The task of detecting and separating multiple radio-frequency signals in a wideband scenario has attracted much interest recently, especially from the cognitive radio community. Many successful approaches in this field have been based on machine learning and computer vision methods using the wideband signal spectrogram as an input feature. YOLO and R-CNN are deep learning-based object detection algorithms that have been used to obtain state-of-the-art results on several computer vision benchmark tests. In this work, YOLOv2 and Faster R-CNN are implemented, trained and tested, to solve the signal separation task. Previous signal separation research does not consider representations other than the spectrogram. Here, specific focus is placed on investigating different time-frequency representations based on the short-time Fourier transform. Results are presented in terms of traditional object detection metrics, with Faster R-CNN and YOLOv2 achieving mean average precision scores of up to 89.3% and 88.8% respectively. / Dissertation (MEng (Computer Engineering))--University of Pretoria, 2017. / Saab Grintek Defence / University of Pretoria / Electrical, Electronic and Computer Engineering / MEng (Computer Engineering) / Unrestricted

Modern classification of organisms is performed on molecular data. These methods rely on multiple alignment of sequences of characters which make them computationally demanding. Only small parts of genomes can be compared in reasonable time. In this paper, the novel algorithm based on conversion of the whole genome sequences to cumulative phase signals is presented. Dyadic wavelet transform is used for lossy compression of signals by redundant frequency bands elimination. Signal classification is then performed as a cluster analysis using Euclidian metrics where multiple alignment is replaced by dynamic time warping.

Dynamic Cellular Cognitive System (DCCS) serves as a cognitive network for white space devices in TV white space. It is also designed to provide quality communications for first responders in area with damaged wireless communication infrastructure. In DCCS network, diverse types of communication devices interoperate, communicate, and cooperate with high spectrum efficiency in a Dynamic Spectrum Access (DSA) scenario. DCCS can expand to a broad geographical distribution via linking to existing infrastructure. DCCS can quickly form a network to accommodate a diverse set of devices in natural disaster areas. It can also recover the infrastructure in a blind spot, for example, a subway or mountain area. Its portability and low cost make it feasible for commercial applications. This dissertation starts with an overview of DCCS network. DCCS defines a cognitive radio network and a set of protocols that each cognitive radio node inside the network must adopt to function as a user within the group. Multiple secondary users cooperate based on a fair and efficient scheme without losing the flexibility and self adaptation features. The basic unit of DCCS is a cell. A set of protocols and algorithms are defined to meet the communication requirement for intra-cell communications. DCCS includes multiple layers and multiple protocols. This dissertation gives a comprehensive description and analysis of building a DCCS network. It covers the network architecture, physical and Medium Access Control (MAC) layers for data and command transmission, spectrum management in DSA scenario, signal classification and synchronization and describes a working prototype of DCCS. Two key technologies of intra-cell communication are spectrum management and Universal Classification and Synchronization (UCS). A channel allocation algorithm based on calculating the throughput of an available is designed and the performance is analyzed. UCS is conceived as a self-contained system which can detect, classify, and synchronize with a received signal and extract all parameters needed for physical layer demodulation. It enables the accommodation of non-cognitive devices and improves communication quality by allowing a cognitive receiver to track physical layer changes at the transmitter. Inter-cell communications are the backhaul connections of DCCS. This dissertation discusses two approaches to obtaining spectrum for inter-cell communications. A temporary leasing approach focuses on the policy aspects, and the other approach is based on using OFDMA to combine separate narrowband channels into a wideband channel that can meet the inter-cell communications throughput requirements. A prototype of DCCS implemented on GNU radio and USRP platform is included in the dissertation. It serves as the proof of concept of DCCS. / Ph. D.

The Universal Classifier Synchronizer (UCS) is a Cognitive Radio system/sensor that can detect, classify, and extract the relevant parameters from a received signal to establish physical layer communications using the received signal's profile. The current implementation is able to identify signals including AM, FM, MPSK, QAM, MFSK, and OFDM. The system is constructed to run on a Universal Software Radio Peripheral (USRP) with the GNU Radio software toolkit and also runs on an Anritsu™ signal analyzer. In both prototypes, the UCS system runs on a host computer's General Purpose Processor (GPP) and is constructed in Matlab™. The aim is to then create a portable and standalone version of the UCS system as an intermediate step towards building a future commercial implementation. This application and particular implementation aims to run on a Lyrtech SFF SDR platform and uses its FPGA and DSP modules for implementation. This platform is one of the more advanced SDR platforms available, and the aim is to develop parts of the UCS system to run on this platform. The aim is to eventually develop the complete UCS cognitive radio system on the Lyrtech SFF SDR platform that can act as a standalone portable cognitive radio system. The modules created and implanted/implemented on the SDR hardware are the Bandwidth Estimation, and Symbol Timing & Coarse Classification modules. This is the system decision path towards classification, synchronization, and demodulation of digital phase modulated signals (QAM and MPSK signal types) and also analog signals. The Digital Receiver Module (DRM) is implemented on the FPGA and takes care of all the digital down conversions, mixing, decimation, and low pass filtering. The FPGA is connected to the DSP module via a bus subsystem where the DSP receives real-time base-band complex IQ samples for further signal processing. The main UCS algorithm runs on the platform's DSP and is compiled from executable embedded C-code. Therefore, this system can then be implemented on virtually any setup that has an RF front end, digital receiver module, and processing module that will execute floating and fixed point C-code with minor changes. / Master of Science

Wireless sensor networks and data fusion has received increasing attention in recent years, due to the ever increasing computational power, battery and wireless technology, and proliferation of sensor modalities. Notably, the application of acoustic sensors and arrays of sensors has expanded to encompass surveillance, teleconferencing, and sound source localization in adverse environments. The ability to passively locate and track acoustic sources, be they gunfire, animals, or geological events, is crucial to a wide range of applications. The challenge addressed herein is how to best utilize the massive amount of data collected from spatially distributed sensors. Localization in two acoustic propagation scenarios is addressed: the free-field assumption and the general case. In both cases, it is found that performance is highly dependent on the array-source geometry which in turn drives the design of localization strategies. First, the general surveillance problem including signal detection, classification, data association, localization and tracking is studied. Signal detectors are designed with a focus on robustness and capacity for real time implementation. Specifics of the data association problem relevant to acoustic measurements are addressed. Assuming free-field propagation, a localization algorithm is developed to harness some of the vast potential and robust nature of a sensor networks. In addition, a prototypical sensor network has been constructed to accompany the theoretical development, address real world situations, and demonstrate applicability. Experimental results obtained confirm the practicality of theoretical models, support numerical results, and illustrate the effectiveness of the proposed strategies and the system as a whole. In many situations of interest, obstacles to wave propagation such as terrain or buildings exist that provide unique challenges to localization. These obstacles introduce multiple paths, diffraction, and scattering into the propagation. The second part of this dissertation investigates localization in the general propagation scenario of a multi-wave, semi-reverberant environment characteristic of urban areas. Matched field processing is introduced as a feasible method and found to offer superior performance and flexibility over time reversal techniques. The effects of uncertainty in model parameters are studied in an urban setting. Multiarray processing methods are developed and strategies to mitigate the effects of model mismatch are established. / Ph. D.

A large body of work has been produced in the area of productivity enhancers for the design of both Software-Defined Radio and Field Programmable Gate Arrays systems. These tool are created with the goal of aiding the user in the process of instantiating a design. They do not address, however, a specific use-case in which the user does not know or care about what the design of his system is. In this work, analysis-based design is presented and applied to FPGA-based SDRs. The RapidRadio framework abstracts away much of the knowledge required for analyzing an unknown signal and building an FPGA-based receiver. Resource utilization is traded-off for reduced implementation time and increased exibility. Automatic modulation classification is done with blind parameter estimation. Unlike other contemporary work, no a priori knowledge about the signal being classified is assumed. This leads to the development of a system that does not depend on perfect synchronization to classify the signal. A new quasi-generic synchronization architecture that allows the synchronization of multiple modulations schemes is presented. The result of the modulation classification is used to automatically create an FPGA-based radio receiver. / Ph. D.

For successful identification of structural defects in plates and pipes, it is essential to understand structural wave propagation characteristics such as dispersion relations. Analytical approaches to identify the dispersion relations of homogeneous, simple plates and circular pipes have been investigated by many researchers. However, for plates or pipes with irregular cross-sectional configurations or multi-layered composite structures, it is almost impossible to obtain the analytical dispersion relations and associated mode shapes. In addition, full numerical modeling approaches such as finite element (FE) methods are not economically feasible for high (e.g., ultrasonic) frequency analyses where an extremely large number of discretized meshes are required, resulting in significantly expensive computation. In order to address these limitations, Hybrid Analytical/Finite Element Methods (HAFEMs) are developed to model composite plates and pipes in a computationally-efficient manner. When a pipe system is used to transport a fluid, the dispersion curves obtained from a “hollow” pipe model can mislead non-destructive evaluation (NDE) results of the pipe system. In this study, the HAFEM procedure with solid elements is extended by developing fluid elements and solid-fluid boundary conditions, resulting in the dispersion curves of fluid-filled pipes. In addition, a HAFEM-based acoustic transfer function approach is suggested to consider a long pipe system assembled with multiple pipe sections with different cross-sections. For the validation of the proposed methods, experimental and full FE modeling results are compared to the results obtained from the HAFEM models. In order to detect structural defect locations in shell structures from defect-induced, subtle wave reflection signals and eliminate direct-excitation-induced and boundary-reflected, relatively-strong wave signals, a time-frequency MUSIC algorithm is applied to ultrasonic wave data measured by using an array of piezoelectric transducers. A normalized, structurally-damped, cylindrical 2-D steering vector is proposed to increase the spatial resolution of time-frequency MUSIC power results. A cross-shaped array is selected over a circular or linear array to further improve the spatial resolution and to avoid the mirrored virtual image effects of a linear array. Here, it is experimentally demonstrated that the proposed time-frequency MUSIC beamforming procedure can be used to identify structural defect locations on an aluminum plate by distinguishing the defect-induced waves from both the excitation-generated and boundary-reflected waves.

Through-the-wall Imaging (TWI) is crucial for various applications such as law enforcement, rescue missions and defense. TWI methods aim to provide detailed information of spaces that cannot be seen directly. Current state-of-the-art TWI systems utilise ultra-wideband (UWB) signals to simultaneously achieve wall penetration and high resolution. These TWI systems transmit signals and mathematically back-project the reflected signals received to image the scenario of interest. However, these systems are diffraction-limited and encounter problems due to multipath signals in the presence of multiple scatterers. Time reversal (TR) methods have become popular for remote sensing because they can take advantage of multipath signals to achieve superresolution (resolution that beats the diffraction limit). The Decomposition Of the Time-Reversal Operator (DORT in its French acronym) and MUltiple SIgnal Classification (MUSIC) methods are both TR techniques which involve taking the Singular Value Decomposition (SVD) of the Multistatic Data Matrix (MDM) which contains the signals received from the target(s) to be located. The DORT and MUSIC imaging methods have generated a lot of interests due to their robustness and ability to locate multiple targets. However these TR-based methods encounter problems when the targets are behind an obstruction, particularly when the properties of the obstruction is unknown as is often the case in TWI applications. This dissertation introduces a novel total sub-MDM algorithm that uses the highly acclaimed MUSIC method to image targets hidden behind an obstruction and achieve superresolution. The algorithm utilises spatio-temporal windows to divide the full-MDM into sub-MDMs. The summation of all images obtained from each sub-MDM give a clearer image of a scenario than we can obtain using the full-MDM. Furthermore, we propose a total sub-differential MDM algorithm that uses the MUSIC method to obtain images of moving targets that are hiddenbehind an obstructing material.

Vehicle acoustic signals have long been considered as unwanted traffic noise. In this research acoustic signals generated by each vehicle will be used to detect its presence and classify its type. Circular arrays of microphones were designed and built to detect desired signals and suppress unwanted ones. Circular arrays with multiple rings have an interesting and important property that is constant sidelobe levels. A modified genetic algorithm that can work directly with real numbers is used in the circular array design. It offers more effective ways to solve numerical problems than a standard genetic algorithm. In classifier design two main paradigms are considered: multilayer perceptrons and adaptive fuzzy logic systems. A multilayer perceptron is a network inspired by biological neural systems. Even though it is far from a biological system, it possesses the capability to solve many interesting problems in variety fields. Fuzzy logic systems, on the other hand, were inspired by human capabilities to deal with fuzzy terms. Its structures and operations are based on fuzzy set theory and its operations. Adaptive fuzzy logic systems are fuzzy logic systems equipped with training algorithms so that its rules can be extracted or modified from available numerical data similar to neural networks. Both fuzzy logic systems and multilayer perceptrons have been proved to be universal function approximators. Since there are approximations in almost every stage, both of these system types are good candidates for classification systems. In classification problems unequal learning of each class is normally encountered. This unequal learning may come from different learning difficulties and/or unequal numbers of training data from each class. The classifier tends to classify better for a well-learned class while doing poorly for other classes. Classification costs that may be different from class to class can be used to train and test a classifier. An error backpropagation algorithm can be modified so that the classification costs along with unequal learning factors can be used to control classifier learning during its training phase. / Ph. D.

The tremendous ongoing growth of wireless digital communications has raised spectrum shortage and security issues. In particular, the need for new spectrum is the main obstacle in continuing this growth. Recent studies on radio spectrum usage have shown that pre-allocation of spectrum bands to specific wireless communication applications leads to poor utilization of those allocated bands. Therefore, research into new techniques for efficient spectrum utilization is being aggressively pursued by academia, industry, and government. Such research efforts have given birth to two concepts: Cognitive Radio (CR) and Dynamic Spectrum Access (DSA) network. CR is believed to be the key enabling technology for DSA network implementation. CR based DSA (cDSA) networks utilizes white spectrum for its operational frequency bands. White spectrum is the set of frequency bands which are unoccupied temporarily by the users having first rights to the spectrum (called primary users). The main goal of cDSA networks is to access of white spectrum. For proper access, CR nodes must identify the right cDSA network and the absence of primary users before initiating radio transmission. To solve the cDSA network access problem, methods are proposed to design unique second-order cyclic features using Orthogonal Frequency Division Multiplexing (OFDM) pilots. By generating distinct OFDM pilot patterns and measuring spectral correlation characteristics of the cyclostationary OFDM signal, CR nodes can detect and uniquely identify cDSA networks. For this purpose, the second-order cyclic features of OFDM pilots are investigated analytically and through computer simulation. Based on analysis results, a general formula for estimating the dominant cycle frequencies is developed. This general formula is used extensively in cDSA network identification and OFDM signal detection, as well as pilot pattern estimation. CR spectrum awareness capability can be enhanced when it can classify the modulation type of incoming signals at low and varying signal-to-noise ratio. Signal classification allows CR to select a suitable demodulation process at the receiver and to establish a communication link. For this purpose, a threshold-based technique is proposed which utilizes cycle-frequency domain profile for signal detection and feature extraction. Hidden Markov Models (HMMs) are proposed for the signal classifier. The spectrum awareness capability of CR can be undermined by spoofing radio nodes. Automatic identification of malicious or malfunctioning radio signal transmitters is a major concern for CR information assurance. To minimize the threat from spoofing radio devices, radio signal fingerprinting using second-order cyclic features is proposed as an approach for Specific Emitter Identification (SEI). The feasibility of this approach is demonstrated through the identification of IEEE 802.11a/g OFDM signals from different Wireless Local Area Network (WLAN) card manufactures using HMMs. / Ph. D.