Improving Node Popularity Calculation Using Kalman Filter in Opportunistic Mobile Social Networks

Abstract

Opportunistic mobile social networks (OMSNs) exploit human mobility to physically carry messages to the destinations. Routing algorithms in these networks typically favour the most popular individuals (nodes) as optimal carriers for message transfers to achieve high delivery performance. The state-of-the-art routing protocol BubbleRap uses a cumulative moving average technique (called C-Window) to identify a node's popularity level, measured in node degree, in a time window. However, our study found that node degree in real-life OMSNs varies quickly and significantly in time, and C-Window moreover slowly adapts to this node degree changes. To tackle this problem, we propose a new method of node degree computation based on the Kalman-filter theory. Using simulation, driven by real human contact traces, we showed that our approach can increase BubbleRap's performance, in terms of delivery ratio and traffic (load) distribution fairness.