Dynamic trafile congestion detection in VANETS using a Fuzzy rule-based system and K-means clustering

Vehicular traffic congestion poses a serious challenge to a green environment by contributing to air pollution, noise pollution, and unnecessary fuel consumption. This paper aims to meet one of the requirements of a smart city that detects traffic congestion in an area so that the traffic can be diverted on an alternate route in a vehicular adhoc network (VANET). VANET is a medium of wireless communication among vehicular commuters, which enables them to intimate their instantaneous physical characteristics such as speed, brake frequency, rain, fog, acceleration, and position to surrounding vehicles within a periphery so as to determine the level of congestion and find suitable ways to divert the traffic. The dynamic nature of the vehicular nodes makes the topology unpredictable, which is constantly monitored using a vehicular adhoc network (VANET). In this work, an integration of fuzzy inference rule based system (FRBS) and K-means clustering technique is explored to detect the traffic congestion under a dynamic traffic environment. FRBS is implemented through an Arduino Uno microcontroller. In this paper, four different physical vehicle attributes such as rain or fog, speed and brake frequency are considered for traffic congestion detection. This paper presents a detailed description regarding the co-ordination between vehicular units and a web server, which maintains a cloud database that preserves the data for future use.