Estimation of traffic density based on past density information for adaptive traffic management

Traffic congestion which causes economic, environmental and even individual psychological troubles not only in Turkey but all over the world, is a very serious problem. Developed intelligent traffic systems in recent years aim to produce solutions to this problem by monitoring the traffic and applying adaptive decision strategies based on estimations about the future situations. The core problem of these systems consists of the estimation of traffic density reliably. In the literature part of the thesis we will summarize the most commonly used density estimation methods. In the application part of the thesis we will examine four different algorithms proposed for traffic density estimation which are inspired by the methods used to estimate the spectral holes in cognitive radio applications. Data used in this study is received from Istanbul Traffic Control Center and has been converted into ternary and binary versions depending on the average speed of the traffic flow. In the proposed algorithms, density state at the 60th minute of the considered road is estimated by looking at the past 50 minutes of density data of the same road or adjacent two roads. Different simulations have been performed using these algorithms and results are evaluated based on several performance criteria. Keywords: Density estimation, traffic, congestion, cognitive radio