Design and Evaluation of Floating Content Services for Vehicular Applications

Manzo, Gaetano (2020). Design and Evaluation of Floating Content Services for Vehicular Applications. (Dissertation, Institute of Computer Science, Faculty of Science)

	Text PhD_Thesis_G_Manzo_.pdf - Other Restricted to registered users only Available under License BORIS Standard License. Download (14MB)
	Text PhD_Thesis_GManzo_with_consent.pdf Restricted to registered users only Available under License BORIS Standard License. Download (15MB)

By offloading vehicular data transfers from the telecommunication infrastructure to direct vehicular-to-vehicular communication, opportunistic communications reduce infrastructure investment, overload, and latency. However, performance studies of opportunistic communication models in vehicular networks mainly focus on content persistence without accounting for the system conditions that enable desired performance, such as the effectiveness with which the content object is replicated and made available. Thus, how to efficiently engineer a vehicular application characterized by an opportunistic communication model remains an open and challenging issue, crucial for the provision of high quality-of-service for vehicular applications.

This thesis aims to provide the tools to efficiently engineer a vehicular application characterized by opportunistic network models. We leverage on Floating Content (FC), an infrastructure-less opportunistic communication scheme that binds the local dissemination of information. The contributions of this thesis are summarized as follows. First, we design an enhanced method for configuring FC schemes in vehicular ad hoc networks. Our results suggest that it is always possible to find a reasonable size of the communication area such that the content object persists for the whole target duration. Second, we propose approaches for fine-tuning FC parameters (e.g., replication and caching) to guarantee a minimum target performance level while minimizing resources used, such as bandwidth and storage. Numerical evaluations show that our deep learning architecture provides content replication and storage strategies much more efficient than analytical techniques. Third, we provide an efficient communication scheme for content retrieval in vehicular networks that adapts to a wide range of network topologies and settings. Our approach outperforms delay-tolerant models reducing the content storage by 30% and the content replication by 43% with designed content availability, success content delivery, and delay targets.

Our approaches lay the foundation for the practical use of vehicular applications based on FC schemes in real scenarios.

Interest & Impact

Downloads

30 since deposited on 17 Nov 2020
8 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item