D.Ing. / This work comprises a study of reflective interferometric fibre optic sensors. The use of Bragg gratings, multilayer quarter-wave stacks, and rugate mirrors for fibre optic sensing applications is discussed in this work. Rugate mirrors are presented in more detail since they form an important part of the research. The control system for an electron cyclotron resonance plasma enhanced chemical vapour deposition reactor was developed for the growth of inhomogeneous dielectric filters. The system is intended to control the growth of films of the required refractive index profile for optical applications on different substrates including fibre ends. The system also includes an automatic microwave tuner. Rugate mirrors deposited directly on optical fibre were used in a dual cavity Fabry- Perot interferometric strain sensor. It was found by computer simulation that reflectances of 40% for the two outer mirrors and 68% for the centre mirror allow the interferometer to have maximum fringe amplitude. The sensor was evaluated experimentally as a strain gauge. The maximum strain applied to the sample was about 0.12% and the corresponding phase change was about 800 radians. The discrepancy between the readings of this sensor and those of a resistive strain gauge, applied to the same structure, did not exceed 5%. Two novel fibre optic sensors were developed during this work: a merged Sagnac- Michelson interferometric sensor for distributed disturbance detection, and a disturbance location sensor using modified Sagnac and Mach-Zehnder interferometers. Both are intended for distributed impact location along the fibre. The magnitude of impact can also be measured with these sensors. The merged Sagnac-Michelson interferometric sensor uses two light sources and a frequency selective mirror to separate the Michelson and Sagnac signals. Birefringence in the fibre was used to bias the Sagnac interferometer to have a sine response. Computer simulations yielded the optimal biasing conditions: the state of linear polarisation of the input light should be rotated by r/4 with respect to the horizontal axis; birefringence in the Sagnac loop must provide retardation of ff/2 with the optical axes rotated by K/4 with respect to the horizontal axis. We verified the sensor concept experimentally. The discrepancy between measured and actual locations of disturbances applied to the fibre did not exceed 2.5 m for a 200 m long sensing loop. The sensor using the modified Sagnac and Mach-Zehnder interferometers makes use of phase modulation and synchronous detection to obtain the sine-biased Sagnac signal. A modified Sagnac interferometer configuration that incorporates an additional coupler and a mirror, allows separation of the Sagnac and Mach-Zehnder signals. Operation of the new configuration was verified experimentally in the system with a sensing fibre of 100 m long. The discrepancy between measured and actual locations of disturbances applied to the fibre did not exceed 2 m.

D.Phil. (Electrical & Electronic Engineering Science) / Recent advancement in optical fibre communications technology is partly due to the advancement of optical thin-film technology. The advancement of optical thin-film technology includes the development of new and existing deposition and optical filter design methods. Genetic algorithm is one of the new design methods that show promising results in designing a number of complicated design specifications. The research is entirely devoted to the investigation of the genetic algorithm design method in the design of producible polarised and non-polarised edge filters for optical fibre communication applications. In this study, a number of optical filter design methods such as Fourier Transform and refining are investigated for their potential in designing those kinds of structures. Owing to the serious limitations to which they are subject, they could not yield the kind of results anticipated. It is the finding of this study that the genetic algorithm design method, through its optimisation capability, can give reliable and producible designs. This design method, in this study, optimises the thickness of each layer to get to the best possible solution. Its capability and unavoidable limitations in designing polarised and non-polarised beam splitters, edge filters and reflectors from absorptive and dispersive materials are well demonstrated. It is observed that the optical behaviour of the non-polarised filters designed by this method show a similar trend: as the angle of incidence increases the inevitable increase in the percentage of polarisation, stop bandwidth and ripple intensity is well controlled to an acceptable level. In the case of polarised designs the S-polarised designs show a better response to the optimisation process than the P-polarised designs, but all of them are kept well within an acceptable level. It is also demonstrated that polarised and non-polarised designs from the genetic algorithm are producible with great success. This research has accomplished the task of formulating a computer program using genetic algorithm in a Mathlab® environment for the design of producible polarised and non-polarised filters from materials of absorptive and dispersive nature.

In this thesis two novel single-end methods are applied to measure and characterize polarization mode dispersion in single mode optical fibres. Polarization mode dispersion (PMD) is an important factor negatively affecting the successful implementation of high speed long haul optical fibre networks operating at bit rates of 10Gb/s and above. PMD measurements are thus important for quality control during manufacturing and cabling processes. It is also useful for network operators planning to upgrade bitrates in existing networks to 10Gb/s and beyond. In an optical fibre link, sections with particularly high PMD may act to increase the entire PMD of the link. Identifying and replacing such sections can greatly reduce the PMD of the link. PMD measurements can be forward or single-end. In forward measurements, both ends of the fibre are used for input and detection. In single-end configuration, only one end of the fibre is used. For this reason, single-end measurements are more practical for the field where fibre ends are situated several kilometres apart. Single-end techniques can be implemented with a continuous wave for non-local PMD measurements (by Fresnel reflection). If a pulsed wave is used, local measurements can be achieved (by total power due to Rayleigh scattering). Two single-end schemes, one based on Fresnel reflection and the other due to Rayleigh scattering have been applied to measure non-local and local PMD of standard single mode optical fibres. For the non-local PMD measurements, the general interferometric technique (GINTY) was modified to operate in a round-trip configuration. In this configuration, the fibre was treated as a concatenation of two identical fibre segments. Three different sets of fibres were investigated, each set representing a particular mode coupling regime. For polarization maintaining fibres, (PMFs), with no mode coupling, a factor of two was found between forward and single-end measurements. For long single mode fibres in the long length regime, the factor was 1.4. For a combination of PMF and single mode fibres, a factor of 1.6 was obtained. The method which is accurate, repeatable, low cost and robust is very suitable for field applications. The second method is the polarization optical time domain reflectometric (P-OTDR) technique. This technique performs local birefringence measurements by measuring the evolution of the states of polarization (SOP). The birefringence information from such measurements was extracted and analysed to characterise four different fibres. Beat lengths and correlation lengths extracted from the P-OTDR were used to calculate the differential group delay (DGD) of the fibres. Next an expression for the root-mean-square differential group delay was derived and applied to the birefringence measurements to calculate the DGDs at a single wavelength. This method which operates at a single wavelength has a huge advantage. Firstly it is able to measure completely all the fibre characteristic parameters. Secondly it can measure mean DGD, root mean square DGD and instantaneous DGD. A plot of instantaneous DGD vs. length enables one to identify and eliminate sections with particularly high DGD. Finally since the P-OTDR system operates with a single wavelength, real time monitoring of PMD is possible via multiplexing. The results obtained are repeatable, accurate and are in good agreement with the standard Jones Matrix Eigenanalysis (JME) technique.

The demand for faster data transmission is ever increasing. Wavelength division multiplexing (WDM) presents as a viable solution to increase the data transmission rate significantly. WDM systems are based on the ability to transmit multiple wavelengths simultaneously down the fibre. Unlike time division multiplexing (TDM) systems, WDM systems do not increase the data transfer by increasing the transmission rate of a single channel. In WDM systems the data rate per channel remains the same, only multiple channels carry data across the link. Dense wavelength division multiplexing (DWDM) promises even more wavelengths packed together in the same fibre. This multiplication of channels increases the bandwidth capacity rapidly. Networks are looking into making use of technology that will ensure no electronic signal regeneration at any point within the DWDM network. Examples are; reconfigurable optical add/drop multiplexers (ROADM) and optical cross connect (OXC) units. These components essentially enable network operators to split, combine and multiplex optical signals carried by optical fibre. WDM allows network operators to increase the capacity of existing networks without expensive re-cabling. This provides networks with the flexibility to be upgraded to larger bandwidths and for reconfiguration of network services. Further, WDM technology opens up an opportunity of marketing flexibility to network operators, where operators not only have the option to rent out cables and fibres but wavelengths as well. Cross phase modulation (XPM) poses a problem to WDM networks. The refractive index experienced by a neighbouring optical signal, not only depends on the signal’s intensity but on the intensity of the co-propagating signal as well. This effect leads to a phase change and is known as XPM. This work investigates the characteristics of XPM. It is shown that, in a two channel WDM network, a probe signal’s SOP can be steered by controlling a high intensity pump signal’s SOP. This effect could be applied to make a wavelength converter. Experimental results show that the degree of polarization (DOP) of a probe signal degrades according to a mathematical model found in literature. The pump and probe signals are shown to experience maximum interaction, for orthogonal probe-pump SOP vector orientations. This may be problematic to polarization mode dispersion compensators. Additionally, experimental results point out that the SOP of a probe signal is much more active in the presence of a high intensity pump, as compared to the single signal transmission scenario.

This thesis provides in-depth information on the high speed optical transport requirements for the Square Kilometre Array. The stringent data rates as well as timing and synchronization requirements are dealt with respect to the optical fibre technology. Regarding the data transport, we draw a clear comparison between a typical telecommunication access network and a telescope network. Invoking simulations and experiments on the field and laboratory test bed, we successfully implement a suitable telescope network using vertical cavity surface emitting laser (VCSEL) technology. Polarization effects on the KAT-7 telescope network, an operational prototype for the SKA is studied so as to estimate the expected effect in the MeerKAT telescope with transmission distances _ 12 km. The study further relates the obtained values to the expected impact on the distribution of the time and frequency reference in the MeerKAT array. Clock stability depends on the differential group delay (DGD) and polarization stability. On a 10:25 km link that includes the riser cable a DGD of 62:1 fs was attained. This corresponds to a polarization mode dispersion (PMD) coefficient of 19:4 fs=km1=2. This is a low PMD value considering telecommunication network. The PMD value is within the allowed budget in the telescope network. However, this may not be the case at longer baselines extending to over 1000 km as expected in SKA 2. The fibre's deployment contribution to the DGD is measured by comparing the deployed fibre to the undeployed of equal lengths. On the 10:25 km deployed single mode fibre, the maximum and mean DGDs measured were 217:7 fs and 84:8 fs respectively. The undeployed fibre of similar type and equal length, gave a maximum and minimum DGDs of 58:6 fs and 36:3 fs respectively. The deployment is seen to increase the maximum and minimum DGDs by factors of 3.7 and 2.3 respectively. This implies that fibre deployment is very critical in ensuring the birefringence is minimized. Polarization fluctuation recorded a maximum of 180o during the 15 hour real time astronomer use of the antenna. To ascertain the contribution of the riser cable, state of polarization (SOP) of the buried section of the single mode fibre in the link was established. A maximum SOP change of 14o over 15 hour monitoring was measured. From the stability realized on the buried section of the fibre, the change in polarization is contributed by the riser cable. The fluctuation in polarization can cause the phase of a clock signal to drift between the birefringent axes by an equal amount corresponding to DGD. We experimentally demonstrate how polarization stabilization can be attained using the polarization maintaining fibre. We also demonstrate the applicability of VCSEL technology in the SKA unidirectional data flow especially for shorter baselines < 100 km. The VCSEL is a low cost light source with attractive advantages such as low power consumption, high speed capabilities and wavelength tuneability. This work entails the use of traditional amplitude modulation commonly known as non-return-to-zero (NRZ) on-off keying (OOK) because of its simplicity and cost. For the MeerKAT typical distances, we show that even in a worst case scenario, the use of VCSEL on different fibres in MeerKAT distance is achievable. Using the impairment reduction approach, we successfully manage to achieve transmission distance beyond MeerKAT. Several in-line dispersion compensation mechanisms in telecommunication have been successfully employed. The work focused on the use of negative dispersion fibre to mitigate the chromatic dispersion effects in the optical fibre. The inverse dispersion fibre (IDF) is proposed for compensation in the conventional zero dispersion wavelength fibres, G.652 that are used at the third window. Similarly, the chromatic dispersion compensation of non-zero dispersion shifted fibre (NZDSF) is experimentally demonstrated using negative dispersion submarine reduced slope (SRS), G. 655 (-). With dispersion management, we demonstrate how transmissions beyond MeerKAT baselines can be achieved error free. A systematic investigation of the use of distributed Raman amplification to overcome the attenuation losses is provided. High on-off gains of up to 15 dB, 8 dB and 5 dB for bidirectional, forward and backward pumping respectively is achieved on a 25 km Raman optimized NZDSF-Reach fibre. Combined dispersion mitigation technique and low noise distributed Raman amplification, up to about 80 km transmission was achieved on a 4:25 Gbps modulated VCSEL using a single pump. Using bidirectional pumping, more than 100 km of transmission was achieved error free. The high gains enhance the VCSEL transmission distance. We further suggest a novel way of using the Raman pump to distribute the clock signal while amplifying the data signal streaming the astronomical data from the remote placed telescope receivers. In summary, the work presented in this thesis has demonstrated the potential use of VCSEL technology for data collection in the telescope array. We have studied the optical effects and mitigation so as to improve the clock and data transmission. This work is relevant and valuable in providing SKA with VCSELs, an option for extremely high network performance at reasonable costs.

In this study, focus is centred on the measurement and emulation of first-order (FO-) and second-order (SO-) polarization mode dispersion (PMD). PMD has deleterious effects on the performance of high speed optical transmission network systems from 10 Gb/s and above. The first step was characterising deployed fibres for PMD and monitoring the state of polarization (SOP) light experiences as it propagates through the fibre. The PMD and SOP changes in deployed fibres were stochastic due to varying intrinsic and extrinsic perturbation changes. To fully understand the PMD phenomenon in terms of measurement accuracy, its complex behaviour, its implications, mitigation and compensation, PMD emulation is crucial. This thesis presents emulator designs which fall into different emulator categories. The key to these designs were the PMD equations and background on the PMD phenomenon. The cross product from the concatenation equation was applied in order to determine the coupling angle β (between 0o and 180o) that results in the SO-PMD of the emulator designs to be either adjustable or fixed. The digital delay line (DDL) or single polarization maintaining fibre (PMF) section was used to give a certain amount of FO-PMD but negligible SO-PMD. PMF sections (birefringent sections) were concatenated together to ensure FO- and SO-PMD coexist, emulating deployed fibres. FO- and SO-PMD can be controlled by altering mode coupling (coupling angles) and birefringence distribution. Emulators with PMD statistics approaching the theoretical distributions had high random coupling and several numbers of randomly distributed PMF sections. In addition, the lengths of their PMF sections lie within 20% standard deviation of the mean emulator length. Those emulators with PMD statistics that did not approach the theoretical distributions had limited numbers of randomly distributed PMF sections and mode coupling. Results also show that even when an emulator has high random mode coupling and several numbers of randomly distributed PMFs, its PMD statistics deviates away from expected theoretical distributions in the presence of polarization dependent loss (PDL). The emulators showed that the background autocorrelation function (BACF) approaches zero with increasing number of randomly mode coupled fibre sections. A zero BACF signifies that an emulator has large numbers of randomly distributed PMF sections and its presence means the opposite. The availability of SO-PMD in the emulators made the autocorrelation function (ACF) x asymmetric. In the absence of SO-PMD the ACF for a PMD emulator is symmetric. SO-PMD has no effect on the BACF. Polarization-optical time domain reflectometry (P-OTDR) measurements have shown that certain fibre sections along fibre link lengths have higher FO-PMD (HiFO-PMD) than other sections. This study investigates the impact of a HiFO-PMD section on the overall FO- and SO-PMD, the output state of polarization (SOP) and system performance on deployed fibres (through emulation). Results show that when the wavelength-independent FO-PMD vector of the HiFO-PMD section is greater than the FO-PMD contributions from the rest of the fibre link, the mean FO-PMD of the entire link is biased towards that of the HiFO-PMD section and the SO-PMD increases (β ≠ 0o or 180o) or remains fixed (β = 0o or 180o) depending on the coupling angle β between the HiFO-PMD section and the rest of the fibre link. In addition, the FO-PMD statistics deviates away from the theoretical Maxwellian distribution. However, experimental results show that the HiFO-PMD section has negligible influence on the SOPMD statistical distribution. An increase in the amount of FO-PMD on a HiFO-PMD section reduces the output SOP spread to a given minimum, in this study the minimum was reached when the HiFO-PMD ≥ 35 ps. However, the outcome of the output SOP spread depends on the location of the HiFO-PMD section along the fibre link length. It was found that when the HiFO-PMD section introduces SO-PMD, the bit error rate (BER) is much higher compared to when it does not introduce SO-PMD.

Polarization mode dispersion (PMD) is stochastic in nature and continues evolving in an unpredictable manner according to the changing environment. Nonlinear birefringence in multichannel systems alters the polarization states of the bits, so that they vary from one bit to the next in a way that is difficult to predict. These are the two major signal-impairment effects that are inherent in optical fibre transmission links which can seriously degrade network performance. It is therefore extremely challenging to compensate for both linear and nonlinear birefringence in multichannel systems. The purpose of this thesis is to investigate the interaction between PMD and nonlinear induced birefringence in a fibre with consideration of mode coupling. A sound knowledge of this interaction is necessary in designing a linear and nonlinear polarization mode dispersion compensator for WDM systems, as was successfully carried out in this study. The investigation shows that the effect of nonlinear birefringence alone depolarizes the signal, while in high PMD links where polarization mode coupling is high, the nonlinear birefringence effect couples with second-order PMD such that it may reduce the penalty and improve the signal DOP. Further investigation shows that when nonlinear birefringence becomes significant, asymmetry arises between the two principal axes of the fibre, such that it is only one axis which experiences the effect of nonlinear birefringence. It is found out that along this vii axis, there exists a critical point in pump power where the nonlinear birefringence cancels PMD in the link and improves the signal. An adaptive compensator to cancel PMD and nonlinear birefringence was designed based on feedforward DOP-monitoring signal. The compensator was tested both at laboratory level and on the Telkom buried fibre link and found to be functioning as intended. It was able to adaptively track and compensate PMD in the link in less than a second. The compensator was able to cancel PMD in the link up to a maximum of 30 ps. The compensator improved the DOP of the worst signal by more than 100 percent.

Frequency and intensity noise are characterized for Erbium-Doped Fiber and Brillouin random lasers based on Rayleigh distributed feedback mechanism. We propose a theoretical model for the frequency noise of an Er-doped fiber random lasers using the property of random phase modulations from multiple scattering points in ultra-long fibers. We find that the Rayleigh feedback suppresses the noise at higher frequencies by introducing a Lorentzian envelope over the thermal frequency noise of a long fiber cavity. The theoretical model and measured frequency noise agree quantitatively with two fitting parameters. A similar model, which also includes additional acoustic fluctuations and a distributed gain profile in the fiber, has been speculated for the Brillouin random laser. These random laser exhibits a frequency noise level of < 6 Hz^2/Hz at 2 kHz, which is lower than what is found in conventional narrow-linewidth EDF fiber lasers and Nonplanar Ring Laser oscillators (NPRO) by a factor of 166 and 2 respectively.

A class of low nonlinearity dispersion-shifted and dispersion-flattened fibers for broadband and long haul applications is presented. The refractive index profiles of these fibers assume a depressed-core multi-clad geometry in order to achieve effective-areas much larger than those in conventional optical fibers. A systematic approach for designing large effective-area dispersion-shifted fibers, using a reference W-index profile to initiate the design, is presented. Transmission properties, including effective-area, mode-field-diameter, dispersion, dispersion slope, cutoff wavelength, and bending, microbending and splice losses are evaluated for several design examples. To ascertain that the proposed fibers can be practically fabricated, the effects of varying fiber dimensions and indices on effective-area, mode-field-diameter and dispersion are assessed. It is shown that there is a trade-off between effective-area and mode-field-diameter and, generally, larger effective-areas are associated with larger mode-field-diameters. In other words, less signal distortion due to fiber nonlinearity (larger effective-area) is associated with higher power loss due to bending of fiber (larger mode-field-diameter). Thus, a large effective-area and low bending loss are conflicting requirements. A parameter Q is defined as a performance indicator, considering effective-area and mode-field-diameter. Dispersion-shifted single-mode fiber designs with effective-areas of 78 μm² to 210 μm² and the corresponding mode-field-diameter of 8.94 μm to 14.94 μm, dispersion less than 0.07 ps/nm.km, and dispersion slope of about 0.05 ps/ nm².km are presented. Numerical simulations for propagation of pulses in few designed fibers are performed.Designs of large effective-area dispersion-flattened fibers are also presented, for the first time we believe. These fibers provide large effective-area and low dispersion over an extended range of wavelengths. For our design, over the wavelength range of 1.48 μm < λ < 1.58 μm, the effective-area is 75 μm² to 100 μm², while the dispersion remains below 0.7 ps/nm.km. / Ph. D.

No description available.