Composite fault diagnosis methodology for urban vehicular ad hoc network

Abstract

Vehicular Ad-hoc NETwork (VANET) is a meteoric growing research area due to the technological advancement in sensing, computation, communication, and radio wireless technology. The demand for VANET is increasing day by day due to the numerous diverse applications. Applications of VANET include safety applications like reduction in the road accident by broadcasting messages to the driver, convenience applications like automatic parking service, commercial applications like selling and buying of products through a network of wheels, productive applications like automatic environmental parameters monitoring, vehicular cloud service like secure toll transaction, etc. The aforementioned VANET applications will become non-operational if the communication unit of the vehicle becomes inoperative or malfunctioning. This paper proposes a composite fault diagnosis methodology by detecting and classifying the fault of the On Board Unit (OBU) of the vehicular ad hoc network. The fault detection includes the detection of two categories of the fault i.e. hard fault in which the communication unit is in the inactive state and not able to communicate with other communication units and soft fault in which the communication unit is in the active state but communicates incorrect data with other communication devices. The hard fault is detected by the repeated time out mechanism and by a statistical method called Kolmogorov–Smirnov test (K-S test). The detection of soft fault (permanent, intermittent, and transient) is performed by a statistical method called the Chi-square test. Apart from the detection of the soft fault, the classification of soft fault is carried out by ECOC-SVM. Finally, the result of the fault is transmitted to the faulty vehicle. The parameters such as Fault Detection Accuracy (FDA) and False Alarm Rate (FAR) are used to validate the proposed fault detection protocol. The parameter False classification rate (FCR) is used to measure the performance of the soft fault classification. Finally, the performance of the transmission of the result to the faulty vehicle is evaluated through the parameters end-to-end (E2E) delay, number of hops, and network gaps by comparing the proposed work with existing routing protocols of VANET like GSR, A-STAR, Bhoi et al.