Characteristics of a vehicular network

If the full range of applications and protocols that exploit vehicular networks are going to be adopted then it is key to understand the performance characteristics of these networks: both the spatial and temporal properties. Many of the applications, services and protocols that could drive deployment of vehicular networking can only be developed if we understand the properties of the network that will be created. Yet, to really understand these properties we need studies of large-scale vehicle traffic traces, which are difficult to obtain. To date, most studies have used traces of small street or freeway sections, or the traces have been generated using unrealistic mobility models. To address this issue we present an extensive analysis of a mobility trace generated by Los Alamos National Laboratory for Portland, Oregon. It is generated using the TRANSIMS micro-traffic simulator, combined with macrolevel traffic information gathered in the city from censuses and studies over a year. Hence, this trace is far more accurate than previously studied synthetic traces. We analyze it to identify and characterize the key properties starting from the characteristics of the vehicle movement. We then explore the properties of the network that would be formed, considering different radio ranges and penetration rates. Finally, we then evaluate how these impact the design space for algorithms that use either one-to-many or one-to-one communication patterns.