Effects of Correlated Shadowing Modeling on Performance Evaluation of Wireless Sensor Networks

Abstract

Wireless links are typically modeled in isolation as independent and parallel links. However, in reality, there is often a correlation in defects and impairments between collocated links. The spatial correlation of shadowing among proximate wireless links has been frequently observed. Although different types of correlated shadowing models have been proposed for wireless channels, these models are not present in popular network simulators. Ignoring such correlations results in the diversity of adjacent links being over- estimated. Our hypothesis is that this could lead to incorrect results when evaluating the performance of wireless systems and network protocols. In this paper, the ns-3 simulation environment has been enhanced with a prototype correlated shadowing model developed in our previously published work. We study the end-to-end packet delivery ratio and delay in wireless sensor networks using both correlated and independent propagation models. Simulation results show that correlated and independent shadowing models produce significant differences in network performance. Moreover, we compare the complexity of different modeling options. It is shown that this correlated propagation model in ns-3 can improve simulation realism without inducing any considerable degradation of scalability.