SuperDARN is an international network of 35 HF radars located near the poles of the Earth dedicated to determining the state of the ionosphere at high latitudes. One of the SuperDARN radars is located at South Africa's base in Antarctica (SANAE IV) and is administered by the South African National Space Association (SANSA). The radar at SANAE IV was recently upgraded to a fully digital transceiver, with the addition of a Field Programmable Gate Array (FPGA) at the core of this upgrade. FPGAs allow for easy hardware reconfiguration and high-performance computing. The aim of this project is to determine the feasibility of using the FPGA on board the radar at SANAE IV to implement a riometer mode to run simultaneously with its main mode of operation, adding a new tool to the radar's set of abilities without any investment in new hardware. The riometer function could easily be ported to other radars in the SuperDARN network, allowing for a significant increase in riometer coverage of the polar regions. As a first step towards achieving this goal, a demonstrator riometer is developed using the Red Pitaya FPGA platform as its backend, and tested at the University of Cape Town, at SANSA in Hermanus and at Fish Hoek. A riometer measures the opacity of the ionosphere with respect to cosmic radiation. Doing this over a wide band of frequencies results in a spectral riometer. This dissertation describes the design and implementation of both a single frequency and a spectral riometer, both implemented on the Red Pitaya, and the results of testing these implementations. Experimentation alongside a La Jolla 38MHz riometer revealed very similar performance for the low-cost demonstrator riometer. It is thereby shown that low cost HF riometry is possible and that it is feasible to implement a riometer on the radar at SANAE IV. However, an additional FPGA is required.

The Super Dual Aural Radar Network (SuperDARN) forms a multinational collaboration consisting of over 30 HF radars used to monitor charged particle convections in the high latitudes of the ionosphere. This research is inspired by the antenna arrangement of one of these HF radars situated at the South African National Antarctic Expedition (SANAE) IV base. This radar consists of a steered 16-element primary array, and a second 4-element interferometer array. Phase wrapping is introduced into echoed returns due to the two arrays being separated by 100 m, several wavelengths of the carrier frequency. As a result, the radar system is only able to unambiguously determine the Angle of Arrival (AOA) for a given target provided it resides below 33ᵒ in elevation when observing along boresight. This dissertation models a subcomponent of this phenomenon and develops an entirely separate pulsed-Doppler radar system on an FPGA platform known as the Red Pitaya. Operating parameters are specifically chosen to be consistent with the SANAE IV radar. An antenna arrangement consisting of two receivers and a transmitter is constructed to replicate the AOA issue experienced by the SANAE radar. Instead of focusing on detecting targets in elevation, as done with the SuperDARN radars, targets are detected in azimuth. Three algorithms are developed to autonomously measure the AOA and to counteract the effects of 2π phase wrapping. The naïve measurement applies no corrections and is only able to detect targets at angles within ±17.10ᵒ of boresight. This is the theoretical maximum measurable AOA when operating at 17 MHz. A second algorithm involving a Standard Deviation (STDEV) search process is used to successfully measure angles up to ±70ᵒ for the same arrangement. It does this by varying the carrier frequency and using numerous received aliased AOAs to correct and expand for a true AOA. Empirical results yield an average Mean Square Error (MSE) of 0.29 between the true and measured target angles. A third algorithm involving the Chinese Remainder Theorem (CRT) was shown to work in theory (albeit only up to ±66ᵒ), but failed to perform in practice. Results presented in this dissertation indicate that the maximum measurable AOA is expanded by a factor of 4 by using the STDEV algorithm. It is postulated that the same result can be achieved using the actual SuperDARN radar. Theory relevant to generating the RF Linear Frequency Modulated (LFM) chirp and its use for transmitting, receiving and processing is covered. Topics of pulse compression, range resolution, environment modelling, reconstruction and matched filtering, coherent integration, sidelobe reduction and demodulation are detailed.

The objectives of this minor dissertation were to investigate relevant theory, models and processes required for the development of a site specific radar coverage and land clutter modelling tool. Various sources of digital elevation model (DEM) and land cover (LC) data were investigated. It was found that the ASTER GDEM and SRTM 30 m DEM datasets can be used to characterise land topography for all intended areas of interest. It was also found that two LC datasets, namely the National Land Cover 2009, and GlobeLand30 m data sources can be used to characterise land cover for all intended areas of interest. For each terrain type found in the GlobeLand30 or NLC 2009 datasets, a decision was made as to which of the terrain types for each land clutter model matches the land cover data terrain type the closest. These classifications were presented in the form of tables. It was concluded that the SRTM 30 m DEM dataset and the GlobeLand30 LC dataset should be used as they are currently the highest quality DEM and LC datasets that are freely available that covers all intended areas of interest. Numerous monostatic land clutter models exist in literature that address specific cases of clutter types and behaviours. Nine such land clutter models were investigated. Measured land clutter data collected over various terrain types in the Western Cape region of South Africa are compared to simulated backscatter data from these land clutter models. From insights gained from the literature study as well as the analysis of these comparisons, a classification was made on each model's compatibility and validity for different grazing angles and frequency ranges. A classification table was presented indicating the appropriate land clutter models to use in order of their validity, with respect to different grazing angle regions and frequency ranges.

Research into multi-static, multi-band networked radar has led to the development of the NeXtRAD radar system. This dissertation will investigate the design and implementation of a dual polarised L-Band prime focus dish antenna with a centre frequency of 1.3 GHz and a HPBW of 10° in the azimuth plane. The antenna is required to handle a peak power of 1.5 kW over a 50 MHz bandwidth and be able to withstand environmental factors such as wind while mounted on a tripod. This dissertation forms part of the larger NeXtRAD project and as such, the antenna design requirements have been set based on the wider system specifications. Previous investigations into the feasibility of various antenna designs have concluded that a prime focus parabolic dish antenna would be the most appropriate to meet the design requirements. The dissertation details the design and manufacturing process followed. All antenna parameters have been simulated using a combination of FEKO v7 and CST 2014 to compare and verify the designs and simulations. Due to manufacturing limitations, the optimal antenna design could not be manufactured and, as a result, compromises had to be made in order for an antenna prototype to be manufactured and tested. These tests include, amongst others, characterisation of the return loss, cross polarisation, gain, beamwidth and beam pattern of the antenna in both planes of polarisation. These results have been recorded, analysed and compared to those found through simulations.

This dissertation discusses the re-development of a subsystem of the tracking radar (TR) at Denel Overberg Test Range (OTR), defined as the Synchronisation and Video Conversion Unit (SVCU). The SVCU performs the task of baseband video processing, generation of all synchronisation triggers, and timing within the TR. The report is based on the Research and Design (R&D) project conducted at Denel OTR and built on the previous study where an SVCU model was designed and tested in SystemVue. A comprehensive measurement and analysis of all functions was first conducted on all (identical) TRs of Denel OTR to verify against the limited literature that is available relating to the SVCU. Various anomalies were discovered between the design specification documents and measured values. Modifications made to functions in the SVCU as a result of development done on other subunits of the TR and not documented were also discovered. A comprehensively revised design specification and high-level description for the SVCU was then generated. A new SVCU architecture was specified to improve the current tasks performed by various discrete analogue and digital components, Single-Board Microcontroller and data bus interfaces using commercial off the shelf (COTS) hardware. A prototype was built on the National Instruments (NI) PCI eXtensions for Instrumentation Express (PXIe) platform, utilising high-speed ADCs and Field Programmable Gate Array (FPGA) modules. All functions of the SVCU were modelled and implemented on these modules using LabVIEW and LabVIEW FPGA software. Fractional Decimators were designed to meet the sample resolution requirement of the range gates (used for range and Doppler measurement). Custom functions were written to integrate samples to increase SNR and apply a correction for errors carried over during the I/Q demodulation. The jitter on the synchronisation pulse responsible for RF energy transmission (TX), the ADC sample clock, and triggers signals that need to travel over excessively long transmission lines used for calibration were found to be the most critical aspects of the SVCU. The rootsum- square of all the jitter on these synchronisation pulses was calculated to ensure that the specified measurement accuracy of the TR is satisfied. Based on these findings, the PXIe based SVCU is recommended for deployment. Further development of other subunits for the TR receiver on the current platform is recommended, and an outline for future work is provided.

This dissertation presents the design of the RF front end for use on the NeXtRAD radar system. The system is intended for research purposes to investigate potential target detection benefits to be derived from a multistatic, dual-band (X- and L-band), polarimetric radar architecture, particularly within dense clutter environments such as the maritime environment. By examining the high-level system requirements and objectives, requirement specifications for the RF front end were derived and a suitable architecture, making use of commercial off-the-shelf components, proposed. This architecture was modified in order to meet cost constraints - subsequently offering reduced levels of functionality but suitable for an initial build. Using this modified RF front end architecture, design verification and system analysis was conducted, both analytically and with the aid of SystemVue, in order to predict both the front end and overall radar detection performance. Once the front end design was found to be satisfactory, it was built and tested in a laboratory environment. Test results revealed a general improvement in performance when compared with the design predictions, yielding peak transmitter power levels in excess of 61dBm at L-band, and 54dBm at X-band. Some non-conformances were also identified, but these were as a result of component problems and not system design. Since the front end could not yet be integrated into the radar, performance modelling was repeated using the final lab test results. This indicated a negligible improvement in receiver single-pulse signal-to-noise ratio, but confirmed that the system performed as predicted. Based on the lab test results, it was concluded that the 'as-built' front end design closely matched the design goals and would be suitable for eventual integration into the first revision of the NeXtRAD system. It was, however, recommended that a concerted effort be made to secure funding to implement the original front end architecture in order to achieve the full system functionality originally desired.

The aim of this study was to research TOA based tracking and deinterleaving algorithms suited to radar emitters in an EW environment for application on the CSIR 5th generation DRFM platform. The research problem statement stipulated that the only defining characteristic of the different emitters be the time of arrival (TOA) of their pulses. The pulse repetition interval (PRI) schemes considered in the study was constant, jittered, staggered and dwell and switch. The different TOA based deinterleaving algorithms investigated were sequence search (SS), TOA difference histogram, CDIF, SDIF, CDIF with SS (CDIF SS), SDIF with SS (SDIF SS) and interleaved pulse train spectrum estimation. The interleaved pulse train spectrum estimation algorithm results could not be replicated and were not included in simulations. The TOA based tracking algorithms that were also investigated were Delta-t histogram, Kalman filter, alpha-beta filter and alpha-beta-gamma filter. The alpha-beta-gamma filter became unstable during simulations and hence their results have also been excluded. The algorithms were simulated in MATLAB against EW environments with varied TOA measurement noise, number of emitters, PRI schemes and interference pulses (missing and spurious). General conclusions drawn from the deinterleaving simulations were the success of the algorithms decrease with the increase of emitters in the EW environment, interference pulses increased the success of some algorithms and the success of algorithms increased with TMNR (time measurement to noise ratio). General conclusions drawn from the tracking simulations were track loss of the algorithms decrease with increase in TMNR, tracking error decreases with increase in TMNR and interference pulses affected the initial estimates used to initialise the filters. The performance of the deinterleaving (CDIF & CDIF SS) and tracking ( Delta-t histogram & alpha-beta filter) algorithms were compared on the DRFM platform. On the DRFM platform, the CDIF algorithm deinterleaved in fewer pulses but had more false detections as compared to the CDIF SS algorithm. The alpha-beta filter performed better with lower TMNR than the Delta-t histogram, on the DRFM platform. The CDIF SS algorithm and alpha-beta filter were chosen, based on their performance on the DRFM, to be implemented on a DRFM based system that would deinterleave and then track emitters in an EW environment. The system was successfully implemented and met all requirements that were placed on it. Possible improvements to the system and the future improvements to the research are also discussed.