Estimators for global information in mobile opportunistic network

Lack of global knowledge in a delay tolerant network proves detrimental, for a replication based routing strategy, in terms of resource utilization. It has often been noticed that nodes keep on replicating a message although it has already been delivered to the destination since such information remains unknown to most of the nodes in the network. Moreover recovery mechanisms (like VACCINE for epidemic routing) take time to propagate and incur additional transmission overhead. Global information such as number of nodes in the network, node meeting schedule, and whether a message has already been delivered to the destination is vital for optimal network performance. In this paper we propose model for estimating some of these vital information and show that replication based routing algorithm like epidemic routing, can perform better with minimum resource utilization. First, we use a uniformly minimum variance unbiased estimator (UMVUE) to estimate the number of nodes in the network. Second, we estimate the number of replicas of any given message in the network, at any given instant (t). Third, we show that epidemic routing with a node population of N can perform best with only N/2 message replicas for any given message and that further replication only leads additional overhead without any performance improvement. We show the simulated results using random waypoint mobility model and Shanghai taxi trace.