Cognitive radio networks (CRNs) are becoming an important part of wireless communications, which can improve the utilization of the limited and scarce spectrum resources. However, the flexibility and the unique characteristics of CRN make it vulnerable to varied types of attack. Moreover, security is a key requirement at every step of its functionality (spectrum sensing, spectrum analysis, spectrum decision and spectrum act). Numerous researches have mainly focused on to provide security at the cognition capability (spectrum sensing, spectrum analysis and spectrum decision) of CRN for detecting the appropriate spectrum holes. However, after obtaining access to the spectrum hole (Spectrum Act), a Cognitive Radio (CR) may behave maliciously to achieve own benefits or for some other reasons. Such maliciousness can severely affect the normal activities of the whole network. Therefore, there is a need to track and record the behaviour during the spectrum access of a CR user, which can encourage the CR users to obey the opportunistic spectrum access policy. In this paper, we construct a trust-based approach for secure spectrum access in CRN. The CR nodes trust value is determined from its past activities, and based on which, it is decided whether the CR node will get access the primary users free spectrum or not.

Cognitive radio (CR) is an emerging technology for efficient utilization of the limited and scarce radio spectrum resources. Spectrum sensing is a key task in CR system for detecting the unused radio spectrum or spectrum holes by the primary users. In this regard, collaborative spectrum sensing (CSS) has shown to as an efficient way to improve such spectrum holes detection in cognitive radio network (CRN). However, this cooperative nature makes it vulnerable to many types of security attacks. In this paper, we focus one of these potential CRN specific security attack called spectrum sensing data falsification attack or Byzantine attack. In which the malicious internal member of the network, reports false sensing results with a aim to increase the sensing errors. In order to ensure the security requirements of CRN, we proposed a novel trust management mechanism that evaluates the trustworthiness of each node participating in the CSS scheme called sensing reputation (SR). To reflect the complexity, the SR calculation involves multiple decision factors like history based trust factor, active factor, incentive factor and consistency factor. Based on this SR value, the suspicious users are identified and the malicious users are filtered out from the decision making process of the CSS scheme. We further introduce the concept of sensing reputation chain to record and track the future behavior of the identified suspicious users. Theoretical analysis and simulation results demonstrate the effectiveness of our proposed malicious user detection technique with a higher accuracy and lower false alarm rate.

Wireless applications have been growing rapidly in the past years, leading to the problem of spectrum scarcity. Cognitive radio network (CRN) is an emerging technology which can provide a promising solution to resolve this spectrum scarcity problem in wireless communication. However, the security concerns in cognitive radio (CR) is an unavoidable issue, since it introduces significant new classes of threats due to its unique characteristics and functioning techniques and also has received more attention by the researchers recently. In this paper, we provide a brief overview of CRN technology and analyse its associated security vulnerabilities and discuss a number of available solutions for countering those threats. We identify the security threats based on the targeted open system interconnection (OSI) layer and classified those into traditional wireless threats and CRN specific threats. The paper will also focus on Byzantine attack or spectrum sensing data falsification (SSDF) attack, which is a common CRN specific threat and reviewed the distinguished proposed techniques by research community to detect and isolation of such attack. Finally, a road-map for future research is addressed.

The objective of this paper is to predict software reliability using non-parametric neural network of computational intelligence (CI). The study uses data sets containing failure history such as number of failures, failure time interval etc. In this paper, we explore the applicability of feed-forward neural network with back-propagation training as a reliability growth model for software reliability prediction. The prediction result is compared with that of traditional parametric software reliability growth models. The results described in the proposed model exhibits an accurate and consistent behavior in reliability prediction. The experimental results demonstrate that the proposed model provides a significant difference respect to accuracy and consistency.