Fast Adaptive Mean Shift Clustering and Cluster Based Routing Protocol for Delay-Tolerant Mobile Networks

Delay Tolerant Networks is fundamentally an opportunistic communication system, where communication links only exist temporarily. Due to the lack of continuous communications among mobile nodes and possible errors in the estimation of nodal contact probability, convergence and stability become major challenges in distributed clustering in Delay Tolerant Mobility Networks. In this paper, the basic idea is to group mobile nodes with similar mobility pattern into a cluster and routing the data. In cluster formation our objective is to group all nodes with high pair-wise contact probabilities together, a node joins a cluster only if its pair-wise contact probabilities to all existing members are greater than a threshold. A node leaves the current cluster if its contact probabilities to some cluster members drop below the threshold. Clustering in DTMN is unique and non-trivial, because the network is not fully connected. Due to the lack of continuous communications, mobile nodes may have inconsistent information and therefore respond differently. To this, an Exponentially Weighted Moving Average (EWMA) scheme is used for online updating nodal contact probability. By using this nodal contact probabilities cluster formation and gateway selection are done. And a set of functions Sync (), Leave (), Join () also used. At last, the gateway nodes exchange network information and perform routing in an adaptive way.