Transportation reliability based on information theory

In this study, a novel travel time reliability measure is proposed that is developed based on entropy from information theory. The authors present the classical methods of measuring travel time reliability. Classical reliability measures rely fundamentally on the statistical standard deviation quantity. However, entropy is a direct tool to measure the uncertainty present in any random variable. The quantity of entropy not only contains the standard deviation but also all other moments. Based on this, the authors constructed a novel travel time reliability measure that can be applied to any random variable regardless of its distributional characteristics. The authors then demonstrate the use of the proposed novel travel time reliability measure by applying it to travel time data obtained from stationary detectors. Furthermore, the authors extended this work to demonstrate measuring the reliability of networked roadway segments. In this study it is found that the algebraic structure of the travel time reliability of a network is consistent with the semi-ring min-max algebraic structure. This provides us with well established properties that can simplify the algebra over networks´ reliability.