Gossip - based Information Spreading in Mobile and Social Networks

HUANG, YUFAN. Gossip-based Information Spreading in Mobile and Social Networks. (Under the direction of Dr. Huaiyu Dai.) We study the performance of gossip-based information spreading in different network models. Our study is conducted in two major different circumstances: mobile networks and social networks. In mobile networks, we mainly study two problems: single-piece information spreading with general transmission range and multi-piece information spreading in connected networks. Towards understanding the gossip algorithm, first, we study single-piece message spreading with a general transmission range. While the results in connected networks are relatively easy to obtain, some critical approximation fails to hold in the disconnected scenario, and we propose general mobile conductance with a new analytical method.With general mobile conductance, we observe a penalty factor of nr2 induced by network connectivity deficiency. We also combine the concepts of mobile conductance and static k-conductance to define a new metric, mobile k-conductance, which enables us to analyze multi-piece message spreading in connected mobile networks. In social networks, we consider the celebrated Preferential Attachment model. Inspired by the Preferential Attachment process, we propose a new non-uniform gossip algorithm called PA-gossip for effective information spreading in social networks. Largescale network simulation is conducted to demonstrate its great potential in facilitating information spreading in social networks. © Copyright 2014 by Yufan Huang