CPCR: Contact-prediction clustering-based routing in large-scale urban Delay Tolerant Networks

Routing in Public Transport Networks is particularly challenging due to the high mobility, rapidly changing topology. Though clustering of nodes can aid forwarding decision in these Delay Tolerant Networks (DTNs), centralized clustering cannot adapt to large scale DTNs for the temporary connections. Moreover, it is critical to choose proper clustering metrics. In this paper, we propose a contact-predict clustering-based routing algorithm, CPCR, for large-scale urban DTNs. This algorithm computes the probability for each node pair and cluster nodes into higher contact-strength clusters using nodal local contact information in a distributed way. Depending on the clusters, intra-cluster and inter-cluster routing strategies will be adopted. Extensive simulations indicate that CPCR maintains relatively stable clusters and enhances the routing performance. The results demonstrate that clustering-based routing algorithm is rational and promising for large-scale urban DTNs.