In recent years, in addition to the significant growth in intelligent transportation systems with new concepts such as SDN (Software Defined Networks), fog computing, and new machine learning methods, the advancement of communication technologies and automotive industries has been considerable. Vehicular Adhoc NETworks (VANETs) can effectively detect traffic congestion, which can be classified into Repetitive Congestion (RC) and Non-Repetitive Congestion (NRC). In particular, NRC in an urban area is mainly due to accidents, congested areas with high loads, special events, and weather conditions. This paper presents a new architecture using two concepts, SDN and fog computing, called SDN and Fog computing-based VANET (SFVN), to classify the causes of congestion in VANET. This architecture is proposed for distributed and real-time congestion classification, considering its components in an urban road network in VANET. The proposed scheme has been evaluated with CCSANN, FogJam, C4.5 classification tree, and the Na¨ıve Bayes method, which indicates the effectiveness of the proposed method. By knowing the leading causes of congestion, this scheme can help transport companies develop effective policies to reduce urban congestion.

Keywords