Vehicular Ad Hoc Networks: Interplay of Geometry, Communications, and Traffic

VEHICULAR AD HOC NETWORKS: INTERPLAY OF GEOMETRY, COMMUNICATIONS, AND TRAFFIC FEBRUARY 2013 MOHAMMAD NEKOUI B.Sc., UNIVERSITY OF TEHRAN M.Sc., UNIVERSITY OF TEHRAN Ph.D., UNIVERSITY OF MASSACHUSETTS AMHERST Directed by: Professor Hossein Pishro-Nik Vehicular Ad Hoc Networks (VANETs) have been proposed to enhance the safety and efficiency of transportation systems. Such networks hold unique characteristics and fulfill new goals that necessitate their study from a whole new perspective other than what has been the prevailing paradigm for conventional Mobile Ad Hoc Networks (MANETs). The mission of this dissertation is to identify such unique characteristics and propose design strategies for VANETs that target the new system goals. We argue that the road and obstacle geometry are two important factors that should be appropriately addressed when studying the communications throughput of VANETs. To this end we first study the effect of traffic conditions and road geometry on VANET throughput scaling laws. We use graph-theoretic and geometrical concepts to derive the throughput scaling of single roads, downtown grids, and general geometry road systems. vi Moreover, since vehicular communications are supposed to operate in the high frequency ranges, line-of-sight between communicating vehicles picks up importance in VANETs. We use computational geometry tools to study how the specific geometry of obstacles (such as buildings) affects the capacity of urban area VANETs. Finally, the design goal in MANETs is mostly to enhance the communications metrics (such as throughput and/or delay) of the network, whereas in VANETs, is mainly to improve the safety and efficiency of commute. Yet, better performance in terms of the communications metrics does not necessarily lead into improved safety and efficiency of driving. To this end, the main theme of this dissertation is dedicated to the application-oriented design of VANETs for safety applications. To this end we bring the drivers’ application needs to the forefront of our attention and provide an analytic framework for VANET safety application design during both sparse and dense vehicular traffic conditions. We use tools from stochastic geometry to derive the optimal MAC parameters that satisfy the safety requirements of the system and validate our results through NS-2 simulations. Our ultimate goal there is to fillin the current gap between purely traffic-based studies that fail to account for the non-idealities of communications, and communications-based ones which neglect the application needs of the system.