Section: New Results

Characterizing and measuring urban networks

Participants: Marco Fiore, Diala Naboulsi, Razvan Stanica, Sandesh Uppoor

The goal of Sandesh Uppoor's PhD thesis [4] was to model and understand the mobility dynamics of high-speed vehicular users and their effect on wireless network architectures in an urban environment. Given the importance of developing the study on a realistic representation of vehicular mobility, we first survey the most popular approaches for the generation of synthetic road traffic and discuss the features of publicly available vehicular mobility datasets. Using original travel demand information of the population of a metropolitan area (Cologne area, Germany), detailed road network data and realistic microscopic driving models, we propose a novel state-of-art vehicular mobility dataset that closely mimics the real-world road traffic dynamics in both time and space [25] . We then study the impact of such mobility dynamics from the perspective of wireless cellular network architecture in presence of a real-world base station deployment. In addition, by discussing the effects of vehicular mobility on autonomous network architecture, we hint at the opportunities for future heterogeneous network paradigms and demonstrate how incomplete representations of vehicular mobility may result in over-optimistic network connectivity and protocol performance [8] .

Motivated by the time-evolving mobility dynamics observed in our original dataset, we also propose an on line approach to predict near-future macroscopic traffic flows. We analyze the parameters affecting the mobility prediction in an urban environment and unveil when and where network resource management is more crucial to accommodate the traffic generated by users on-board. Such studies unveil multiple opportunities in transportation management either for building new roads, installing electric charging points, or for designing intelligent traffic light systems, thereby contributing to urban planning.

Despite the growing interest in a real-world deployment of vehicle- to-vehicle communication, many topological features of the resulting vehicular network remain largely unknown. We still lack a clear understanding of the level of connectivity achievable in large-scale urban scenarios, of the availability and reliability of connected multi-hop paths, and of the evolution of such features over daytime. In [14] , we investigate how the instantaneous topology of the vehicular network would look like in the case of a typical middle-sized European city, using the example of the Cologne mobility trace. Through a complex network analysis, we unveil the low connectivity, availability, reliability and navigability of the network, and exploit our findings to derive network design and usage guidelines.

Community urban sensing is one of the emerging applications enabled by the growing popularity of mobile user devices, like smartphones and in-vehicle monitoring systems. Such devices feature sensing and wireless communication capabilities, which enable them to sample large-scale phenomena, like air pollution and vehicular traffic congestion, and upload these data to the Internet. In [10] , we focus on the above scenario and investigate the level of accuracy that can be achieved in estimating the phenomenon of interest through a mobile crowdsourcing application. Specifically, we take a signal processing-based approach and leverage results on signal reconstruction from sets of irregularly spaced samples. We apply such results to a realistic scenario where samples are collected by vehicular and pedestrian users, and study the accuracy level of the phenomenon estimation as the penetration rate of the sensing application varies.

The growing ubiquity of mobile communications has offered researchers new possibilities to understand human mobility over the last few years. In [22] , we analyze Call Detail Records (CDR) made available within the context of the Orange D4D Challenge, focusing on calls of individuals in the city of Abidjan, Ivory Coast, over a period of five months. Our results illustrate how aggregated CDR can be used to tell apart typical and special mobility behaviors, and demonstrate how macroscopic mobility flows extracted from these cellular network data reflect the daily dynamics of a highly populated city. We discuss how these macroscopic mobility flows can help solve problems in developing urban areas.