Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Know thy neighbor - a data-driven approach to neighborhood estimation in VANETs

 
: Roscher, Karsten; Nitsche, Thomas; Knorr, Rudi

:
Postprint urn:nbn:de:0011-n-4846172 (248 KByte PDF)
MD5 Fingerprint: 6bf40ecb7f978c2f580f26bf1437bccb
© IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Erstellt am: 28.2.2018


Institute of Electrical and Electronics Engineers -IEEE-:
IEEE 86th Vehicular Technology Conference, VTC Fall 2017. Proceedings : Toronto, Canada, 24–27 September 2017
Piscataway, NJ: IEEE, 2017
ISBN: 978-1-5090-5935-5
ISBN: 978-1-5090-5934-8
ISBN: 978-1-5090-5936-2
5 S.
Vehicular Technology Conference (VTC Fall) <86, 2017, Toronto>
European Commission EC
H2020; 636220; TIMON
Enhanced real time services for an optimized multimodal mobility relying on cooperative networks and open data
Englisch
Konferenzbeitrag, Elektronische Publikation
Fraunhofer ESK
classification; feature selection; link estimation; machine learning; neighbor selection; neural network; simulation; vehicular ad hoc network; VANET; V2X; Vehicle-to-X; unicast; standard; reliability; estimation; Cams; adaptation model

Abstract
Current advances in vehicular ad-hoc networks (VANETs) point out the importance of multi-hop message dissemination. For this type of communication, the selection of neighboring nodes with stable links is vital. In this work, we address the neighbor selection problem with a data-driven approach. To this aim, we apply machine learning techniques to a massive data-set of ETSI ITS message exchange samples, obtained from simulated traffic in the highly detailed Luxembourg SUMO Traffic (LuST) Scenario. As a result, we present classification methods that increase neighbor selection accuracy by up to 43% compared to the state of the art.

: http://publica.fraunhofer.de/dokumente/N-484617.html