Research On Geographic Routing And Voids Handling Techniques In Wireless Sensor Networks

Geographic routing algorithms for wireless sensor networks usually use greedy idea to select an adjacent node, which is closest to the destination node, as the next hop node, and then forward data to it. When the sensor nodes'density is relatively high and the nodes'residual energy is generally sufficient, the algorithm is efficient enough and the routing path nearly is the shortest path. But when the nodes'density becomes lower due to uneven node placement or part of nodes'failure, it leads to the appearance of routing voids. That is, a node can not find an adjacent node closer to the destination node, to forward data greedily. The existence of routing voids undermines the network connectivity, and even may lead to the failure of data transmission. Because each node makes routing decision according to the local information, these algorithms are scalable and can adapt to networks with rapid changes in topology. However, since the routing voids handling techniques often require global information to generate a planar graph of the whole network, consequently these techniques are not truly localized, and will affect the routing algorithms'practicality.To avoid routing voids, the paper proposes a predictive neighborhood feedback mechanism. Once a node encounters a routing void, it will notify their upstream nodes about the routing void. The upstream nodes then adjust the routing information to avoid forwarding data to this direction, thus resulting in a lower frequency of the startup of void handling techniques. The mechanism also can prevent the nodes with lower residual energy from depleting their energy rapidly due to forwarding other nodes'data, slow the formation of the new routing void as much as possible.In order to handling the above problem, this paper proposes an isolated subtree re-routing algorithm based on the idea of looking upon the nodes, which will route data to a routing void, as an isolated subtree. The algorithm only uses the local information corresponding to the subtree to find the key node of the subtree. All member nodes of the subtree select the key node as their local target node, and use the key links to establish new paths to the Sink node. Then, the paper additionally proposes a distributed isolated subtree split algorithm to find and use multiple key nodes of the subtree. In the algorithm, each member node of the subtree.selects the closest key node as its local target node according to the principle of proximal integration. As a result, the isolated subtree is splitted into multiple smaller subtrees. The communication burden of the key nodes is balanced.