|Abstract:||The broadcast nature of the wireless medium and its susceptibility to interference not only affects the capacity of wireless networks but also has an impact on the fidelity, accuracy and efficiency of wireless network simulation. This is because transmission of a wireless signal has to be received and processed by a potentially large number of nodes that operate on the same channel. How to faithfully and yet effectively carry out simulation has thus been an important issue to realizing real time, large scale wireless network simulation.
In this paper, we take a systematic approach and investigate the process of signal transmission in the various stages: signal propagation, signal interference, and interaction with the PHY/MAC layers. We identify the key components that affect the simulation performance and quantify their impacts whenever possible. Specifically, we first investigate the impact of the signal propagation limit on the accuracy and efficiency of simulation. Second, we devise a simple computation model to capture the effect of the grid size --- the parameter that characterizes how nodes that are subject to interference are being located --- on simulation efficiency in a concave manner and derive its optimal value. Third, we make a key observation on whether or not, and when, a node needs to respond to a signal arrival event from the perspective of the MAC layer and propose a reactive channel model (RCM), which can effectively reduce the large number of events related to signal transmissions by an order of magnitude.
The simulation results have demonstrated that RCM, with the appropriate choices of the signal propagation limit and grid size, gives an order of magnitude improvement in the execution time. This, coupled with the fact that RCM can be implemented in a modular manner, requires only modest changes to existing PHY/MAC layers in the simulator, and does not incur extra memory consumption, suggests that RCM is a promising approach to enabling large-scale wireless network simulation.