Expression Of Interest

Expression Of Interest Modelling Complex Warfighting Strategic Research Investment Joint & Operations Analysis Division, DST Points of Contact: Management and Administration: Annette McLeod and Ansonne Yap Defence Science Institute Technical: Dr Cayt Rowe Defence Science and Technology Group T: 02 6128 7338 E: cayt.rowe@dst.defence.gov.au Initial contact: MCWSRI@defencescienceinstitute.com 03 83441402 Introduction: The Defence Science and Technology (DST) Group and Defence Science Institute (DSI) are partnering to identify research experience in fields that are relevant to Modelling Complex Warfighting including mathematical, computer, and human behaviour sciences. The identified centres of expertise from Australian academic sector may find opportunities for long-term engagements in future research activities and partnerships with the DST Group. Joint and Operations Analysis Division undertakes rigorous, scientifically-based analysis of Defence operations and capability to provide independent, impartial, timely advice. Our Mission is to develop and employ trusted analytical methods and decision support tools that give Defence and national security decision superiority across all aspects of force design, operational planning, command and control, and support to Australian Defence Force on operations. Together, we aim to be Australia s most trusted and influential source of evidence-based analysis shaping and enhancing Defence and national security decision making across the capability lifecycle. Background: The Modelling Complex Warfighting (MCW) Strategic Research Investment (SRI), led by the Joint and Operations Analysis Division, expressly seeks to revolutionise how we undertake operations analysis in DST Group to better handle the interaction of complex geopolitical, social, technological, economic and cultural factors for design of the future force. Force design is the planning and 1

decision-making process regarding military equipment and future defence operations. Such decisions are often made under conditions of high uncertainty. The MCW SRI aims to address the force design and future defence force employment problems under four broad research themes: Conquering Uncertainty, Innovative Simulations, Knowledge Synthesis, and Modelling Complexity. The MCW SRI is a five-year research initiative and is one of a portfolio of strategic research programs sponsored by the Chief Defence Scientist. This initiation has only just begun and you have a chance to influence it. At present there are seven active research areas in this initiative: Machine-discovered behaviour, Simulation-based concept exploration, Modelling complex human systems, Capability decision evaluation under uncertainty, Concepts for complexity-enabled warfare, Force design data culture, and Modelling unknowns. Areas of interest: Schematic of force design timeline and MCW SRI Research themes Academic research programs in the following three focus areas are being investigated: I. Simulation and Data Visualisation 1. Deep learning for real time strategy games. Forces on force adversarial scenarios are essentially a real-time strategy game. Recent work by Google DeepMind and others are 2

applying deep learning techniques to these types of games. It is critical that the SRI has an in depth understanding of this space as it is directly relevant to complex joint scenarios. 2. Generative adversarial networks. Investigate the use of Generative Adversarial Networks (GANs) to automatically generate tactical behaviour for the multiagent simulation of military operations. GANs have had significant success in automatically generating content that is indistinguishable from content that is human generated. This research is concerned with developing techniques for generating creative, novel and innovative tactical behaviour for individual agents, teams of agents and forces (teams of teams) of agents in adversarial games and military simulations. 3. Simulation infrastructure as a service. Develop concept for a cloud based infrastructure to support Data Farming. Defence Science & Technology s researchers develop stochastic simulations to model the dynamic interactions between multiple military entities involved in sea, air and/or land operations. The properties and behaviour of these entities are controlled by numerous multivalued parameters provided as inputs to a batch of execution runs. To efficiently explore the response space of the simulation to gain new insights the technique of Data Farming is used. The project consists of two related infrastructure aspects. The first is the development of a software framework for managing data and orchestrating the steps involved in Data Farming, from robust definition and design of experiments to parallel execution of simulations to analysis and visualisation of results. The second aspect is the design of a distributed computing architecture to support parallel stochastic simulation and management of data products for these problems, using Defence Science & Technology s and/or hybrid secure remote infrastructure. 4. Approaches for data visualisation and analysis of simulation experiments. Data farming for analysing future operating concepts is expected to produce large volumes of high dimensional data. Development of approaches and techniques for supporting exploratory analysis is required to make sense of relationships between variables in simulation experiments. Visualisations will be used to optimise the speed of exploring the data as well as to convey understanding. It is envisaged that a suite of techniques will be developed to account for the strength and limitations of any one particular technique and the need to support a range of simulation models and associated analysis approaches. 5. Explore application of Bayesian optimisation for Bayesian network (BN) design and structure learning from data. Bayesian network meta-model has been used for causal and what-if analysis for simulation. The number of parameters and data input/output from complex warfighting simulation will be very large. It is desirable to explore the use of Bayesian Network structure learning techniques to construct the BN meta-model from the experts and large simulation data, and applying Bayesian optimisation and/or heuristics algorithm techniques for optimising force design options. The expected scientific output is to demonstrate the applicability of Bayesian optimisation and/or heuristic algorithm techniques in fine tuning the parameters in high-level force structure analysis via Bayesian network model. 6. Uncertainty Quantification techniques options for complex, discrete time modelling and simulation. The Australian Defence Force is a large, complex system of systems that 3

functions at different levels and across different levels, at different times, in different places, with different partners, against different adversaries, in different social contexts. Understanding complexity, the system behaviours that result, and the interdependence between different systems is an important problem for decision makers. Behaviour that could be deemed to be undesirable at one level (e.g. disordered, inefficient) may be necessary for desirable outcomes (e.g. effectiveness) at another level. Our research aims to provide methods for identifying, characterising and quantifying the uncertainty inherent in complex modelling and simulation of these systems of systems, and how these unknowns can be effectively communicated with Force Design decision-makers. 7. Taming complexity and emergent behaviour in simulated force-level tactical conflict. Using constructive modelling and simulation, investigate the effects of scale and complexity on the observable behaviours of the systems-of-systems force elements (agents) that occur within a tactical conflict. Explore, with the aim of developing an understanding, the conflict scenario attributes (environment, information, interaction, mission objective and agent-level behaviour) that enable or cause emergent behaviour at the whole-of-system level. Understand how to identify the inherent and emergent system behaviours that occur in complex closed-loop simulations and translate these observations and measurements into meaningful assessment outcomes for principal stakeholders and decision makers. II. Social Sciences and Complex Systems 8. Explore the use of advanced international relations (IR) methodologies for analysis of the trends of the future operating environment and their implications to force design. Demonstrate the applicability and utility of the IR techniques to analysis of the interplay and intersubjectivity of material and non-material factors that have the potential for a fundamental shift in the strategic landscape, informing gaps in force design. 9. Complex systems design methods in Defence. Co-develop approach and application of holistic analytical method/s to design and develop Strategy, Concepts and Structures for the ADF. Linking the theories of Systems Thinking with Human Centred Design to explore novel and creative options for force design, such as; organisational, management and business processes, design thinking, systemic design, to both improve understanding of the strategic and operational contexts and incorporate bottom up innovative design. 10. Novel methods and tools to generate plausible future situations (conceptual futures) that the Australian Defence Force may encounter which will influence shape and inform force development. Demonstrate alternative methods that enhance defining plausible future situations. III. Decision Making under Uncertainty 11. Novel methodologies for ranking force design options with high level of uncertainty and interdependencies between criteria. Force design and complex warfighting modelling characterised by large numbers of variables, parameters and inter-dependence between context, environments, systems and sub-systems in which the performance and effectiveness data is scarce, uncertain, qualitative and/or quantitative. There is a need to 4

Activities: develop a consistent approach for combining different data types and difference model results, and combining qualitative and quantitative modelling methodology to evaluate and to rank force design options. Bayesian Network (BN) and Multi-criteria Decision Analysis (MCDA) modelling techniques have been identified as suitable to address this problem. The expected scientific output is a newly developed methodology and toolset for evaluating and ranking the force design options via BN and MCDA techniques. 12. Applications of project benefit management approach and portfolio selection methods in the context of force design. Demonstrate the utility of project benefit management approach in articulating the relative value of Defence investment in different Integrated Investment Plan projects, and portfolio selection methods in creating a balanced Integrated Investment Plan to meet Government strategic objectives. 13. Probability theory of avalanche processes on networks. Theoretical probability analysis and large-scale simulation of adaptation of avalanche models as models of spontaneous information flows through a combat-engaged force triggering coherent behaviour. 14. Frameworks for understanding human decision-making within environments that have unknowns, high stakes and high risk due to uncertainty. This topic is interested in the potential of a range of techniques to address this area including mathematical modelling, field studies, clinical practice and laboratory experimental research (e.g. microworld simulations and elementary cognitive tasks). 15. Approaches for eliciting, understanding, modelling and managing unknowns. Unknowns play an intrinsic role in force design decision-making. Force design decisions have long lifespans, outlasting even the careers of their decision makers, and have far-reaching implications for future force capability and future capability decisions. Decisions are made in light of a threat, social, political, and technological environment that won t exist for another 20 or 30 years, and may not be predicted or imagined. And these decisions will ultimately influence not just Defence capability, but society as well, in terms of opportunity costs and impacts on health, education, and the economy in general. What approaches can be developed to elicit, understand, model and manage the unknowns faced in force design decision-making to ensure the best Defence outcomes for the people of Australia? Phase 1: Initial call for proposals. Interested organisations will be encouraged to identify relevant fields of endeavour or expertise where they would be willing to engage and partner with DST Group s Joint and Operations Analysis Division on Modelling Complex Warfighting. Organisations are requested to submit short organisational profiles including relevant experience and proposed approaches for addressing the questions of interest. The intent of this call is to enable short listing for Phase 2. No funds will be exchanged for this stage. The call for expressions of interest will commence for this phase on 30 October 2017 and be due on 17 November 2017. Phase 2: Partnering opportunity. Depending on the outcomes of Phase 1, some organisations may choose to engage in partnering relationships through long-term collaborative agreements covering research into various aspects of Modelling Complex Warfighting. DST Group will continue to play a 5

central role in this relationship, but academic organisations may wish to form their own connections. Exploration of partnering opportunities will commence at the Modelling Complex Warfighting Symposium to be held in Melbourne on 14-15 th December 2017. Schedule: 30 October 2017 - Phase 1 EOI released 17 November 2017 - Phase 1 EOI returns due 1 December 2017 Selected partners are notified and invited to attend the Modelling Complex Warfighting Symposium 14-15 December 2017 - Modelling Complex Warfighting Symposium (at own cost) December 2017 Development of research agreement (contracts up to $100K for current financial year) 30 May 2018 Research Agreement Program completed August 2018 Potential open call for the development of 3-year partnerships agreements Submitting Expressions of Interest: Responses to this call for Expressions of Interest are to be made via MCWSRI@defencescienceinstitute.com Submissions must be received before 5:00pm (EST) on 17 November 2017 6