Research On Distributed Storage Mechanism In Opportunistic Networks

Opportunistic networks make use of the mobility of nodes to establish links whose connections cannot be guaranteed. That is the basic communication method in the environment which is not all connected. Characters such as poor controllability and resource limitation have greatly restricted the performance of opportunistic networks. Information in opportunistic networks needs to be stored moved and transmitted by nodes. Nodes have to face a huge storage pressure if the connection situation of the network is poor or there are too many data requests at the same time. According to recent studies, we can make effort to improve the performance of data sharing and reduce the storage pressure of nodes by setting up some static infrastructures and building up clustered opportunistic network architectures. So how to build up the distributed storage relationship between infrastructures and nodes under this architecture has become an important research point.At the beginning this paper described the basic characteristics of opportunistic networks and introduces some kinds of popular distributed storage technology. Then we have analyzed current situations of research on architectures and distributed storage mechanism in opportunistic networks. The third chapter of the paper put forward the initial clustered architecture for opportunistic networks and has given the data information list and behavior models for super node and normal nodes within one cluster.Chapter4has designed the data management behaviors including data replica adding and removing algorithm and configuration process in the cluster. Simulation result shows that this design can improve the performance of the network storage. In chapter5, dynamic sub-regions are classified in each cluster according to the features of super node and the movement regularity of normal nodes. After that the paper has described a kind of strategy used for dissemination of data updating messages in one sub-region and the whole cluster. The author set up a network overhead model and got the conclusion that the strategy established in this paper could get a lower and more stable overhead in some conditions compared to the flood spread method as well as the method that only relies on the super node.