The Department of Computer Science at the University of Cyprus cordially invites you to the PhD Defense entitled:
Distributed Traffic Information Systems Based on V2X Communication in Large-Scale Urban Environments
Speaker: Mr. Nicholas Loulloudes
Intelligent Transportation Systems (ITS) aim to assist commuters in taking informed decisions concerning travel safety and efficiency. Vehicular Ad Hoc Networks (VANET) are one of the key platforms for ITS, wherein vehicle-to-vehicle and vehicle-to-infrastructure communication - collectively designated as V2X - can support information dissemination. With high-fidelity sensors becoming mainstream in modern vehicles, VANETs are expected to adequately support the diffusion of disparate sensory artifacts, including the prevailing traffic conditions. In turn, V2X-based Traffic Information Systems (TIS) will provide drivers with dynamic route planning and congestion estimates. Nevertheless, the intrinsic properties of mobility, the lack of clear understanding of vehicular network connectivity, as well as the stringent time and quality constraints for traffic data diffusion, impose significant barriers in the process of designing, implementing and deploying TIS. Moreover, the research community has yet to adopt thorough and realistic performance evaluations approaches for such systems and their peripheral components thereby introducing unnecessary risks of failure, simply because they will confront a demanding environment they were not designed for. This talk will present our research efforts towards providing solutions to the aforementioned problems. To investigate the potentials and understand how distributed VANET-based TIS will eventually be realized, we study through extensive analytical and simulative methods the following problem domains: (i) VANET dynamics in Urban Environments, (ii) Traffic Sensing and Acquisition, and (iii) Large-Scale Traffic Information Dissemination. At first, we explore the basic premise behind V2X, specifically the constantly evolving spatio-temporal nature of urban-based VANETs through which traffic information will flow. Here, V2X connectivity is modeled as sequences of contact graphs wherein complex network analysis measures are employed to gain insights on the structural properties of VANETs from a city-wide perspective. In contrast to other low-dimensional networks, we unveil the VANET's high sensitivity to fragmentation, extremely variable connectivity and total lack of small-world and scale-free features. Additionally, we present why and where particular network phenomena occur in the underlying road network and exploit the implications in the design of effective information dissemination mechanisms. Next, we consider the constrained bandwidth of VANETs, and thus we investigate the possibility of in-vehicle data caching. Particularly, we examine its capacity to improve the efficiency of TIS by reducing the time-to-serve requests and network overhead, while abiding established time and quality constraints. Through large-scale realistic simulative evaluations, we discuss that the use of simple TTL-based replacement policies can achieve significant improvements under both normal traffic conditions and unscheduled traffic events. Finally, we present V-Radar a reactive, VANET-based, traffic information dissemination protocol for metropolitan environments. In contrast to other approaches in the literature, V-Radar enables the querying and acquisition of traffic information along a number of composite road-paths, starting from a vehicle's current position towards its final destination. Ultimately, this effective approach enables the driver to establish a more broad view of the traffic conditions that will be encountered further ahead in a timely manner. Finally, we present the analytical and simulation testbeds developed to support the above research, which are open-source released to the community for future studies.
Nicholas Loulloudes is a PhD candidate at the Department of Computer Science and member of the Laboratory for Internet Computing, University of Cyprus. His research interests are focused in the area of Vehicular Ad Hoc Networks (VANET), specifically spatio-temporal topology analysis via complex network science, design and optimization of data dissemination and routing algorithms. Nicholas has also strong interest for the areas of Elastic Cloud and Grid Computing. In 2010, Loulloudes was awarded a 2-year research grant (V-Sense: http://vsense.cs.ucy.ac.cy) co-funded by the Republic of Cyprus and the European Regional Development Fund, in order to investigate novel concepts in Vehicular Sensor Network infrastructures. Previously, he was on a research internship on Intelligent Transportation Systems with Siemens Corporate Research, Princeton, NJ, USA. His work was published in a Springer book chapter, conference proceedings such IEEE/ACM MASCOTS, IEEE MDM, Euro-Par, IEEE VNC, VTM2012 and ERCIM. Loulloudes is also serves as a frequent peer-reviewer in several international journals and conferences in the area of VANET, ITS and Mobile Computing. He has strong background in network and traffic simulation platforms such as NS-3, SUMO, TRANSIMS where he actively contributes in their respective open-source ecosystems.
|Other Presentations Web: https://www.cs.ucy.ac.cy/colloquium/presentations.php|
|Colloquia Web: https://www.cs.ucy.ac.cy/colloquium/|