Using Time Series Forecasting for Adaptive Traffic Signal Control

This paper presents a method of adaptive traffic signal control using time series forecasting and real time signal phase adjustment. The situation in which a green light turns red before passing an intersection is a familiar and frustrating experience to drivers. The proposed forecast based traffic signal adjustment attempts to predict and alleviate this situation by extending green lights in real time. The procedure and thought process of the implementation of the system are discussed in this paper along with results from a simulation on three intersections over the course of a week. Experimental results have shown an increase in traffic efficiency based off of a decrease in total waiting vehicles and time.