Traffic flow prediction using Kalman filtering technique

Abstract

Traffic flow prediction is an important research problem in many of the Intelligent Transportation Systems (ITS) applications. The use of Autoregressive Integrated Moving average (ARIMA) or seasonal ARIMA (SARIMA) for traffic flow prediction requires huge flow data for model development and hence it may not be possible to use ARIMA in cases where sufficient data are unavailable. To overcome this problem, a prediction scheme based on Kalman filtering technique (KFT) was proposed and evaluated which requires only limited input data. Only previous two days flow observations has been used in the prediction scheme developed using KFT for predicting the next day flow values with a desired accuracy. Traffic flow prediction using both historic (previous two days flow data) and real time data on the day of interest was also attempted. Promising results were obtained with mean absolute percentage error (MAPE) of 10 between observed and predicted flows and this indicates the suitability of the proposed prediction scheme for traffic flow forecasting in ITS applications.