Study On Clustering Algorithms In Ad Hoc Networks

The Ad Hoc network doesn't rely on fixed based stations or a wired backbone infrastructure. The network is composed of multiple mobile nodes without a fixed router. Ad Hoc network is without any basis for setting and any of the central control nodes. The statuses of nodes are equal and have a packet forwarding capability. The communication between nodes is through the node multi-hop transfer to achieve. Ad Hoc networks have distributed control, network hoc, dynamic topology, limited bandwitdh and energy constrained nodes, multi-hop routing and security-constrained and so on. With fast networking and the advantages without the need to set the default base, Ad Hoc network have been widely used in military, sensor networks, emergency disaster relief, and personal communications and so on. Minimum ID clustering algorithm maintains the smaller cost. The drawback is that the cluster head nodes well consume more energy, and does mot take into account load balancing factor. The central idea of the maximum node degree is to minimize the munber of clusters. Low mobility algorithms can significantly reduce the update frequency of the cluster head, so that stable cluster structure. Because the mode moving speed is random, the mode will cause frequent changes in the weight calculation of cost inctrases, without taking into account the network load balancing. Portfolio weighting algorithm is taking into account additional network elemengts. Each node participating in the elections when the algorithm parameters can refer to more, but the load balancing problems remains unresolved.This paper introduces the Ad Hoc network node structure. In the planar structure, the nodes are completely equal with a faster networking. There is no advantage of the bottleneck node , but the network scalability is not srtong. In the cluster structure, if we can cluster reasongable, cluster membership function would be relatively simple and do not need to maintain particularly complex routing information. It can greatly reduces the network routing control the amount of information and make Ad Hoc clustering network has very good scalability. As the cluster head can be elected by the cluster head algorithm, the entire network survivability is also very strong. The disadvantage of the structure of cluster is that the cluster head node may become the bottleneck of network expansion because of constraintion. This paper has discussed the clustering algorithm and the advantages and disadvantages of various algorithms are analyzed and discussed.The basic objective of Ad Hoc Network Routing Algorithm is to transmission of data correctly and effectively. The routing algorithms of traditional wired networks can not be used Ad Hoc networks, so people have a large number of developments for the Ad Hoc netword routing algotithm. This paper discusses tow types of routing algorithms. The inplementation of table-driven routing algorithm is to maintain the network of nodes from its own node to all other nodes of the latest routing information. All nodes is to maintain consistent routing information. Each node need to safeguard one or more of the routing table, which make modes and networks big pressure. When th enetwork topology changes, update information is spread throughout the network to maintain a unified message. DSDV, GSR and WRP are the representative of the table-driven routing algorithm. Demand-driven routing algorethm only build routing when the source node needs to establish routing. Compared with table-driven routing, demand-driven routing nodes do not need to maintain multiple routing table and can greatly reduce network overhead. DSR, AODV and TORA are the representative of the demand-driven routing algorethm.Based on the above results of exposition, we found that control of the bottleneck node clustering algorithm can be by strengthening the network load balancing and strengthening the cluster head node to compete to achieve fairness. On the basis of the combined weighting algorithm, we have introduced and the state machine timers. The role of the timer is to limit the mode as thecluster head of time, to ensure that the cluster head node will not run out of the high consumption of energy. The role of state machine indicatings election status of the cluster node. The use of timers and state machine is that the cluster head node can fulfill responsibilities in the timer. When the timer reaches the door threshold, the mode will no longer serve as cluster head the task, according to rules of the cluster head task shifting to other nodes, and no longer participate in the election of cluster head, while the node state machine to change. When the network node-state device has a number of state machine for the election after the state, the network is to algorithms. Each cluster head can make a fair participation in cluster head election. It can ensure the stability of the structure of the network and reduce the cost of clustering algotithms. The results show that the inproved algorithm is better in HFI. In traditional routing protocols, it use table-dirven routing algorithm in one cluster and use on-demand routing algorithm between clusters. Cluster head node need to install two kinds of routing algorithms while maintaining cluster head and the routing tables. The cost of network is relatively large. When the route is broken, the node need to buld process. We suggest using an on-demand routing algotithm and use multi-path transmission.Local repair can guarantee that the data during transmission do not particularly worried about eht routing fault, for the network's service quality and rtansmission efficiency to provide a guarantee. The results show that our strategy in the data arrival rate and delay is to perform better. It is particularly inportant in Ad Hoc networks.