| dc.description.abstract | Geographic routing algorithms are an attractive alternative to traditional ad hoc routing algorithms. Although geographic routing like Greedy Perimeter Stateless Routing (GPSR) is efficient to mobile ad hoc wireless networks, it requires that nodes be aware of their physical positions to inform their neighbors. In addition, if there are exist geometric holes within the network topology, which is called “dead end”, routing across dead ends in GPSR [5] will lead to a lot of redundant routing paths during data transmission and overloaded nodes in the boundaries of dead ends. In this thesis, we proposed an approach, according to virtual coordinate system based on detecting possible dead ends among nodes to improve the routing performance of existing geographic routing algorithms. When exchanging beacon message among nodes, each node will construct a local Voronoi diagram on the basis of the message information which receives from its neighbors simultaneously, each node draws perpendicular bisectors with all of its neighbors and then it checks if any segments on its transmission circle are not covered, which is called “dead end area”. Encountering those nodes which is existed the dead end area when data packets were greedily forwarded. It may cause transmission end up. In order to reduce this kind of situation, we assume every mobile node within network was a positive electron and destination node is a negative electron. In accordance with field theory [25] the packet forwarding among the shortest path from source node to destination node is inevitable, if the node exist one or more dead end area within its broadcast range, there was existed an pseudo positive electron which locate out of node’’s broadcast range, the pseudo electron’s location was sum of two vectors, one is mobile node to the start point of dead end area, and the other is mobile node to the end of dead end area. And pseudo electron will give a repulsive force to mobile node which will force virtual coordinate of nodes moving away from the dead end area. For the purpose of reducing the risk of encountering dead end situation. After the mention above, the nodes in the network are aware of the relative virtual coordinates and each of them was notified to the neighbors by exchanging beacon message. In the final of this study, results of the simulation show that our approach can improve geographic routing more efficient, In addition, our approach is resilient to various network shapes especially for U-shaped-like topology within the sparse network. In this thesis, some programming effort was implemented for comparison using the NS2 simulator platform[17]. The simulation results show that the average number of dead end occurrences was significantly reduced about 5 to 10 times of packet to encounter the dead end situation. When our dead end avoidance enhanced mechanism was applied. The packet delivery ratio was improved also. | en_US |