Abstract
Simulation results can only approximate testbed ones but, at the best of our knowledge, only very few papers such as [11, 8, 12, 15] studied the reliability of simulators. In this work, we used the results of some experiments on Mote lab [9] testbed as a baseline to evaluate the accuracy of NS-2 [3] and Castalia [10] simulators. Our results show that the use of simulators in default configurations may produce unreliable results in the considered scenario, but a simple tuning can significantly increase accuracy.