Study On Key Technologies Of MANET

The Mobile Ad Hoc Network (MANET) is a temporary multi-hop autonomous network with all nodes having wireless transmitters. Each node in it plays dual roles as usual node and router at same time. As a network without infrastructure, the MANET has wide prospects of application not only in military domains, but also in civil fields. It has now drawn interest of many researchers around the world. This paper mainly focuses on the research of cluster? architecture, service-oriented MANET and the opportunistic network, which are some key issues for the MANET .In the first part of this paper, the topology structure of MANET is discussed. To make cluster more stable and sustainable, the paper proposes a series of cluster algorithms with two cluster headers. Firstly, it initializes the devices in MANET into clusters by means of communications among devices, and all the nodes are classified as member-node or gateway-node in different clusters. Then, by comparing the factors such as the adjacent degree with gateway nodes in a cluster, the adjacent degree with member nodes and the degree of available power, the node with the highest value is selected as the cluster head (CH). It ensures that the cluster head is a stable node with longer duration and higher connectivity, and strengthens the cluster's stability. Furthermore, substitute node (SD) is introduced to improve the stability of cluster. Once CH was unable to work due to low power or movement condition, SD would replace the role of CH to prolong the life cycle of cluster and reduce the cost of cluster maintenance. In the end, the paper presents a strategy to monitor or predict the movement of CH in a cluster, which is easier and more efficient than traditional algorithms.The second part of the paper focuses on the problem of service discovery on MANET. The service discovery is used to find service in network automatically. And its final aim is to realize the auto-management network and liberate human from the tasteless network maintaining work. Thus, a service discovery protocol with simpleness, efficiency, strong flexibility and good expansibility is of a vital technique to put MANET into practice. In order to design such a good protocol, the paper introduces a novel service register and discovery mechanism. Firstly, a new meta-data description model is constructed, which combines the service info, user info and device info, and divides info into static description and dynamic description. By this way, the model can meet the requirement of weak computation of device, the big difference between devices and narrowband of network. And the meta-data description model is also helpful to the service discovery. Secondly, this paper introduces a new service register strategy different from the common UDDI protocol. In the new register strategy, the service provider registers the service info to the cluster header and the service match is processed in the cluster first. Only when the service match failed to reach the cluster, the match process would continue to search in the neighbor clusters by communicating with cluster headers. With this strategy, the service applications can work well in MANET with limits of MANET environment and devices. And it reduces the loads among the clusters during the register or discovery process. The distributed service registry mechanism solves the problem that UDDI can't work in a dynamic and varied topology network. Also it implements a structure of 'local centralized, global distributed'. In addition, the mechanism uses SOAP-over-UDP to overcome the limitations of TCP protocol, and adopts a fault-tolerance service discovery method to make up for the weakness of UDP. The simulation results show that the mechanism and architecture has better practicality.The final part of the paper brought out with the Opportunistic Network. This paper analyzes the strategy of the data transmission from source node to target node in opportunistic network. To make the selection of intermediate nodes more reasonable and fairly, it introduces a message transmitted route strategy based on two-player non-cooperation game. By concluding the status info of nodes and messages, the income statement is established for the game theory with the help of Kalman filter which can predict the future values. And the game theory can also be used for the negotiation of two nodes. By this strategy, the decision of message transmission is much fair because the condition of accepter is taken into consideration in the negotiation. But, during the game process, the attendance node may do selfishness behaviors to gain more profits from the negotiation. And such behaviors are called 'cheating'. How to avoid the cheating behavior is a new exploration. In the paper, the idea of credit-cooperation and repeated games are involved to solve the problem. For a node, every game is considered as a part of repeated game in its lifecycle. If it did a cheating-operation in a game, it would face the punishment with long time. And the profit gained from the cheating-operation would be counteracted mutually from the loss of punishment time. Moreover, the punishment is usually bigger than profits in fact. The method minimizes the probability of cheating in the game theory and makes the game more efficient.