An Analysis of the Effects of Intelligent Location Prediction Algorithms on Greedy Geographic Routing in Mobile Ad-Hoc Networks

Geographic routing uses the physical locations of devices for routing purposes instead of the conventional routing protocols which make use of logical addressing schemes representing an abstract hierarchy. Greedy geographic routing is a popular method favoured for its efficiency and its simplicity that considers only one hop neighbours. Greedy routing needs only minimal network information and as such is resilient to change and dynamic behaviour. Recent advances have seen the development of location prediction algorithms that use a variety of methods to determine a node’s future location based on their previous movements. Such prediction schemes can potentially benefit greedy geographic routing by allowing nodes to make routing decisions based on where a node will go rather than where it was according to the last update. The use of location prediction algorithms therefore allows geographic routing protocols to make decisions that are more intelligent, thus not only improving routing performance but providing a crucial step towards fully autonomous computer communications. The main contribution of this paper is the implementation and analysis of two existing location prediction schemes on top of the existing Greedy Perimeter Stateless Routing (GPSR) protocol ran in greedy mode.