User traces analysis based on crowdsourced data

Our paper analyses the distributions of various parameters of GPS user traces collected in four measurement campaigns during 2008–2016 across Europe and US and including 1036 volunteer users and more than 43 million trajectory points. The aim is to shed new lights on eight selected parameters of user mobility models and to compare the distributions of these parameters between different measurement datasets and existing theoretical models. We find that the user-level based statistics for the angle distributions are rather independent on the environment, while the user-level based statistics for the other parameters such as speed, acceleration, flight and pause times and steps are environment dependent. We also find out that exponential and lognormal distributions are the most encountered distributions that fit best the user traces parameters. We also show that users are usually at maximum 15 Km away from the spatial means of their trajectories.