Electrical and Electronic Engineering - Theses
Permanent URI for this collection
Search Results
1 - 5 of 5
-
ItemManaged DC power reticulation systems( 1999-11)Electric power engineering, as it applies to low-voltage power reticulation in buildings and industrial sites, is ripe for a ‘paradigm shift’ to bring it properly into the Electronic Age. The conventional alternating-current approach, now over a hundred years old, is increasingly unsatisfactory from the point of view of plant and appliance requirements. Alternative approaches can deliver substantial cost savings, higher efficiencies, power quality improvements, and greater safety. Power reticulation systems in the future can be expected to differ from present systems in two key respects. The first is a greatly increased role for direct current; the second is the augmentation of the power system with a wide range of ‘management’ technologies. Combining these two trends, which can already be observed today, leads to consideration of ‘managed DC’ power reticulation systems, operating from AC bulk supply mains via AC-DC converters.
-
ItemPerformance analysis of Hidden Markov Model based tracking algorithms( 1997)This thesis investigates the performance of Hidden Markov Model (HMM) based tracking algorithms. The algorithms considered have applications in frequency line tracking and target position tracking. The performance of these algorithms are investigated by a combination of theoretical and simulation based approaches. The theoretical based approach focuses on deriving upper bounds on probabilities of error paths in the output of the tracker. Upper bounds on specific error paths, conditioned on typical true paths are derived for a HMM based frequency line tracker that uses continuous valued observation vectors. These bounds are derived by enumerating possible estimated state sequences, and using necessary conditions on the Viterbi scores of these sequences. The derived upper bounds are found to compare well with simulation results. Next, upper bounds on average error event probabilities (averaged over all possible true paths) are derived for the same HMM based frequency tracker. Here, 'error event' refers to a brief divergence of the estimated track from the true path. Numerical computation of the derived upper bounds are shown to compare well with simulation results. Using these bounds a theorem is established which states that optimum tracking, corresponding to minimum error probability, is achieved when model transition probabilities are matched to 'true' transition probabilities of the underlying signal. Other interesting features of this algorithm are analysed, including robustness of the algorithm to variations in model transition probabilities, and characterisation of the benefits of using HMM based tracking as opposed to a simple approach based on isolated Maximum Liklihood estimators. The theoretical analysis is extended to two other HMM based frequency line trackers that use discrete valued observation vectors. A comparative study of the three HMM based frequency line trackers is carried out to arrive at conditions for the superiority of one algorithm over another. The simulation based approach to analysing performance consists of a combination of Monte-Carlo (MC) and Importance Sampling (IS) simulations. MC simulations are carried out at moderate SNR where required computation time for estimating performance measures is feasible. At high SNR, the error probabilities are small and the required computation time becomes infeasible. To overcome this, importance sampling schemes are designed which reduce the computation time by orders of magnitude. Importance sampling is a modified Monte-Carlo method which is useful in the simulation of rare probabilities. The basic principle is to use a different simulation density to increase the relative frequency of "important" events and then weight the observed data in order to obtain an unbiased estimate of the parameter of interest. In this thesis, a systematic procedure based on minimizing an upper bound to the IS estimator variance is used in the simulation density design. High efficiency gains, of order 1013 are demonstrated with the proposed scheme.
-
ItemTechnologies for millimetre-wave fibre-radio systems( 1999)The recent advances in mobile wireless technology have increased the need for more transmission bandwidth to accommodate for future telecommunication services. The millimetre-wave (mm-wave) hybrid fibre-radio system is one of the potential technologies for provision of these broadband services. With the advent of low-loss and high bandwidth optical fibre in telecommunication networks, it is an ideal medium for distributing the broadband mm-wave information. In addition to the provision of broadband services and high capacity, the hybrid fibre-radio system also enables customer mobility. This thesis investigates the performance of different architectures for distributing radio signals over fibre in mm-wave fibre-radio applications. In particular, networks incorporating radio signal distribution as mm-wave frequency and as baseband data over fibre are investigated. There have been many different downlink implementations incorporating mm-wave radio signal distribution over fibre for mm-wave fibre-radio system, however there has been little focus on the uplink path. For the first time a simple mm-wave uplink path in a mm-wave fibre-radio system is presented in this thesis which incorporates direct modulation of a multi-section mode-locked distributed-Bragg reflector (DBR) laser at the base station. The performance of the multi-section laser as a mm-wave optical transmitter is presented, and a detailed characterisation of the multi-section laser stabilised via hybrid mode-locking at fundamental and subharmonic frequencies is carried out. The uplink path implementation using the multi-section laser is further extended to incorporate multi-channel transmission. Efficient multi-channel transmission incorporating the multi-section laser requires the separation of the stabilisation and modulation functions of the drive signal applied to the laser. Two different stabilisation techniques namely fundamental hybrid and subharmonic synchronous mode-locking are considered for multichannel operation, and their performance investigated. A detailed theoretical analysis is then presented that quantifies the impact of fibre chromatic dispersion-induced rf power penalties when using the multi-section laser to transmit mm-wave frequencies over fibre. The model is also used to study the performance of the laser as a function of its operating conditions. The thesis also presents a detailed investigation of the implementation and performance of a mm-wave hybrid fibre-radio system with baseband data delivery over optical fibre, a technique which has not been previously reported. The first realisation of a mm-wave fibre-radio system with baseband data delivery over the optical fibre network is presented. This system incorporates a novel modulation scheme for simultaneous transmission of baseband digital data and remote local oscillator (LO) signal delivery. The modulation scheme exhibits several unique features including the use of a single dual-electrode Mach-Zehnder modulator and the delivery of the LO signal is such a way that it is not affected by fibre dispersion. A full-duplex mm-wave fibre-radio system comprising broadband baseband data transmission is implemented with custom-designed mm-wave diplexers and antennas. A theoretical model is also developed to quantify the performance of the baseband data modulation scheme. The analysis is used to determine the sensitivity of the modulator input parameters on the system performance and good agreement is obtained between the model and measurements of the fibre-radio link incorporating baseband data modulation.
-
ItemEM algorithms for state and parameter estimation of stochastic dynamical systems( 1997)This thesis studies the use of the Expectation Maximization (EM) algorithm for state and parameter estimation of stochastic dynamical systems. Optimal maximum likelihood parameter estimates via the EM are computed for: i) Gaussian state space models, where explicit rules are given for optimally partitioning the parameter space to update some parameters using the Newton-Raphson method and the EM method (on the remaining parameters). The partitioning is optimized to ensure minimum processing time for computing the estimates, ii) Errors-in-variables models driven by additive Gaussian and finite-state Markovian disturbances, and iii) 1-bit quantized Markov modulated autoregressive process, generalizing the binary time series algorithm of [Kedem 1980] for linear time series to Markov modulated time series. While the EM is widely used as an iterative numerical method for maximum likelihood parameter estimation of partially observed stochastic models, a significant contribution here is to use the EM for optimal state estimation. In particular, the EM is used to compute maximum a posterior state sequence estimates of jump Markov linear systems. An off-line optimal state estimator is derived, which iteratively combines a Hidden Markov Model Smoother and a Kalman Smoother. Two applications are extensively studied: i) Maximum a posterior state estimates of a maneuvering target in clutter. and ii) Narrowband interference suppression in spread spectrum code division multiple access systems. This thesis also studies open-loop control strategies using information theoretic criteria. Optimal observer paths are derived for the bearings-only tracking problem. Optimal optimization techniques, such as forward and backward dynamic programming and enumeration with optimal pruning are derived. Furthermore, a number of computationally tractable suboptimal optimization techniques, such as approximate reduced complexity forward dynamic programming and one-step ahead (suboptimal) control strategies are presented.
-
ItemStudies in nonlinear filtering theory: random parameter linear systems, target tracking and communication constrained estimation( 1998)The focus of this thesis is nonlinear filtering for discrete-time stochastic systems. In particular, we consider optimal and suboptimal filtering algorithms for random parameter linear systems and state estimation for Markov chains in the presence of communication constraints. The thesis presented in three parts. (For complete abstract open document)