A Support Vector Machine Implementation for Traffic Assignment Problem

Abstract

Simulating urban mobility scenarios is a useful tool for researchers in multiple fields like Urban Planning, Traffic Optimization, CO$^2$ Emissions Analysis, Performance Evaluation of Protocols for Connected Vehicles, among others. SUMO handles microscopic traffic simulations and allows communication to Python language through an API which is also shared by VEINS. This communication channel lets researchers interact with the simulation on-live, facilitating the implementation of state-of-the-art algorithms from Machine Learning (ML) and Artificial Intelligence (AI). On the other hand, OMNeT++ is a framework to manage and analyze communication protocols of mobile networks. We experimentally evaluated the training of a Support Vector Machine (SVM) in the SUMO-VEINS-OMNeT++ framework. Our experiments show the best classification model for a particular traffic light assignment scenario.