RFRA: Random Forests Rate Adaptation for vehicular networks
2013 (English)In: IEEE 14th International Symposium and Workshops on a World of Wireless, Mobile and Multimedia Networks (WoWMoM), 2013, IEEE conference proceedings, 2013, 6583398- p.Conference paper (Refereed)
Rate adaptation in vehicular networks is known to be more challenging than in WLANs due to the high mobility of stations. Nevertheless, vehicular networks are subject to certain recurring patterns particularly if stations communicate to roadside units. This has lead to the proposal of learning-based rate adaptation schemes which are trained for a certain propagation environment. In general, these schemes outperform other approaches at the price of being specific for a particular environment. In this paper we present RFRA, a novel rate adaptation scheme for vehicular networks. It is based on the machine-learning algorithm Random Forests which is known to be superior to most other learning approaches. Firstly, we show that RFRA outperforms other learning-based methods significantly. We also study the question how sensitive RFRA is to changes of the learned environment, especially with respect to the propagation characteristics. We show that, although this reduces the gain of our scheme, RFRA still provides a much higher performance than state-of-the-art rate adaptation schemes.
Place, publisher, year, edition, pages
IEEE conference proceedings, 2013. 6583398- p.
Learning approach, Learning-based methods, Propagation characteristics, Propagation environment, Rate adaptation, Rate adaptation schemes, Roadside units, Vehicular networks, Computer networks, Information systems, Decision trees
IdentifiersURN: urn:nbn:se:kth:diva-133291DOI: 10.1109/WoWMoM.2013.6583398ScopusID: 2-s2.0-84883738667ISBN: 978-146735827-9OAI: oai:DiVA.org:kth-133291DiVA: diva2:660474
2013 IEEE 14th International Symposium on a World of Wireless, Mobile and Multimedia Networks, WoWMoM 2013; Madrid; Spain; 4 June 2013 through 7 June 2013
QC 201310302013-10-302013-10-292013-12-13Bibliographically approved