Optimal Replication Based on Optimal Path Hops for Opportunistic Networks

Abstract

Opportunistic Networks are highly mobile networks which may lack a reliable path between some source and destination. Therefore, this type of network uses Store, Carry and Forward and delivers messages based on hop-by-hop routing. Epidemic is the simplest routing protocol for Opportunistic Networks, as it replicates messages to all encountered nodes. For messages spread by Epidemic replication, we study and analyze the trade-off between the messages' delivery ratio, delay, and overhead. We consider that all members of the network, at any given time, know the amount of relay nodes that pass the message and each message carrier knows the amount of infected nodes that already have acquired the message. We address the problem of deriving the optimal closed-loop control for the replication strategy in our network. We draft this issue as a controlled, discrete-time and finite-state Markov Chain with an All Hops Optimal Path formulation. In real life scenarios, however, due to the intermittent correspondence in Opportunistic Networks the nodes do not have insight of the network's global state. We try to solve this issue by obtaining an Ordinary Differential Equation approximation of the Markov Chain for the replication process of messages. Furthermore, our model considers All Hops Optimal Path as network graph analysis for the optimally controlled replication. Lastly, we present the performance evaluation of these replication control conditions in finite networks. Our results show that this proposed Optimal Replication Based on Optimal Path Hops performs better than the omni-directional and contact-based Epidemic routing replication.