Quasi-opportunistic contact prediction in delay/disruption tolerant network

Communication in Disrupt Tolerant Networks (DTNs) is a challenge because it presumes the absence of an end-to-end path at the time of sending a message to a destination. An efficient selection of a contact node to forward a message is a key in the routing process. Prediction techniques can be used to assist in routing decisions. This paper presents an approach to predict the next node and the moment of contact, based on artificial neural networks (ANN). The hit rate of a prediction function based on ANN has been compared by simulation with the hit rate of a function based on contact frequency. The ANN showed better results in a quasi opportunistic scenario. We expect to apply this approach to support a end-to-end routing protocol.