Real-Time Traffic Data Smoothing from GPS Sparse Measures Using Fuzzy Switching Linear Models

Traffic is one of the urban phenomena that have been attracting substantial interest in different scientific and industrial communities since many decades. Indeed, traffic congestions can have severe negative effects on people's safety, daily activities and quality of life, resulting into economical, environmental and health burden for both governments and organizations. Traffic monitoring has become a hot multi-disciplinary research topic that aims to minimize traffic's negative effects by developing intelligent techniques for accurate traffic states’ estimation, control and prediction. In this paper, we propose a novel algorithm for traffic state estimation from GPS data and using fuzzy switching linear models. The use of fuzzy switches allows the representation of intermediate traffic states, which provides more accurate traffic estimation compared to the traditional hard switching models, and consequently enables making better proactive and in-time decisions. The proposed algorithm has been tested on open traffic datasets collected in England, 2014. The results of the experiments are promising, with a maximum absolute relative error equal to 9.04%. © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the Conference Program Chairs.