Research On Data Dilivery And Congestion Control In Mobile Opportunistic Networks

In recent years,with the rapid proliferation of smart portable devices and endless development of Io T(Internet of Things),data transmissio ns grow up in circumstance of intermittent connectivity;traditionalone hop communication cannot meet the requirement between “device to device”.Distributed,easy to implement and expandable wireless multi-hop communication networking,as the main communication technology between devices and people and their interconnectionin the future "smart city" scenario.In the intermittent communication link,it is hard to set up stable communication between nodes,these nodes usually adopt opportunistic contact to transmit data,the data delivery in opportunistic network s are referred to as the store-carry-forward mechanism.The merit of this kind of data delivery mechanism is not relying on network topology,in facing the scenarios of the intermittent communication link or unstable data transmission,opportunistic data forward can meet the requirement of data transmission in various extreme conditions.Comparing with traditional Ad Hoc network,opportunistic network s can more satisfy the requirements of data refer in mobile scenario.Opportunistic data transmission route protocol is the foundation technology of the network,how to design a high-efficient and stable route transmission scheme is still the core of wireless multi-hop network.Since more smart devices are carried by humans,considering social characters of humans as a factor to design route protocol has been studied extensively in data opportunistic transmission.Due to the differences of node mobility,transmission ability and transmission type,lead to nodes have different message dissemination ability,which is difficult for traditionalmethods to capture the process of message dissemination.This dissertation adopt epidemic model to simulate message dissemination in opportunistic mobile network s,theoretically analyze different effects on message dissemination in different scenario.To provide foundation for different kinds of data opportunistic deliver in theory.In addition,combining super node with large cache space and high transmission speed in mobile opportunistic networks and establish model function to analyze the effects of route protocol in opportunistic mobile networks.In the following,a copy-control mechanism is analyzed and provides epidemic model based on improved Bloom Filter,introducing Bloom Filter to efficiently decrease number of copies to decrease network overhead.Be different with traditional widely adopted multi-copy mechanism,this message refer scheme could increase network overhead.This dissertation combination the human social characters in social network,mining the social factors of mobile nodes to design distributed data deliver route strategy to decrease network overhead.From data dissemination incentive scheme with selfish nodes in mobile opportunistic netwroks(Non-full cooperation between nodes),data deliver route protocol with no selfish nodes in mobile opportunistic netwroks(full cooperation between nodes),data offload transmission control scheme three aspects to study.The main research contents are as follows:Firstly,Mobile nodes usually contain multi-interrelations information,messages referred between different nodes are result from various information.To combination different treats of human in social network,geography-similarity,social-similarity and interest-similarity,three aspects to choose optimal refer node in order to establish high stable and high efficient data transmission route protocol.To prevent network overhead due to over number of copies,a copy-control scheme is proposed to efficiently decrease number of copies.when selfish nodes exist in the mobile opportunistic networks.This thesis proposes a virtual currency incentive scheme based on Rubinstein-Stahl bargain game to stimulate selfish nodes forward data cooperative.Combinate individual selfishness and social selfishness of nodes in consideration,in addtion with residual resouse of nodes to estblish game price function.A dynamic time control scheme is adopted to alter different factor in different time slots,and analyses of scheme and control is provided.simulation expriments based on real trace indicates that the proposed GSI-IR route outperforms others opportunistic routes in dilivery ratio and average delay.Secondly,due to the cybotactic of human mobility,human in community m ore intend to contact.It is found that data delivery in community structure,both community detection and optimal relay node choose are vital to performance of route protocol.To partition nodes in network based on community detection in complex network.A semi-supervised nonnegative matrix factorization algorithm is proposed to detect community structure in opportunistic mobile network s.Dynamic-centrality is provided to choose optimal relay nodes,which indicate importance of contact frequency to deliver data in community.Relative centrality of nodes is also provided to choose optimal relay nodes to deliver databetween communities.Finally,in combination with mobile edge computing(MEC)and opportunistic mobile networks to analyze data offload and pocket scheduling control.Firstly,time delay and energy consumption of MEC system are analyzed.A pocket transmission to cloud server in low priority queue probabilistic jump to high priority queue scheme is proposed.This scheme could efficiently decrease th e time delay and energy consumption of MEC system.In the mean time,a mobile device decision algorithm is proposed to decide whether or not the tasks compute in local devices or cloud server.To efficiently solve overhead of MEC system,a improved krill h erd metaheuristic algorithm is also proposed to sole the optimal function.The simulation results show the improved krill herd algorithm has more global searcher ability and converge faster than original krill herd algorithm.Based on the analyses of simulations,it proved the proposed PBP scheme efficiently reduce both time delay and energy consumption in MEC system,it also proved validity and feasibility of improved krill herd algorithm in solving multi-objecive function.