Tactful networking as a cornerstone for opportunistic human-aware D2D communication.

Abstract

Among the mobile network challenges, there is data traffic growth, an increasing number of connected devices, spectrum limitations, and costly infrastructure updates. In this context, opportunistic D2D communication appears as a solution to offload data, increase capillarity, deliver content in specific scenarios (e.g., emergency) and foster innovative applications and services. Previous standards like LTE discussed D2D broadly. Still, recent research remarks that 5G/6G will be D2D's real enablers, given the possibility of using human-behavior big data from users' devices. Previous opportunistic D2D forwarding algorithms extracted mobility characteristics for improving content delivery cost-effectiveness. Nevertheless, most initiatives dealt with traditional metrics due to constraints, such as limited availability of real traces or the lack of a human-centered networking vision. Next-generation solutions need a more in-depth vision of human aspects hidden into datasets. Moreover, scenarios practical to mobile carriers lack evaluation. To this end, this thesis guides the reader through the whole process for building a novel Tactful Opportunistic COmmunicaTion Strategy (TOOTS). TOOTS leverages wireless encounter patterns, temporal, spatial, geographic, and direction awareness to improve cost-effectiveness content delivery in a mobile networks scenario. The proposal consists of: learning human-aspect state-of-art best practices; achieving insights to improve strategy's performance; using, proposing, and analyzing human-aware metrics; combining metrics and insights into the strategy targeting an improved performance in a more realistic mobile scenario; and finally, evaluating the strategy through its strengths and shortcomings. This thesis shows that TOOTS improved the performance of an opportunistic content delivery scenario in terms of overhead, delivery rate, and latency by following this process. First, to achieve the results, we survey the human aspect in networking solutions for over a decade. We found that there is an evolution in how the human links to networking challenges. This broader human-aware vision culminates into the Tactful Networking perspective, which we introduce to follow, including discussions concerning application examples and insights. We complete the literature review by discussing state-of-art opportunistic forwarding strategies, including contributions, gaps, and open issues. Second, we discuss a framework for enhancing raw human data hidden into datasets. A case study with the MACACO dataset validates the framework through trace characterization results and analysis. Following, we characterize MACACO, NCCU, and GRM datasets to bring insights and to validate TOOTS. The presented and analyzed strategy's metrics come from such datasets' characterization results and insights. Third, we introduce TOOTS, targeting cost-effective content delivery in a mobile scenario. Afterward, we present the formal definition for choosing the disseminator nodes problem (TOOTS' 1st phase). As motivation, we analyze the overhead and latency of the Epidemic forwarding. Subsequently, we introduce TOOTS' two phases: a Tactful Dissemination Policy and a forwarding algorithm. Finally, we compare TOOTS' effectiveness on its phases with enhanced Store-wait-forward, Epidemic, and Bubble Rap algorithms through real and synthetic traces. Among the main outcomes, in a scenario with a 30m range, TOOTS reached a 100% delivery ratio with 28% and 73% reduced delivery latency, and with 16% and 27% reduced overhead, respectively, in the real and synthetic datasets. With a restricted (10m) range, TOOTS was the fastest strategy and the only one that delivered 100% of the contents within the real trace. Consequently, this thesis contributes to addressing the opportunistic content delivery problem in cellular networks by providing a cost-effective human-aware opportunistic solution. The strategy can be applied in different scenarios to assist the operator in delivering content without not necessarily using their legacy network, e.g., by exploring the capillarity of their mobile users, for data offloading and other applications.