Research On Trust Management Mechanisms For Crowd-Based Cooperative Networks

With the development of Internet technologies and the proliferation of mobile devices,crowd-based cooperative networks have brought significant convenience to people's daily life.In this kind of networks,the users can be the consumers of data and the producers of data in the meanwhile.The properties of crowd-based cooperative networks include the diversity of networking technologies,the different kinds of data and the variant roles of the users.As a great volume of data can be produced in this kind of networks,they are easy to become the target of cyber attacks,especially in the data-driven nowadays.As one of the most efficient auxiliaries of cryptographical tools,trust management mechanisms have been widely adopted here to defend against insider threats.Thus in this thesis,we have conducted several researches on trust management mechanisms on the basis of the properties of crowd-based cooperative networks.Due to the diversity of networking technologies,users can access to the network via different ways.In order to fully utilize different kinds of networks for data exchange,it is necessary to equip the users with the ability of spectrum sensing.However,there can be malicious users to provide false local spectrum sensing data to mislead the fusion center to derive a wrong final result.The malicious users can do this to utilize the idle spectra only by themselves or disrupt the communication of other users.Hence,the first work of this thesis is to design robust and secure spectrum sensing mechanisms based on trust management mechanisms and the detailed scenario is cognitive radio networks.In most existing works,the common control channel(CCC)is assumed to be perfect.However,this assumption may not hold in practice and imperfect CCC makes the existing methods against independent or cooperative data falsification attacks less effective.In this work,we first analyze the impact of an imperfect CCC on the identification of malicious secondary users under independent and cooperative attacks.To better differentiate honest users and malicious users,a reputation threshold is derived for each secondary user.Based on the obtained reputation threshold,we propose a new reputation-based cooperative spectrum sensing method,which is validated to be robust against attacks under imperfect CCC.Due to the different kinds of data produced in crowd-based cooperative networks,the data reliability is an important issue to be addressed.This is because any user can inject false or biased data to the network motivated by financial interests.Thus in the second work of this thesis,we propose a method to compute a reliable reputation value for any given target and the application scenario is the online rating systems for E-commerce.The key point of this work is to detect which ratings are injected by attackers.This method can be adopted by online trading platforms or run by a third party to provide reputation computation as a service.At first,a fine grained two-phase detection method is proposed to detect malicious ratings.After filtering out the ratings detected as malicious,the weights of the remaining ratings are determined by computing the degrees to which the users giving these ratings are interested in a target item.The intuition to do this is that the more interested a user is in this item,the more important the rating given by this user is.Extensive experiments verify that the proposed reliable reputation computation framework is effective to detect different kinds of malicious ratings and determine the interest degrees of users.Apart from the issue of data reliability,the property of producing different kinds data in crowd-based cooperative networks makes it easy to reveal private and sensitive information about the users.From this viewpoint,it is of great significance to design privacy-preserving trust management mechanisms to make crowd-based cooperative networks go further.As a result,the third work of this thesis is to provide two privacy preserving trust management schemes to simultaneously preserve privacy and deal with malicious participants and the application scenario is mobile crowdsensing.In the basic scheme,a novel reputation value updating method is designed based on the deviations of the encrypted sensing data from the final aggregating result.The basic scheme is efficient at the expense of revealing the deviation value of each participant to the reputation manager.To conquer this drawback,we propose an advanced scheme by updating the reputation values utilizing the rank of deviations.Extensive experiments demonstrate that both these two schemes have high cost efficiency and are effective to deal with malicious participants.For the reason that the users of crowd-based cooperative networks can play variant roles at the same time(i.e.the data provider and data consumer),it is easy for everyone to join this kind of networks.As a result,it is necessary to introduced a trusted entity to monitor the behaviors of the users and evaluate data reliability.However,this kind of trusted entities can be easily corrupted or collude with users and thus become the bottleneck of trust management schemes.From this viewpoint,the last work is to design a distributed trust management mechanism on the basis of Blockchain without the need for such a trusted entity.Here,the application scenario is also the mobile crowdsensing.In this work,the edge computing paradigm is introduced and the tasks of processing data and managing trust are loaded via geographically distributed edge nodes.Also,the blockchain are managed by the edge nodes.To improve the efficiency of blockchain management,we utilize the concept of consortium chain where only the edge nodes are allowed to request operations on the chain.To preserve privacy and conquer the high cost of running the SHE scheme in the previous work,the Paillier encryption scheme,which support homomorphic addition,is utilized for aggregating sensing data from different participants.With respect to updating reputation values,we present a different definition of deviation degrees from the previous work and design a novel updating rule based on the newly defined deviation degrees.