Sniffing out correct error frame model for the ns-2 simulator

Network protocols and their efficiency are extremely important to the research community. Evaluating the performance of these protocols has thus also becoming extremely important. However, it is difficult to conduct physical experiments on the performance of these systems, especially wireless networks. Therefore, network simulators, such as ns-2, QualNet and OPNET, have been widely used to evaluate the performance of network protocols, and thus the accuracy of network simulators is critical to the research community. In this paper, we consider the incorrect simulation results generated by the current ns-2 implementation employed in numerous research works. The cause which for the purposes of this paper will be called the incorrect error frame assumption (IEFA), is the incorrect handling of frames from stations located between the carrier sense and transmission ranges of the receiver in the two-ray ground propagation model. Although this model is known to be oversimplified to the extent that it does not consider such critical phenomena as cumulative interference and noise levels, it is the most popular model for simulating wireless networks. Moreover, the incorrect assumption has been widely adopted in the research community, such that the simulation scenarios are limited or the results of simulations are significantly affected by the IEFA. The issue is therefore in need of investigation and clarification. In this paper, we present a detailed description of the IEFA problem and a probabilistic analysis of the problem to determine its root cause. We suggest an implementation that fixes the IEFA and present the current bit-error model for the ns-2 simulator. We then conduct simulations to show how the problem impacts the simulation results from the MAC to the application layers.