This master thesis investigates the possibilities of using GPS-information from trains to create a model of a railway system, register the routes of the the trains and predict how the trains will travel in the railway system.The thesis work was done at Icomera AB which is a company that specializes in internet connected vehicle solutions.We describe how the GPS-information is sent from the trains in the form of samples and investigate limitations and possibilities with the available information. About three million samples from 83 SJ-trains were analysed and the samples were filtered with the intent of removing erroneous information. We here found that about 30 % of the samples from the trains were unreliable.A solution for how the samples from the trains can be used to create a model of the railway system traveled by the trains is presented. We use a large set of historical samples and use cluster nodes and Hermite curves to generate a graph that represent the railway system with stations, nodes and the railway itself. A fully automatic solution is not presented since it did not fit into the time frame of the thesis; instead we created a simple tool which can be used to manually finish the model of the railway system.Furthermore, we describe how the railway system and the samples from the train can be used to register the routes the trains travel.

This master thesis investigates the possibilities of using GPS-information from trains to create a model of a railway system, register the routes of the the trains and predict how the trains will travel in the railway system.The thesis work was done at Icomera AB which is a company that specializes in internet connected vehicle solutions.We describe how the GPS-information is sent from the trains in the form of samples and investigate limitations and possibilities with the available information. About three million samples from 83 SJ-trains were analysed and the samples were filtered with the intent of removing erroneous information. We here found that about 30 % of the samples from the trains were unreliable.A solution for how the samples from the trains can be used to create a model of the railway system traveled by the trains is presented. We use a large set of historical samples and use cluster nodes and Hermite curves to generate a graph that represent the railway system with stations, nodes and the railway itself. A fully automatic solution is not presented since it did not fit into the time frame of the thesis; instead we created a simple tool which can be used to manually finish the model of the railway system.Furthermore, we describe how the railway system and the samples from the train can be used to register the routes the trains travel.