ML-Based Localizing and Driving Direction Estimation System for Vehicular Networks

Abstract

Road traffic injuries continuously claim the lives of millions of people besides millions of disabled people every year and causing tremendous economic loss. Numerous safety-critical applications, for instance, pre-crash sensing, blind spot warning, emergency braking, lane changing, cooperative collision avoiding, curve speed warning, etc, are proposed in Vehicular Ad Hoc Networks (VANETs) to mitigate road traffic injuries. Both, this vehicle’s position and the position of other road participants in the vicinity is a critical condition for the success of safety-critical applications. Many technologies and techniques are proposed, but no one technology is yet able to satisfy the accuracy, integrity, continuity, and availability needed by safety-critical application. Hence, integrating effective positioning or localizing techniques into one system is the solution. This work proposes a Localizing and Driving Direction Estimation System (LDDES) that estimates the position and driving direction of other vehicles in the vicinity. LDDES created using Machine Learning (ML), VANETs, and Multiple Input Multiple Output (MIMO) capabilities. The performance evaluation of LDDES shows that LDDES estimates the location and driving direction of other vehicles with an accuracy percentage of 90 %. LDDES can be integrated with the Global Position System (GPS) and other positioning techniques to enhance the vehicular positioning requirements needed by many safety-critical applications.