Analyzing public transportation mobility data for networking purposes

Utilizing vehicles for networking purposes has always been a challenge. This is mainly due to the minimum density of connected-vehicles required. The locations of these vehicles should be shareable and reasonably predictable for efficient position-based routing protocols to be implemented. Their Vehicle-to-Vehicle (V2V) communication cooperation should be well-incentivized for efficient networking to be realized. Regular vehicles struggle to have all of these properties. Public transportation vehicles, on the other hand, are well-positioned in this regard; their number is proportional to the number of city residents while being uniformly distributed throughout the day, their locations have no privacy concerns while being highly predictable and their V2V communication cooperation is easily enforceable by the single administration authority they usually have. With efficient networking, public transportation vehicles can become the reliable communication backbone for other vehicle categories. In order to investigate their networking potential, we present for the firs time, in this paper, a data analysis study of realistic public transportation mobility datasets representing the Grand River Transit bus service offered throughout the Region of Waterloo, Ontario, Canada. We show both the data preprocessing and processing phases. The processing phase is mainly based on discovering bus groups using hierarchical clustering. This is done while varying the minimum degree of intra-cluster connectivity and the maximum intra-cluster communication range. Based on this data analysis approach, we show the promising networking potential of public transportation vehicles and provide design guidelines for future networking solutions utilizing them.