Terrain based routing protocol for sparse ad-hoc intermittent network (TRAIN)
Traditionally, routing protocols for ad-hoc networks have been developed without taking into consideration the effects of the surrounding environment. In this thesis, we propose and investigate a new approach of terrain and location based routing and its effect on the routing layer on sparse ad-hoc networks. This approach is particularly important for sparse networks, where relatively large node separations can result in periods of disconnectivity. Terrain blockage compounds the problem, as the likelihood of inter node communication is further diminished. We anticipate that terrain and location based routing would cause significant decrease in latency and would increase the throughput of sparse ad-hoc mobile nodes. It would also provide a new area of research which would provide more realistic and intelligent view of the surrounding. The use of terrain maps from U.S. geographical survey provides a realistic view of the network environment, through which a node can determine whether there is a Line of Sight (LOS) or Non Line of sight (NLOS) path to another node by combining location awareness and terrain data. When nodes are mobile, knowledge of trajectories can be used to predict future locations, and combined with terrain information, to forecast link conditions at a future time. As an additional feature of our routing protocol; when a node has data to send it calculates the stability of the link based on terrain and trajectory information and deterministically predicts the duration for which this link is going to be stable. For the performance assessment of using location, trajectory and terrain information, a simulation model including the terrain map of Yellowstone National Park have been developed using the discrete event simulator OPNET Modeler TM 10.5. Simulation scenarios have been created for single hop as well as for multi-hop networks based on the characteristics of 802.11 wireless technology. From the results, we have observed that using the terrain and location information provides nodes with an intelligent and realistic view of the network topology. Our approach has proved effective in make deterministic routing decisions based on LOS predictability, thus increasing the reliability, stability, and throughput of the network.