International Journal of ITS Research, Vol. 6, No. 1, 2008 Preliminary Analysis on Link Travel Time for Probe-Based Estimation Method by Microscopic Simulation Qiang Li *1 Tomio Miwa *2 Taka Morikawa *3 Department of Civil Engineering, Nagoya University *1 (Furo-cho, Chikusa-ku, Nagoya City, Japan, 464-8603, +81-52-789-3730, liqiang@trans.civil.nagoya-u.ac.jp) Department of Civil Engineering, Nagoya University *2 (Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan, +81-52-789-3565, miwa@civil.nagoya-u.ac.jp) Graduate School of Environmental Studies, Nagoya University *3 (Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan, +81-52-789-3564, morikawa@civil.nagoya-u.ac.jp) A set of travel time reports from probe vehicles is a sample. The adequate sample size required to estimate mean travel time of all vehicles (population) reliably has been studied in many literatures. However, in these literatures, the population is simply assumed as normal or approximate normal. In this paper, the properties of the population are discussed. The population is divided into three stages and the properties of the population in each stage are described in qualitative analysis and the description is verified by microscopic traffic simulation model. From this, some conclusions that are important to develop probe-based estimation method are made. Keywords: Link Travel Time, Probe Vehicle, Population, Simulation 1. Introduction Travel time-based measure, such as mean travel time, space-mean speed, and delay, is easy to understand by both the professional transportation community (transportation engineers, planners, and administrators) and the traveling public (commuters, business persons, and consumers). Travel time-based measure is also applicable across modes and is a common measure of effectiveness for all modes. These advantages make travel time-based measure extremely powerful, versatile, and desirable and an increasing number of transportation agencies are switching to travel time measure to monitor traffic condition [1]. Travel time or space-mean speed can be obtained by probe vehicles (PVs) directly and there is therefore a growing attention to probe vehicle technique with the progressive implementation of Advanced Traveler Information Systems (ATIS). However, the cost and capacity of the communication between PVs and the operation center impose restrictions on the number of PVs. Therefore, it is impossible to make all vehicles as PVs. In other words, a set of travel time reports from PVs on a link in a time interval is a sample. The adequate sample size required to estimate mean travel time of all vehicles (population) reliably has been an imperative issue since probe vehicle was recognized as a method to collect traffic information and there are many literatures, for example [2-7]. In these literatures, the population is simply assumed as normal or approximate normal and the properties of the population have not received attention. In this paper, the properties of the population that are important to PV-based estimation are discussed. In qualitative analysis, the population is divided into three stages and the properties of the population in each stage are described. Then, a microscopic traffic simulation model is employed to verify and extend the description in the qualitative analysis. 2. Qualitative Analysis 2.1. Definition of link travel time This section clarifies what is meant by the term ‘link travel time’ in this paper. Different definition will lead different properties of link travel time and thus it is necessary to define the concept explicitly. Traditionally, the link travel time is defined as the travel times of all vehicles traveling a link during a time interval regardless of turning movement at the link’s upstream and downstream intersections; that is, the effect of turning movement at intersection is neglected in the population definition (e.g., [4, 5, 7, 8]). The traditional traffic detectors, such as loop detector, cannot observe a vehicle’s turning movement at a link’s ends and it is one of the reasons why the effect of turning movement is neglected in the link travel time definition. 21 International Journal of ITS Research, Vol. 6, No.1, July 2008