Research On Key Problems Of Information Communication In Social Internet Of Things

Recently,the Social Internet of Things(SIoT)has been a new research area of the Internet of Things.It is a convergence of Social Networks and Internet of Things.Observing that people tend to communicate with friends rather than strangers,the SIoT can be established according to social connections of devices' owners.Information is exchanged throughout the SIoT.Therefore,it is indispensable to investigate problems of information communication in the SIoT.Based on an IoT application instance of smart home in a community,we identify three key problems that require to be solved well.These problems reflect three aspects of information communication in SIoT,respectively,i.e.,local information exchanges,global information spreading and the utilization of information transmission resources.In this thesis,we focuses on dealing with these problems,and the major contributions as well as novelty are summarized as followings:From the aspect of local information exchange,we investigate the distributed optimization problem.Targeting at this problem,we propose a method for distributed information processing.By designing rules of information exchange between adjacent nodes,this method can accomplish the distributed optimization with heterogeneous information sensed by each node.The method is comprised of two algorithms.First,given the practical consideration that nodes of heterogeneous sensing functionalities cannot sense full dimensional records,we design a distributed record completion algorithm.In this algorithm,each node only needs to exchange data with its neighbors in social networks.Thus it can quickly complete the data with all the other nodes' data,and the elimination of each node's sensing error as well as the global information consensus can be achieved,which creates a foundation for the following distributed optimization solution.Second,we propose a distributed dual averaging algorithm.The algorithm targets at deal with the distributed optimization problem.By the information exchange between adjacent nodes,this algorithm can achieve consistent optimization results for all the nodes.Moreover,this algorithm can solve the practical non-smooth optimization problem,and meanwhile,can achieve a more efficient computation with fluctuations of input data.From the aspect of global information spreading,we investigate how to identify powerful information initial spreaders.This investigation includes two parts.First,from the spectral radius perspective,we establish the relationship between the information spreading size and the spectral radius of the network,based on which,we propose an algorithm for identifying the initiators.In this investigation,in order to incorporate the conformity effect on the individual's information adoption in Sociology,we introduce the epidemic model to describe information spreading.Moreover,with regard to the determination of weights in the weighted graph,we employ the expectation maximization algorithm to extract useful parameters that can truly imply information spreading ability from the raw reposting data,and use such parameters to define the weight of each link.Comparing with existing approaches,the initiators identified by our algorithm can lead to faster spreading speed.Second,aiming at the case that the SIoT may exist multiple groups of tight-knit nodes,and taking advantage of the nature that information can spread quickly in such groups,we propose an approach of seeking initiators from the dense group perspective.This algorithm only needs to be executed within the dense groups which includes fewer nodes than the entire network.Therefore,comparing with the centralized method,our approach is more efficient.From the aspect of the utilization of information transmission resources,we investigate the long-term transmission resource occupancy problem.The human behavior is one of key considerations in the SIoT.Due to the limited communication resources,devices in the SIoT are required to queue for using the resources.However,there may exist some greedy users that occupancy spectrum resources for longer time than the normal ones to transmit more information and gain more revenue.To address this problem,first,we investigate the impact of greedy users on the system performance with quantitative studies.Then,we provide a quick detection strategy to identify such behavior.Finally,we propose a protection approach for normal users by adjusting queueing rules of accessing spectrum.As the abnormal behavior may be malicious in the SIoT,comparing with the existing protection strategies which are based on economic approaches,our proposed strategy can be more effective in mitigating the influence of malicious users.From the mathematical viewpoint,we address the challenge of analyzing multidimensional queueing system with the blocked customers delayed strategy.Our proposed principles for constructing the state-transition diagram and analytical methods for the performance evaluation of queueing system can serve as universal techniques for analyzing such type of queueing problem.With increasingly close connections between social networks and the Internet of Things,we believe that the future SIoT will have a wide range of applications,along with a number of problems require to be addressed.Therefore,in the end of this thesis,we provide our future research outlooks.