive real hardware measurements obtained in different scenarios. In addition, accurate model parameters are provided. This model is compared with the well-known one slope path-loss model. We demonstrate that the two slope log-normal model provides more accurate WSN simulations at almost the same computational cost as the single slope one. It is also shown that the radio propagation characterization heavily depends on the adjusted model parameters for a target deployment scenario: The model parameters have a considerable impact on the average number of neighbors and on the network connectivity.