Find in Library

Routing in MANET is very challenging due to movement of nodes. Predicting the node position and routing based on predicted positions helps to establish routing path with much longevity. Most predictions approaches are based on the past locations of the node. In this work a node location prediction based on the temporal and spatial characteristics with respect to its neighborhood is applied to estimate the probable locations using a hybrid model. A multi path routing protocol based on estimated probability locations with path diversion at necessary places along path is proposed for improved routing performance without larger packet overhead.