Feasibility of dynamic traffic management based on data exchange between a traffic centre and vehicles. In theory data originating from vehicles can be applied to support dynamic traffic management. This concept is called Floating Car Data (FCD), and has been extensively studied for the past decade. The main advantage of an FCD system is that it allows coverage of an extensive area in a short amount of time. This is in contradiction with the approach of extending the classical sensing infrastructure based on inductive loops and cameras. The downside is that the FCD technology is less mature then the classical infrastructure. It is not clear if in reality a FCD system will be able to effectively provide the same services with the same quality as existing systems. The final cost and organizational approach of an FCD roll-out is also uncertain. Hence policy makers face difficulties when deciding rather to invest in the further expansion of the classical infrastructure based on inductive loops and cameras, or to invest in the roll-out of an FCD system. The goal of this report is therefore to provide well-founded insights in the feasibility of using FCD in the context of dynamic traffic management. In contradiction to many existing studies we will adopt a top down approach instead of a bottom up approach. Within this research context we aim to answer the following questions: What are the requirements for the FCD data? What is the required penetration rate? Which data should be part of the FCD samples? What is the sampling and transmission interval? * Which functionality can be provided using this data? Are they different for the different types of roads (highway, arterial road, urban environment)? * Are there reliability issues? What is the impact of the FCD system on the supporting mobile data network in case of high traffic concentrations? * How much will the roll-out of an FCD system cost? and * How can the roll-out of an FCD system be organized best? To define an answer to these questions the report starts with an extensive literature study. From this study estimations can be derived regarding the first two groups of questions. However, they are not accurate enough. To further refine them a specially developed platform is utilized. This platform is based on microscopic traffic simulation. Concerning the last three groups of questions (impact on the mobile data network, cost, organization), no existing studies could be found in literature. To further research these questions several techniques are applied: an adjusted model for the determination of the load on a mobile data network that was developed in previous work, a specially developed cost model and a specially developed organizational model (so-called value networks). Based on the obtained results it can be concluded that it is best to aim for a FCD configuration with a penetration rate of 1% and a sampling rate of 10 seconds. Samples are first stored locally and contain accurate information regarding position and speed of the vehicle, and exact moment of sampling. Every 30 seconds an aggregate of 3 samples is then sent to the FCD server. During connection setup a security optimization is applied: the so-called SSL restart handshake. An FCD system as described above will be able to make accurate speed estimations in a highway environment. In this environment it will also be able to accurately determine the location of an incident and the tail of a traffic jam. On arterial roads and in urban environments the FCD system will not be able to make accurate speed estimations. It will however be able to classify road segments in two categories: normal or congested traffic. In this environment it will also be able to accurately determine the location of an incident and the tail of a traffic jam. The impact of the proposed FCD system on the mobile data network can be neglected. Hence no reliability problems due to network overload are expected. Every month, an FCD client will transmit about 1 MB of data to the server. Regarding cost the most feasible option on the short and medium long term is to buy the FCD data from an external party. Currently several companies possess their own FCD data. Examples are TomTom, Be-Mobile, Coyote Systems and Wikango. It should be mentioned that this list is not exhaustive, a market study should be performed to define it in more detail. This is however out of the scope of this report. A careful cost estimation was made, an annual cost of 160.000 euro seems feasible. However, this cost estimation should be further verified by the government in bilateral contacts with the different companies. Regarding organizational aspects, the most feasible option on the short and medium long term again is to buy the FCD data from an external party. For the government to organize an FCD system themselves, either being an open or a closed system, requires the coordination of much more actors compared to the case of just buying the data. Based on these conclusions it is recommended to the government to perform a market study to finalize the list of possible FCD data suppliers, to contact all of them and to decide based on the received feedback if the roll-out of an FCD system is desired and possible. (Author/publisher)
Abstract