Knowledge-based dynamic channel selection in vehicular networks (Poster)

In this paper, we propose a channel selection algorithm for vehicular dynamic spectrum access (VDSA) networks employing instance-based learning methods. Due to the high mobility and spatially variant spectrum availability across large geographical regions associated with this transmission environment, we propose using VDSA methods for non-safety applications such as traffic efficiency and local information dissemination. Additionally, we propose a distance-based multidimensional indexing approach to enable learning of a vehicle communications environment. Our results suggest that the multidimensional approach can improve the channel selection and channel switching performance, especially in either unknown environments or when limited learning information is available due to circumstances such as reduced storage requirements.