Abstract
This paper studies cooperative spectrum sensing based on crowdsourcing in cognitive radio networks. Since intelligent mobile users such as smartphones and tablets can sense the wireless spectrum, channel sensing tasks can be assigned to these mobile users. This is referred to as the crowdsourcing method. However, there may be some malicious mobile users that send false sensing reports deliberately, for their own purposes. False sensing reports will influence decisions about channel state. Therefore, it is necessary to classify mobile users in order to distinguish malicious users. According to the sensing reports, mobile users should not just be divided into two classes (honest and malicious). There are two reasons for this: on the one hand, honest users in different positions may have different sensing outcomes, as shadowing, multi-path fading, and other issues may influence the sensing results; on the other hand, there may be more than one type of malicious users, acting differently in the network. Therefore, it is necessary to classify mobile users into more than two classes. Due to the lack of prior information of the number of user classes, this paper casts the problem of mobile user classification as a dynamic clustering problem that is NP-hard. The paper uses the interdistance-to-intradistance ratio of clusters as the fitness function, and aims to maximize the fitness function. To cast this optimization problem, this paper proposes a distributed algorithm for user classification in order to obtain bounded close-to-optimal solutions, and analyzes the approximation ratio of the proposed algorithm. Simulations show the distributed algorithm achieves higher performance than other algorithms.
Highlights
According to a Cisco report [1], wireless traffic has significantly increased over the last few years
Due to the lack of prior information on the number of user classes, this paper casts the problem of mobile user classification as a dynamic clustering problem that is NP-hard
The paper uses the interdistance-to-intradistance ratio of clusters as the fitness function for evaluating the clustering effect, and aims to maximize the fitness function. To cast this optimization problem, this paper proposes a distributed algorithm for user classification in order to obtain bounded close-to-optimal solutions, and analyzes the approximation ratio of the proposed algorithm
Summary
According to a Cisco report [1], wireless traffic has significantly increased over the last few years. The authors proposed a clustering strategy for cooperative spectrum sensing, with the clustering considering the differences in underlying hidden Markov models associated with the detection of distinct licensed users. Mobile users are assigned spectrum sensing tasks by the crowdsourcing method in this paper. Based on other mobile users’ sensing reports, each mobile user makes a decision about the channel state independently. Mobile users should be divided into several classes, rather than only two classes (honest and malicious). The paper uses the interdistance-to-intradistance ratio of clusters as the fitness function for evaluating the clustering effect, and aims to maximize the fitness function To cast this optimization problem, this paper proposes a distributed algorithm for user classification in order to obtain bounded close-to-optimal solutions, and analyzes the approximation ratio of the proposed algorithm.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
