無線網路中的加權積極式幾何路由

Geographic routing ensures network scalability through routing decisions made only from local state information. This feature is essentially at the cost of routing path length particularly in networks having sparse nodes or topological holes. We present ways to mitigate the unappealing factor by further exploring local state information. We first propose a forwarding scheme that utilizes neighbor positions with one-step forward expectation for making routing decisions. In terms of total forwarding distance, the scheme turns out to outperform the locally optimal one, greedy forwarding. Furthermore, we select the maximum angle of a node with its circular contiguous neighbors as the representative angle that economically characterizes local topology and is also exchanged with neighbors. Besides analyzing its valuable inferences, we prove that nodes with representative angle larger than 5pi/3 can be logically removed from networks without breaking a given source and destination originally connected. For reliable routing, we then propose joint forwarding and perimeter routing schemes that thoroughly utilize representative angles and associated design principles to reduce path hop counts. In particular, our novel routing schemes not only judiciously determine a relay node from a trimmed network graph at each hop in greedy forwarding but also allow perimeter routing to walk on a virtually trimmed planar graph. Consequently, routing path hop counts significantly shorten as compared to that performed by greedy perimeter stateless routing (GPSR), validated by simulations.