Abstract

Wireless connectivity for vehicles is a fast-growing market, with a plethora of different network technologies already in use. Surveys of the numbers of IEEE 802 .11b/g access points in cities point to hundreds to thousands of networks within each square kilometre, with coverage areas that are not easily predicted due to the complexities of the urban environment. In order to take advantage of the diversity in wireless networks available, we need data concerning their coverage. Methods of generating such coverage maps that are accurate, space-efficient and easy to query are not a well addressed area. This talk will overview novel algorithms for the creation of such coverage maps, and how these have been evaluated using a large corpus of real-world data. The conversion of coverage maps into graph form to enable optimal network selection over a geographical route, whilst taking into account handover costs, will also be described. This is (to the author’s knowledge!) the first approach of this kind, particularly since it has been evaluated with real traces from a vehicle.