Research On Topology Control Algorithm Based On The Key Nodes Elimination In Ad Hoc Networks

Ad Hoc network is a novel form of network which is self-constructed, self-organized and self-managed without centralized administration or wired infrastructure. Because of its many advantages such as flexible networking, quick deployment and distributed control, Ad Hoc network has wide application in the military and civilian fields. However, limited resources, as well as wireless communications and node mobility make the performance of Ad hoc network closely bound up with its topology. Unreasonable network topology will reduce network capacity, increase packet transmission delay and weaken network robustness when faced by node failures. Hence, topology control is an effective measure of great significance to improve the overall performance of Ad Hoc networks.In accordance with the characteristic of Ad Hoc networks, the key node called partition node for short was defined, which connects two or more areas. The key nodes failure will cause network segmentation. In order to eliminate the partition nodes, a topology control algorithm was proposed. The algorithm has three phases, which are partition node detection, partition node elimination and topological optimization. In the first phase, every node judges whether it is partition node according to arrival relationship collected. In the second phase, the partition nodes choose representation nodes from their each subset. Then these representation nodes are connected in the way of chordal ring connection. Considering the capacity of nodes, there is need to remove some links when the connectivity reaches or exceeds the threshold value because of adding new links. In the third phase, with the use of power estimation, the nodes estimated minimal transmission power when they are receiving or hearing their neighbor nodes. Then, the nodes adjust their transmission power to guarantee good network performance in the running phase.The network simulation software NS2was adopted to evaluate the proposed topology control algorithm. Simulation results demonstrate the algorithm can eliminate effectively the partition nodes and optimize network topology, such as reducing the existing time of partition node without the overhead increasing, guaranteeing good invulnerability and improving the ability of anti-segmentation.