Alex M. Suarez

.

Faculty Mentor: Yosuke Bando

Direct Supervisor: Henry Holtzman

Home University: University of Puerto Rico, Mayaguez

Major: Computer Engineering

 .

Biography

Senior Computer Engineering student at the University of Puerto Rico, Mayaguez Campus. Born and raised in the island of Puerto Rico. My research interests revolve around technology that’s pervasive, interactive and simple to use. I tend to focus on high level applications work, low level systems development and electronics design. Young adult novels are my vice, along with an interest in human behavior and social mechanics.

Abstract

Server-aided, globally-aware content dissemination over delay tolerant mobile ad-hoc networks

The proliferation of wirelessly enabled mobile devices has renewed the interest in Mobile Ad Hoc Networks (MANETs). With the increasing gap between mobile data traffic and the current network infrastructure’s ability to cope, MANETs could become a platform with which to alleviate congestion. The existing infrastructure could offload traffic to a network formed out by people’s mobile devices. This MANET would, in turn, opportunistically disseminate the data between devices. Some of the key challenges with these networks lie in dealing with their dynamic topology, intermittent connectivity and inherent latency. Delay Tolerant Network (DTN) protocols, which specifically target sparsely connected networks, address these issues. These protocols follow a “store and forward” approach to increase the probability of message delivery. For those that selectively disseminate content, they typically do so with the information nodes infer from their surroundings. This knowledge is inherently limited and may result in suboptimal routing decisions. Here, we explore a protocol that communicates sparsely with a server as a means to forward content based on complex node social patterns. Such calculations would likely be impossible to perform on the mobile device. By aggregating all the data collected by each node, the server computes a social graph of the network to probe for patterns. A profile is then made for each node for them to use in routing decisions. We test the performance of this approach by simulating the protocol using a modified version of an open source DTN simulator, with a realistic node mobility model of people’s working day.