| In this work, we propose techniques to systematize the modeling of wireless networks. In spite of the tremendous progress in wireless network research, there still exist a few major issues in wireless network modeling that remain relatively overlooked. These issues affect the results of simulation and analytical modeling of wireless networks and have a considerable impact on research studies. We consider three such issues in depth, detail several problems caused by them, and propose techniques to address these problems.; The large number of parameters. Wireless network models typically have a large number of parameters because of the intricate nature of wireless networks which introduces a large degree of freedom in choosing parameters and their values. This abundance of parameters makes it difficult to interpret, compare, and replicate research results. We argue that we need fewer, more meaningful parameters that capture the essence of the network while hiding the specifics. Moreover, absolute parameters and values are usually not useful in analyzing results. We show how to reduce the parameters into a smaller set and also propose relative parameters which do not deal with absolute values. Specifically, we introduce a novel concept, steady state utilization, to capture the inherent network capacity and use it to simplify network analysis.; The assumption of homogeneity. Wireless network models frequently utilize a large degree of homogeneity to simplify their implementation and the subsequent analysis of results. We challenge this assumption and try to answer the question: How does heterogeneity affect network behavior? Using simulations and analytical models, we demonstrate the effects of heterogeneity and show that it is important to consider them in order to avoid inaccuracy and misinterpretation of results. We also consider hotspots as a special case of heterogeneity. We show that they can occur due to various reasons and based on the causes, their, effects on network behavior are different.; The predictability of mobility. Predicting the mobility of users is a critical issue in wireless networks, especially in providing quality of service (QoS). We study predictability of user mobility in the context of making advance reservations to improve QoS. Some mechanisms of making reservations depend on the usage of statistical data to predict user movement. We show that the QoS depends on the accuracy of this data. To do this, we utilize relative parameters which we first introduce in addressing the issue of a vast parameter space. |
