DBCC: Leveraging Link Perception for Distributed Beacon Congestion Control in VANETs
Feng Lyu,Nan Cheng,Haibo Zhou,Wenchao Xu,Weisen Shi,Jiayin Chen,Minglu Li
DOI: https://doi.org/10.1109/jiot.2018.2844826
IF: 10.6
2018-01-01
IEEE Internet of Things Journal
Abstract:Under the IEEE 802.11p-based dedicated short range communication modules, vehicular safety applications rely on periodical broadcasts of safety beacons by each vehicle. However, the channel can be easily congested by high-frequency periodic beacons when the vehicle density becomes heavy. In this paper, through real-trace-based empirical study on vehicle-to-vehicle communication, we find that nonline-of-sight (NLoS) condition is the key factor on link performance degradation and blindly sending more packets in harsh NLoS conditions can hardly succeed but increase interferences to neighboring vehicles. Inspired by this, we propose a distributed beacon congestion control (DBCC) scheme to control beacon activities with considering link conditions, i.e., vehicles with more neighbors and better conditions of links with its neighbors, will be assigned with higher beacon rates. In DBCC, we first utilize two machine learning methods, i.e., naive Bayes and support vector machines, to train the features and output a classifier model which conducts online NLoS link condition prediction. With link status information, we then formulate a link-weighted safety benefit maximization (L-SBM) problem of the rate-adaptation under a TDMA broadcast MAC, which is proved to be NP-hard. A greedy heuristic algorithm for L-SBM is then proposed and the performance of the algorithm is evaluated. Extensive trace-driven simulations demonstrate the efficiency of DBCC design; particularly, the rate of beacon transmissions can be effectively controlled without exceeding the resource limit and the rate of transmission/reception collisions are greatly reduced.