| Other Research Projects |
| Networked Control Systems |
| Smart Electricity Grid |
| Micro-actuator Control |
Past Projects
Project 1: Quantized Feedback Conrol
Collaborators: Lihua Xie (Nanyang Technological University, Singapore); Carlos de Souza (LNCC, Brazil)
Project Summary: The paradigm of control over communication links is depicted as above. Traditional control theory assumes that the feedback channel is analog and solely dedicated to control purposes. However, more and more industrial systems are controlled via digital communication links such as Fieldbus, local area networks, and even the internet. These communication links are shared with other network functions. Moreover, feedback signals must be sampled and quantized. Feedback control using communication links is a new challenge for control design. It raises a number of fundamental questions which cannot be answered using classical control and estimation theories. One of the important questions is how to jointly design quantizers and feedback controllers. This question is fundamental because if we ignore transmission dealys and transmission errors in the network, which is practical assumption in many applications, quantization and its effect to control design is the only major problem. Our recent research focuses on two types of quantized feedback control problems. The first type uses static (memoryless) quantizers, whereas the second type uses dynamic quantizers. For stabilization problems and many other control problems for linear systems, the use of static quantizers leads to the so-called logarithmic quantization. Our work shows how to optimize logarithmic quantization by relating the design problem to a robust control design problem. When dynamic quantizers are used, we introduce a design scheme which makes the quantized feedback system with good control performance by using a very moderate bit rate for feedback. We have also studied quantized feedback control problems for sampled-data systems and for uncertain systems. Currently we are also studying quantized estimation problems.
Project 2: Interpolation Approach to Spectral Estimation
Collaborators: Kaushik Mahata (University of Newcastle).
Project Summary: Spectral estimation is a well-studied research area which numerous techniques available. The so-called parametric specrtal estimation typically assumes that the spectrum of a given time-domain signal can be fitted using a rational function model. The problem involves estimating the parameters of the model from a set of samples of the signal. Most methods start with computing some sort of statistics of the signal and then use these statistics to estimate the spectral model. In this project, we study the use of interpolation methods for spectral estimation. The rough idea is as follows: Instead of estimating the spectrum directly, we estimate the so-called half spectrum which is uniquely related to the spectrum via a standard spectral decomposition. The half spectrum is a strictly positive-real rational function. The given samples are first passed through a set of linear filters, called input-to-state filters. The output of these filters can be used to directly estimate the values of the half spectrum at a set of pre-specified locations in the complex plane. These values can then be used as the interpolation constraints on the half spectrum and the latter can be determined by solving an interpolation problem (Nevalinna-Pick type). This approach was originally proposed by T. T. Georgiou and his co-authors for discrete-time processes. Our research generalises the above method in several directions. We have developed a similar approach to spectral estimation for continuous-time processes. Compared with the common approach which estimates a discrete-time model first, our approach is a direct approach, avoiding difficulties arising in nonlinear conversion between discrete-time and continuous-time models. We have also developed a robust interpolation method which allows us to use redundant interpolation data to gain more accurate spectral fitting. In addition, we have studied more efficient algorithms for the interpolation appraoch to spectral estimation.
Project 3: Subband Approach to System Identification and Euqlisation
Collaborators: Damian Marelli (University of Newcastle).
Project Summary: The main advantage of this method is to save computation, especially when the linear model in the full-band case is an finite-impulse-response (FIR)Â model with a long tap size. It turns out that for the same estimation accuracy, the amount of computation is inversely proportional to the number of subband channels. However, the subband approach involves the overhead of filterbank computation, which increases as the number of subband channels increases. The optimal number of subbands is typically between 2 to 10. We have studied the tradeoffs among the number of subbands, identification performance and computational load in detail. Our work provides a theoretical foundation and practical guidelines for the use of subband approach. We have also applied the subband approach to an channel equalisation problem in digital communications (to do with channel estimation and equalisation of orthogonal-frequency-divison-multiplexing).
Project 4: Dynamic Modeling of Iterative Decoding
Project Summary: One of the major breakthroughs since the pioneering work of
Project 5: Design and Analysis of Low-density parity-check Codes
Project Summary: The aim of this project is to better understand how LDPC codes work in terms of both coding and decoding properties. We have discovered a new class of irregular LDPC codes which has only (true) linear complexity for encoding but offers very similar performance to randomly chosen irregular LDPC codes. When the block size of the code is sufficiently large, this performance approaches what can be best offered by irregular LDPC codes (i.e., the so-called loop-free case). This result allows powerful LDPC codes to have true linear complexity for both encoding and decoding. We have also studied the validity of Gaussian approximation for characterisation of the terative decoding for LDPC codes.
Project 6: Iterative Equalisation and Decoding
Project Summary:The aim is to study how to apply iterative algorithms to joint equalisation and decoding.
Project 7: Channel Modelling and Channel Capacity of MIMO Systems
Collaborator: Leif Hanlen (Former Ph.D. student, currently with NICTA, Canberra)
Project Summary: In this project, we study the problems of channel modeling and channel capacity estimation for wireless communications systems with multi-antennas. Multi-antenna design has been used recently in conjunction with space-time coding to significantly improve the channel capacity of a wireless communication system. However, the known theoretical estimates of channel capacities are based on channel models with ideal assumptions on multipath propagation. In this project, we propose to model a multi-antenna system using the fundamental physical principles on radio wave propagation. The aim is to derive an accurate yet practical multi-input-multi-output model for communications between arbitrary antenna arrays in two volumes in a scattering environment. This model can then be used to compute the channel capacity of a multi-antenna system.
