| With the development of wireless communication, cheap and powerful have been widely used. There are some hot topics of wireless mobile communication: improve adaptability and performance, reduce cost etc. Wireless communication networks can be classified into infrastructure networks and mobile ad-hoc ones. A mobile ad-hoc network is a self-organized autonomous system where no fixed infrastructures are required. As nodes of this network are able to move and re-organize arbitrarily, the network topology is likely to change dynamically and unpredictably. Due to its nature of multi-hop, mobility, scalable size, device heterogeneity, constrained bandwidth and power, designing a practical routing protocol becomes a major challenge. Recently, many routing protocols have been proposed for MANET. This thesis studies two MANET routing protocols: AODV and DYMO.Many ad-hoc routing protocols are based on flooding (or some variants of it). Optimized or not, many redundant routing messages are propagated. This thesis introduces a gossip-based approach, where each node only forwards messages with certain probabilities to reduce the number of messages. Gossip exhibits bimodal behavior in sufficiently large networks: in some scenarios, the Gossip fails quickly and rare nodes get messages; in the others, almost all nodes get messages. This thesis studies the relationship among the successful ratio and the gossip probabilities as well as the topology of networks, at the same time analyzes the phase transition phenomenon of Gossip. We get optimal gossip probabilities via simulations: a gossip probability of 0.65 suffices to ensure success of almost every execution in a random network with a degree of 8. Considering unpredictable topology, we also propose a self-adaptive gossip method. It is shown in simulations that the gossip-based DYMO protocol results in a significant performance improvement: the packet delivery percentage is increased by 13% and the normalized routing load is reduced by 35%. |
PDF Full Text Request