Correct, efficient, and realistic wireless network simulations

Simulating wireless networks accurately is a non-trivial task because of the large parameter space that affects the performance of such networks. Increasing the amount of detail in the simulation model increases these requirements by many times. Hence there is a need to develop suitable abstractions that maintain the accuracy of the simulation while keeping the computational resource requirements low. The topic of wireless network simulation models is explored in this research, concentrating on the medium access control and the physical layers.; In this dissertation, we explore the correctness, efficiency, and realism of wireless network simulations. We concentrate on the 802.11-based wireless network simulations, although the methods and results can also be used for various other wireless network simulations too. While many simulators model the IEEE 802.11 wireless networks, almost all of them tend to make some abstractions to lessen the computation burden and to obtain reasonable results.; A comparative study of three wireless simulators is made with respect to the correctness of their ideal behavior as well as their behavior under a high degree of load.; Further, the physical-layer abstraction in wireless network simulations tends to be very simplistic because of the huge computational requirements that are needed to accurately model the propagation, fading, and shadowing models. When mobility is taken into account several other issues like the Doppler Effect should also be accounted for.; This dissertation explores an empirical way to model the physical layer which cumulatively accounts for all these effects. From a network protocol designer's perspective, it is the cumulative effect of all these parameters that is of interest.; Our major contribution has been the investigation of novel empirical models of the wireless physical layer, which account for node mobility and other effects in an outdoor environment. These models are relatively more realistic and efficient when implemented in a simulation environment. Our simulation experiments validate the models and provide simulation results which closely match our outdoor experiments. Another significant contribution is in understanding and design of wireless network simulation models.