UWA Computer Science - Research in the Department

    home > research > areas > formal specification    

MODELLING COMPLEX SYSTEMS

Research Areas

Our research is focussed on the study and increased understanding of naturally occurring complex systems, particularly spatial systems such as traffic networks, bushfire dynamics and spread of infections Intended outcomes are models and simulation technology allowing us to predict future behaviour of such systems: in real-time, allow better management of evolving systems such as traffic and fire; non real-time allows better design of road systems to minimise conjestion and urban spaces for safe egress, and development of fire and infection control strategies.

  • Computational Epidemiology : Pandemic Influenza

    One of the research group's projects is the computational modelling of the spread of infectious respiratory diseases such as influenza. Of particular interest are strains of "pandemic" influenza such as the "Swine flu" strain of A(H1N1) influenza that originated in Mexico in 2009. Highly infectious in nature, these strains have the potential to infect a large proportion of the world population.

    This research programme utilises extant mathematical modelling theory which has had extensive application to the dynamics of other complex systems and brings together researchers with skills in epidemiology, virology, modelling and software development. We have constructed a model to capture the dynamics of pandemic influenza in an Australian urban context with population mobility patterns extracted from ABS and other data sets. The simulation model has been used experimentally, each experiment being a separate simulation run, dealing with a particular scenario which "plays out" as time passes.

    Work to date has concentrated on modelling the effectiveness of non-pharmaceutical "social distancing" interventions. More recent work has involved extending the model to include vaccination.

    In addition to human infectious diseases, we have also investigated the application of spatial computational modelling techniques to animal diseases, such as the spread of classic swine fever among herds of feral pigs in northern Australia.

    Publications

    Milne GJ, Kelso JK, Kelly HA. Strategies for mitigating an influenza pandemic with pre-pandemic H5N1 vaccines.  Journal of the Royal Society Interface, September 15, 2009, doi:10.1098/rsif.2009.0312

    Joel K. Kelso, George J. Milne, Heath Kelly (2009) Simulation suggests that rapid activation of social distancing can arrest epidemic development due to a novel strain of influenza. BMC Public Health 9:117
    doi:10.1186/1471-2458-9-117

    George J. Milne, Joel K. Kelso, Heath A. Kelly, Simon T. Huband, Jodie McVernon (2008) A Small Community Model for the Transmission of Infectious Diseases: Comparison of School Closure as an Intervention in Individual-Based Models of an Influenza Pandemic. PLoS ONE 3(12):e4005.
    doi:10.1371/journal.pone.0004005

    Milne G, Fermanis C, Johnston P “A mobility model for classical swine fever in feral pig populations”, Veterinary Research, 39(6), 53, 2008.

    Johnston P, Kelso J, and Milne G “Efficient simulation of wildfire spread on an irregular grid”, International Journal of Wildland Fire, 17, 614-627, 2008.

     

    Related Research


  • Crowd Modelling

    Modelling Emergent Crowd Behaviour
    Using a cellular automata approach we create models of people movement within buildings. Simple, local update rules capturing the movement of individuals are shown to produce realistic behaviour of crowds, that is, collections of individuals. We demonstrate how distinct crowd behaviour at constriction points is characterised using diㄦent update rules. These distinct rules are produced in an experimental manner; we utilise a simulation environment to examine various crowd movement scenarios with the resulting crown dynamics being revealed by graphical animation.

    Crowd Dynamics

Projects

  • Simulation, modelling and animation tools for real-time prediction of fire development
    Modelling bushfire spread in a meaningful way improves decision-making when it comes to critical scenarios. This project will develop reliable bushfire spread modelling, simulation and animation technology to underpin and support a wide range of fire management activities, including risk analysis, prescribed burning, wildfire suppression and incident control training.
    A computer-based environment will permit rapid and repeatable execution of bushfire simulations under a wide range of conditions, assisting with real-time decisions and "what if" scenarios. Simulations will be based on the latest understanding of physical fire behaviour captured within discrete automata-based models. These simulations are key to improving predictions of fire behaviour and the effectiveness of containment strategies. They also increase our understanding of the non-linear scaling found in extreme fire behaviour, as seen in Canberra in January 2003.
    The overall outcome will be a computer-based simulation environment for training purposes utilising a high-performance visual display and readily useable interface. This toolset will allow users to understand fire behaviour and how it can be affected by changes to the fire environment and firefighter safety and suppression options. Trainees will experience and explore, in virtual reality, the effect of different containment strategies via repeated simulation under differing conditions with readily visible consequences. The resulting cause - and - effect relationships will increase understanding of how the positioning of fire-breaks and deployment of fire suppression resources can alter fire spread under various weather, fuel and topographical conditions.
    The developed system will present the sequence of conditions and events leading to worst-case scenarios and directly influence safest and best containment strategies.
    This project is part of the Bushfire CRC.

    Johnston P, Kelso J, and Milne G “Efficient simulation of wildfire spread on an irregular grid”, International Journal of Wildland Fire, 17, 614-627, 2008.

    PhD Scholarships are available in this area, contact Professor George Milne .  

     

  • NHMRC Project: Modelling of Alternative Control Strategies for Pandemic Influenza
    Current research will develop a spatial simulation model of the dynamics of pandemic influenza within Australia. The resulting program will be of product quality and be readily usable by public health epidemiologists to predict the effect of alternative containment measures on the scale, rate and location of disease spread, through a city, state or the nation. Deployed in .real time. after infection has started in Australia, it would be used to predict infection spread and the containment effect of response measures. In pre-pandemic period the simulation model will be available to predict the containment effect of a range of response measures, such as travel restrictions, social distancing, vaccination and antiviral usage. Our project will specifically apply the simulations model to determine optimal use of limited resources such as targeting of antiviral drugs and initial supplies of vaccine.
    The simulation model allows spatial and temporal features which contribute to the dynamics of disease spread to be captured. Effective containment strategies have a spatial component, eg. where to target antiviral drug therapy and vaccination.
    This research programme utilises extant mathematical modelling theory which has had extensive application to the dynamics of other complex, spatial systems. Prototype spatial simulation software is currently under development developing a set of realistic pandemic scenarios which will simulate outcomes for different nominated values of model parameters and brings together researchers with the necessary blend of skills in epidemiology, virology, modelling and software development.
    We are constructing a model to capture the dynamics of pandemic influenza in an Australian urban context with population mobility patterns extracted from ABS and other data sets. Once constructed the simulation model will be used experimentally, each experiment being a separate simulation run, dealing with a particular scenario which .plays out. as time passes. For example, starting with an index case arriving at an Australian international airport separate simulations maybe used to highlight the spatial spread of influenza with an without geographically target antiviral prophylaxis.
    Other diseases modelled include Foot and Mouth Disease (FMD) and Classical Swine Fever (CSF). The FMD research is part of an international research project funded by the Danish Food and Veterinary Research Institute (DFVF), utilising spatial simulation models to develop FMD vaccination strategies for Denmark.
    Three postdoc researchers are involved in disease modelling, in addition to Professor Milne.

     

  • Modelling and simulation of disease spread dynamics using interacting automata
    The aims of this project are to:
    • verify the suitability of interacting automata as the mathematical foundation for modelling the spatial dynamics of disease spread;
    • discover techniques for capturing geographic features which influence disease spread utilising event-base state transitions;
    • produce and event-based simulator core together with a technique to rapidly instantiate or personalise this core to create a disease-specific simulation engine;
    • create a simulation environment which permits disease managers to have ready access to the simulator via an appropriate user interface and communicates the result of the simulation runs via graphical display.

    ARC Linkage Project Grant, in conjunction with WA Department of Agriculture.

     

  • Epidemic Spread Simulation with Cellular Automata
    Virus and other disease outbreaks have been attracting more and more media attention. Apart from the obvious health effects of virus infections, the economic impact can be substantial is illustrated by the foot-and-mouth disease (FMD) outbreak in Britain in 2000. In order to better understand, predict and eventually control epidemic spread, accurate models need to be devised and analysed.
    Current epidemiological techniques are dominated by statistical and deterministic models that do not take into account spatial variables. Considering that the landscapes in nature are far from homogeneous and uniform, it makes sense to incorporate spatial parameters and produce "better" models. A particular paradigm that is conducive to modelling space is cellular automata (CA). This project focusses on investigating the applicability of CA to modelling epidemic spread.
    Shih Ching pages with images and demonstrations
    Research by Shih Ching Fu and George Milne

Staff

PhD Students

  • Bobby Chu
  • Nilimesh Halder

Other Publications

Milne GJ, Kelso JK, Kelly HA. Strategies for mitigating an influenza pandemic with pre-pandemic H5N1 vaccines.  Journal of the Royal Society Interface, September 15, 2009, doi:10.1098/rsif.2009.0312

Joel K. Kelso, George J. Milne, Heath Kelly (2009) Simulation suggests that rapid activation of social distancing can arrest epidemic development due to a novel strain of influenza. BMC Public Health 9:117
doi:10.1186/1471-2458-9-117

George J. Milne, Joel K. Kelso, Heath A. Kelly, Simon T. Huband, Jodie McVernon (2008) A Small Community Model for the Transmission of Infectious Diseases: Comparison of School Closure as an Intervention in Individual-Based Models of an Influenza Pandemic. PLoS ONE 3(12):e4005.
doi:10.1371/journal.pone.0004005

Milne G, Fermanis C, Johnston P “A mobility model for classical swine fever in feral pig populations”, Veterinary Research, 39(6), 53, 2008.

Johnston P, Kelso J, and Milne G “Efficient simulation of wildfire spread on an irregular grid”, International Journal of Wildland Fire, 17, 614-627, 2008.

"A heat transfer simulation model for wildfire spread" P Johnston, G Milne, J Kelso Forest Ecology and Management, 234 Supplement 1 (2006) S78.

" Modelling Dynamically Changing Hardware Structure", G. Milne. Proceedings of Algebraic Process Calculi 2005. In Electronic Notes in Theoretical Computer Science 162 (2006) pp. 249-254, Elsevier 2006.

" Properties as Processes: Their Specification and Verification" J. Kelso & G. Milne Proceedings of the 25th IFIP WG 6.1 International Conference of Formal Techniques for Networked and Distributed Systems (FORTE 2005), Taipei, Taiwan, October 2-5, 2005. In Lecture Notes in Computer Science 3731, pp. 503-517, Springer, 2005.

" Programming Paradigms for Reconfigurable Computing", G. Lee and George Milne. Microprocessors and Microsystems, Vol. 29, No. 10, pp. 435-450. 2005.

" Property Verification of Asynchronous Systems", Antonio Cerone and George J. Milne. Innovations in Systems and Software Engineering, Vol. 1, No. 1, April 2005, pp. 25-40. Springer-Verlag London Ltd, 2005.

"Modelling Wildfire Dynamics via Interacting Automata", G Milne and A Dunn. Proceedings of the 6th International Conference on Cellular Automata for Research and Industry (ACRI 2004), Amsterdam, The Netherlands, October 25-28, 2004. In Lecture Notes in Computer Science 3305, Editors: Peter M. A. Sloot, Bastien Chopard, and Alfons G. Hoekstra, pp. 395 - 404, Springer, 2004.
pdf (1.2 Mb)

"A Flexible Automata Model for Disease Simulation", G Milne and S.C. Fu. Proceedings of the 6th International Conference on Cellular Automata for Research and Industry (ACRI 2004), Amsterdam, The Netherlands, October 25-28, 2004. In Lecture Notes in Computer Science 3305, Editors: Peter M. A. Sloot, Bastien Chopard, and Alfons G. Hoekstra, pp. 642 - 649, Springer, 2004.

"Epidemic modelling using cellular automata", G Milne and S.C. Fu. In Proceedings of the 1st Australian Conference on Artificial Life (ACAL'03), Canberra, December 2003, pp. 43-57.
pdf (235 kb)

"Modelling emergent crowd behaviour", T.J. Lightfoot and G. Milne. In Proceedings of the 1st Australian Conference on Artificial Life (ACAL'03), Canberra, December 2003, pp. 159-169.
pdf (293 kb)