Cognitive radios (CRs) have been proposed as a promising solution for improving spectrum utilization via opportunistic spectrum sharing. In a CR network environment, primary (licensed) users have priority over secondary (unlicensed) users when accessing the wireless channel. Thus, if a malicious secondary user exploits this spectrum access etiquette by mimicking the spectral characteristics of a primary user, it can gain priority access to a wireless channel over other secondary users. This scenario is referred to in the literature as primary user emulation (PUE). This dissertation first covers three approaches for detecting primary user emulation attacks in cognitive radio networks, which can be classified in two categories. The first category is based on cyclostationary features, which employs a cyclostationary calculation to represent the modulation features of the user signals. The calculation results are then fed into an artificial neural network for classification. The second category is based on video processing method of action recognition in frequency domain, which includes two approaches. Both of them analyze the FFT sequences of wireless transmissions operating across a cognitive radio network environment, as well as classify their actions in the frequency domain. The first approach employs a covariance descriptor of motion-related features in the frequency domain, which is then fed into an artificial neural network for classification. The second approach is built upon the first approach, but employs a relational database system to record the motion-related feature vectors of primary users on this frequency band. When a certain transmission does not have a match record in the database, a covariance descriptor will be calculated and fed into an artificial neural network for classification. This dissertation is completed by a novel PUE detection approach which employs a distributed sensor network, where each sensor node works as an independent PUE detector. The emphasis of this work is how these nodes collaborate to obtain the final detection results for the whole network. All these proposed approaches have been validated via computer simulations as well as by experimental hardware implementations using the Universal Software Radio Peripheral (USRP) software-defined radio (SDR) platform.

Thesis( M. A. (Computer Science)) -- University of Limpopo , 2021 / In Software Defined Cognitive Radio Network (SDCRN), network security is a significant issue. This issue arises when Software Defined Network (SDN) architecture integrates with the Cognitive Radio Network (CRN) technology. SDN is designed to improve network resource management, while CRN technology is meant at improving spectrum management. These technologies are vulnerable to several malicious attacks. These attacks include Distributed Denial of Service (DDoS) and Primary User Emulation (PUE). Both the DDoS and PUE can be disrupt services in the SDCRN. To curb these attacks, schemes which hardens the security of SDCRN need to be designed. Thus, in this study we propose a security mechanism called Extreme_Controller_Mechanism (XCM) that reduce the effects of DDoS and PUE. The proposed XCM scheme was designed and evaluated in three simulation environment, the OMNeT++, Octave, and MATLAB simulators. The SDCRN data set was generated using the Neural Network back propagation algorithms. The data set was then used in Matlab to evaluate the effectiveness of the prosed XCM scheme. XCM proved to be effective and efficient at detection and prevention of DDoS and PUE attacks in SDCRN. In terms of memory and processor utilisation, XCM proved to the best when compared to other schemes such as the Advanced Support Vector Machine (ASVM) and deep learning convolution network (CDLN). But in terms of detection time, the ASVM was found to be the best performing scheme. Regarding our test for detection rate, false positive and false negative, the XCM, ASVM and CDLM performed the same. The results of the XCM were therefore the best and superior to the ASVM and CDLM. This can be attributed to the fact that the XCM scheme is optimised for DDoS and PUE attacks. We can therefore conclude that our XCM scheme is the best performing scheme compared to the ASVM and CDLN schemes.

The tremendous success of various wireless applications operating in unlicensed bands has resulted in the overcrowding of those bands. Cognitive radio (CR) is a new technology that enables an unlicensed user to coexist with incumbent users in licensed spectrum bands without inducing interference to incumbent communications. This technology can significantly alleviate the spectrum shortage problem and improve the efficiency of spectrum utilization. Networks consisting of CR nodes (i.e., CR networks)---often called dynamic spectrum access networks or NeXt Generation (XG) communication networks---are envisioned to provide high bandwidth to mobile users via heterogeneous wireless architectures and dynamic spectrum access techniques. In recent years, the operational aspects of CR networks have attracted great research interest. However, research on the security aspects of CR networks has been very limited. In this thesis, we discuss security issues that pose a serious threat to CR networks. Specifically, we focus on three potential attacks that can be launched at the physical or MAC layer of a CR network: primary user emulation (PUE) attack, spectrum sensing data falsification (SSDF) attack, and control channel jamming (CCJ) attack. These attacks can wreak havoc to the normal operation of CR networks. After identifying and analyzing the attacks, we discuss countermeasures. For PUE attacks, we propose a transmitter verification scheme for attack detection. The scheme utilizes the location information of transmitters together with their signal characteristics to verify licensed users and detect PUE attackers. For both SSDF attacks and CCJ attacks, we seek countermeasures for attack mitigation. In particular, we propose Weighted Sequential Probability Ratio Test (WSPRT) as a data fusion technique that is robust against SSDF attacks, and introduce a multiple-rendezvous cognitive MAC (MRCMAC) protocol that is robust against CCJ attacks. Using security analysis and extensive numerical results, we show that the proposed schemes can effectively counter the aforementioned attacks in CR networks. / Ph. D.