Research On Data Forwarding In Opportunistic Networks Based On The Perception Of Encounter Time

As the proliferation of portable devices,e.g.mobile phones,tablets,laptops,the devices sometimes need to connect to the others by using their short-range wireless communication capability without the aid of network infrastructures.At this time,these devices will constitute an opportunistic network,in which they can connect to each other within meeting duration.In opportunistic network,frequent link disconnection and changeable network topology will lead to unstable end-to-end path,which results in low delivery efficiency and high delay in data forwarding.Nowadays,the rapidly developing mobile Internet and forthcoming 5G technology have brought new opportunities and challenges to the research on opportunistic networks.Particularly,the data forwarding schemes should be carefully designed by considering the social behaviors in opportunistic networks.Basically,the nodes,e.g.,devices,individuals,with close relationship constitute a group named community,which reflects the social relations among nodes.Most of the previous community detection algorithms ignore the variation of social relations.As the community size will increase monotonically with time,the existing data forwarding schemes are likely to be inefficient over a long time.To tackle the aforementioned difficulties,a distributed community detection algorithm that can be implemented on each mobile device is proposed in this work,where the community members are dynamically updated and appropriate relay nodes are selected according to their utility.The main work of the thesis is summarized as follows:1)The log files of campus wireless network users are analyzed to extract data set.A method for calculating the random distance in rectangular region is proposed to calculate the probability of direct communication between nodes.Some classic datasets extracted from conference or campus environment,i.e.,Infocom05,Infocom06,Cambridge,are studied and the connection time characteristics of nodes are obtained.2)Based on the temporal characteristics of the dataset,the social relationship between nodes is quantified by accumulating the connection time.The distributed connection time based dynamic community detection algorithm(CTDC)is proposed in which the expired member removal mechanism is considered.The algorithm checks the relationship of community members and clears outdated members at the end of the time slice,which can not only prevent the problem of oversized communities,but also avoid the waste of network resources when messages are passed among the outdated members.Then the two-hop and three-hop reachable node model is introduced based on the community structure.The CTDC-2hop and CTDC-3hop routing strategies are proposed to verify the effectiveness of the community detection algorithm.Finally,the parameter selection in CTDC algorithm is studied in detail.3)The Social Relationship and Dynamic Community-based data forwarding algorithm(SRDC)is proposed.The indicators such as connection stability,connection frequency,social and community similarity are defined to quantify direct and indirect social relationships between nodes.Then these indicators are used to construct utility functions according to different forwarding situations.The relay selection scheme is determined according to the relationship between the current node,encounter node and the destination of the message.The active nodes are selected for communication between different communities such that messages can be delivered to the target as quickly as possible.The simulation with several data sets and ONE simulator is conducted to evaluate the performances of the proposed scheme and the benchmark schemes,i.e.,Epidemic,DRAFT,Simbet and Bubble,in terms of the delivery ratio,the delivery latency,the network overhead and the average hops.