Prof. Brett Ninness
Research
My research interests are
in the areas of stochastic (ie. noise corrupted) systems as they
occur in control, signal processing and telecommunications
applications.
In particular, I work on problems of system
identification whereby models of systems (such as telecommunications
channels or chemical processes) are obtained directly by observing the input-output
properties of those systems.
My past and future research has been/is funded by the Australian Research Council as
detailed below. Please refer to my publications
and current projects for more detail.
for more detail.
| Period |
Funding |
Title |
| 2007-2009 |
$246,090 |
Advancing System Identification using Modern Optimisation Methods |
| ARC Discovery Grant (Ninness and Wills as chief investigators). |
| An essential part of science, engineering and economics is the
development of mathematical models to describe how certain quatities
relate to one another. For example, such models have proven to be
extremely powerful in predicting the value of financial instruments,
and in providing high performance control of robots, and in detecting
faults or changes in petrochemical processing plants. This project is
directed at developing such models using modern computer based
optimisation methods, but for situations that, on the one hand, have
previously been considered unsolvable, and on the other, are
acknowledged as being of high practical interest. |
|
| Period |
Funding |
Title |
| 2006-2008 |
$336,000 |
New Approaches for the Estimation of Complex Dynamic System Models |
| ARC Discovery Grant (Ninness as solo chief investigator). |
| In engineering, science and economics it is essential to find
equations that describe how some quantities relate to
others. This is because these equations can be used for prediction
and control of the quantities, or for detection of faults or changes
in the mechanisms that produce them. Examples are share market
prices, chemical plant purities, and distortion of radio channels.
This project is directed at determining the necessary describing
equations via automated computer based methods. Given its importance,
there has already been significant international work in this area.
However, this project will employ new techniques in order to address
more sophisticated modelling problems that have previously been
solvable. |
|
Selected Associated publication |
| Period |
Funding |
Title |
| 2005-2009 |
$993,000 |
New Methods and Microelectronics for Wireless Communications Systems |
| ARC Linkage Grant (Joint with Dr Steve Weller and Agere Systems Australia). |
| Global demand for high quality wireless communications poses
significant challenges.
The so-called "physical layer" is crucial, as this is where the
vagaries of the wireless channel, including interference and limited
bandwidth, are mitigated by sophisticated signal processing.
This project will conduct applied research to meet these physical
layer challenges, providing solutions that feed directly into next
generation wireless communication systems.
Uniquely, this project focuses on the transfer of research from
theoretical genesis, through to realisation of silicon integrated
circuit "chips". This will maximise both the impact of the research
and the potential for significant national economic benefits to
accrue.
In the Innovation Action Plan, Backing Australia's Ability, the
Federal Government recognises the vital role played by information and
communications technologies (ICT) in the economic and social fabric of
Australia.
These technologies are drivers of the information economy, the
spawning of new businesses, and the creation of new jobs.
This research proposal brings together an internationally recognised
cross-disciplinary team to work on technologies which underlie the
success of next-generation wireless communications. |
|
| Period |
Funding |
Title |
| 2002-2005 |
$360,000 |
New methods for the estimation of dynamic systems |
| ARC Discovery Grant (Ninness as solo chief investigator). |
| The estimation of dynamic system models from records of normal operating data is a significant problem across many engineering applications ranging from (at least) chemical process control to the implementation of telecommunications networks. This project, based on a promising pilot study, will develop and analyse new methods for solving these problems. This will lead to significant advances in the simplicity, robustness and generality of methods for accurate model generation with attendant benefits of more efficient and higher performance operation of many engineering systems. |
|
Selected Associated publications
|
| Period |
Funding |
Title |
| 2003-2005 |
$177,000 |
Robust Control and System Identification of Highly Resonant Systems |
| ARC Discovery Grant (Joint with Dr. S.O.R. Moheimanhi). |
| The modelling and control of complex and highly resonant systems is of increasing engineering importance due to their occurence in a wide variety of emerging areas in aerospace, acoustics, robotics and ``smart'' structures. At the same time, effective tools tailored towards identifying the necessary models, and synthesising the necessary controllers for these systems are in their infancy. This arises from special difficulties encountered via the high dimensionality of the structures involved. This research project will employ new methods from the fields of robust control and multivariable system identification theory to lead to new and high performance solutions in this area.
|
|
| Period |
Funding |
Title |
| 2002-2004 |
$194,000 |
Advanced Space-Time Coded Multiuser Wireless Communications via Test-bed Development |
| ARC Linkage Grant with Bell Labs (Lucent Technologies). |
| Meeting the global demand for mobile and wireless communications depends critically on reliable and high rate data transfer. Unfortunately, communications medium idiosyncrasies pose formidable challenges. Very recently, in combatting this, major breakthroughs have been achieved whereby the use of multiple antennas allows for drastic data-rate increases. These advances use sophisticated Space-Time coding methods, and while they are causing great excitement in terms of their simulation performance, it is not clear how they will perform in practice, or in fact how they are to be realistically implemented. This project will address this issue by building a world-first testbed that implements a high rate wireless communications system using Space-Time and other coding methods.
|
|
Associated publications |
| Period |
Funding |
Title |
| 2000-2002 |
$186,000 |
The application of Advanced Control in Materials Rolling |
| ARC SPIRT Grant with Industrial Automation Services (IAS). |
| In the context of the Australian Economy, one of the most significant areas of value-adding to export products involves the rolling of materials; steel, aluminium, wire, and others. Central to the competiveness of the resultant products is the quality they derive via computer control of the machinery involved in the rolling process. This project concentrates on applying recent sophisticated control-theoretical developments (arising principally in academic arenas) to practical problems of rolling. Although, via the industrial partner, the focus will be on steel rolling, applications to the rolling of other materials are envisaged. The results will allow for more efficient production of higher quality rolled products. |
|
| Period |
Funding |
Title |
| 1999-2001 |
$176,240 |
New Methods for Nonlinear System Identification |
| ARC Large Grant (Ninness as solo chief investigator). |
| System Identification is a field where one tries to intelligently determine the global behaviour of a system by observing it locally for a fixed period. This global behaviour is captured by a mathematical model, and an important use for such a model is to allow the design of automatic control systems. To date, the vast bulk of system identification work has concentrated on linear system models. However, it has become increasingly apparent that most systems can be better controlled if nonlinear models for them can be found. The purpose of this project is to develop methods together with supporting theory for the estimation of nonlinear system models that are suitable for control design. |
|
Associated publications |
| Period |
Funding |
Title |
| 1996-1998 |
$91,500 |
Examination of the use of Orthonormal Bases in System Identification |
| ARC Large Grant (Ninness as solo chief investigator). |
| System identification is a field where one tries to intelligently
determine the global behaviour of a system by observing it locally for a
fixed period of time. This global behaviour is captured by a
mathematical model, and an important issue is how this model should be described.
The aim of this project is to attack this problem by breaking the description
into simple units, known as basis functions. In particular, the idea
is to choose the basis functions so they are adapted to be appropriate to the system
under study. |
| Associated publications |
| Period |
Funding |
Title |
| 1994-1996 |
$101,000 |
Analysis of
the Interplay between System Identification and Robust Controller Design. |
| ARC Large Grant (Ninness as solo chief investigator) |
| A common engineering problem is to automatically control a
system. Such problems are currently solved by using a mathematical
model for the system. However, there are problems if this model
is not accurate. In response, two very active areas of research have
developed. One, going under the title of `robust control', allows the
inaccuracies to be taken into account. Another,called `system
identification', seeks to refine the model. Unfortunately, the methods
developed in these two areas cannot be bolted together.
This project will examine how this can be remedied in order to produce
design methods that are more widely applicable to
difficult control problems. |
|