Robert H. Kewley

Computational military tactical planning system

A computational system called fuzzy-genetic decision optimization combines two soft computing methods, genetic optimization and fuzzy ordinal preference, and a traditional hard computing method, stochastic system simulation, to tackle the difficult task of generating battle plans for military tactical forces. Planning for a tactical military battle is a complex, high-dimensional task which often bedevils experienced professionals. In fuzzy-genetic decision optimization, the military commander enters his battle outcome preferences into a user interface to generate a fuzzy ordinal preference model that scores his preference for any battle outcome. A genetic algorithm iteratively generates populations of battle plans for evaluation in a stochastic combat simulation. The fuzzy preference model converts the simulation results into a fitness value for each population member, allowing the genetic algorithm to generate the next population. Evolution continues until the system produces a final population of high-performance plans which achieve the commanders intent for the mission. Analysis of experimental results shows that co-evolution of friendly and enemy plans by competing genetic algorithms improves the performance of the planning system. If allowed to evolve long enough, the plans produced by automated algorithms had a significantly higher mean performance than those generated by experienced military experts.

Federated simulations for systems of systems integration

Systems of systems integration is a difficult engineering challenge that places a particular burden on the engineers who must develop simulation models to support that integration. Developing a large scale stand-alone model to support systems integration is a time-consuming process that is often not possible. An alternative approach is to leverage existing models in a federation. This type of work requires a specialized set of engineering skills. The United States Military Academy Department of Systems Engineering SysHub research program is better defining these skills and applying them to different problem domains. This paper highlights how capabilities for information exchange, environmental representation, entity representation, model development, and data collection support the federation development process.

Data Mining for Molecules with 2-D Neural Network Sensitivity Analysis

This paper illustrates a data mining application using two-dimensional (2-D) neural network sensitivity analysis for gaining insight into data strip mining problems. Data strip mining refers to predictive data mining problems where there are a large number of descriptive features, and the number of features is on the order of or exceeds the number of data records (e.g., 100 to 1000 features for 50 to 300 data records). After reducing the number of descriptive features to a manageable set using 1-D neural network sensitivity analysis (e.g., 40 features), a 2-D neural network sensitivity analysis allows the user to visualize variations in the response to identify relevant combinations of features. Each relevant combination can then be analyzed independently to look for interesting patterns and relationships, and can be used in this way to either prune more features or to get insight into the underlying rules for the model. 2-D sensitivity analysis enables the exploration of relevant relationships and features resulting in more robust, meaningful, and efficient models. This methodology was applied to an in-silico drug design problem with 64 molecules and 160 d escriptive features.

WholeSoldier Performance Appraisal to Support Mentoring and Personnel Decisions

We present a multiattribute model called WholeSoldier Performance that measures the performance of junior enlisted soldiers in the U.S. Army; currently there is no formal performance appraisal system in place. The application is unique to decision analysis in that we utilize a common constructed scale and single-dimensional value function for all attributes to match the natural framework of model users and based on operability concerns. Additionally, we discuss model validation in both the terms of decision analysis and psychometrics in models that are used for repeated or routine assessments and thus generate significant quantities of data. We highlight visualization of data for use to support mentoring and personnel decisions to better train, assign, retain, promote, and separate current personnel. Last, we address common cultural concerns related to performance appraisals in organizations by offering a method to standardize ratings and hold raters accountable for their responsibility to mentor subordinates as well as identify their performance to the larger organization.

A systems engineering process supporting the development of operational requirements driven federations

This paper proposes a systems engineering process utilizing the conceptual artifacts of the Model Driven Architecture (MDA) describing platform independent views of models to capture operational requirements, to derive essential tasks, and to combine these tasks into scenarios and vignettes with attributed metrics. This model-independent mission description is then used to identify supporting simulation services that implement the identified military means and capabilities to perform the tasks in the given context. Once the services are identified, the necessary simulation middleware to federate the services is identified and the interfaces are configured using the technical artifacts of the MDA describing platform specific views of systems. This systems engineering process provided support for simulation development for the US Army¿s Program Executive Office - Soldier.

Agent-based model of Auftragstaktik : self organization in command and control of future combat forces

Agent-based modeling is a framework that allows the analysis of distributed command-by-influence using mission-type orders known for over a century as Auftragstaktik in German Army manuals. A combat simulation with embedded decision agents analyzed this type of decentralized command and control. Local commanders relied on their improved situation awareness, mission goals, and constraints from higher commander to drive a small set of robust decision methods. In this sense, they self-organized, and an effective set of actions for mission accomplishment emerged. The improvement was most dramatic for more capable future combat forces. Agent-based modeling provides a laboratory for experiments in command and control of future combat forces.

Further exploration in primitives of meaning

Semantic mismatch between systems is due, in part, to the grouping together of terms who have representations of meaning in different levels of granularity (from different perspectives), and which are composed together into distinct groupings by their own systems of origin. It has been proposed that by referencing these representations when decomposed into elemental concepts (referred to here as primitives of meaning) can assist in interoperability. Insofar as systems describing similar referents are operating within a similar community, the underlying concepts are near-universal (within that community) enough to accommodate this. By decomposing a systems groups of composed terms into primitives of meaning, the building blocks that can be reassembled into the compositions required by another group (of another system, for instance) can be made apparent. While such a de-composition could serve as the basis for an interoperability enabler, having the decomposition available as a common descriptor to highlight areas of semantic misalignment should prove in itself useful. Taking doctrinal statements for US small unit infantry actions as one semantic system, we show how the elemental ideas that are grouped together into commands can be identified and isolated for reconstruction into other groupings. This is the first research step towards relying on primitives of meaning for interoperability.

Co-Adaptive Behavior Algorithms for Insurgent and Counter-Insurgent Techniques in Combat Simulations

This paper proposes an overall strategy and technical approach that allows development of sophisticated behavior algorithms which model insurgent and counter-insurgent techniques in military combat simulations. This capability will allow Department of Defense force providers to estimate, in a combat model, the effects of different approaches to counter-insurgency operations, reconstruction operations, stability operations, and information operations. The overall approach proposes development of robust application programming interfaces that can be employed in several different combat simulations. In a phased strategy, technical development will first focus on task-level behaviors. It will then shift to short-term planning behaviors and long-term planning behaviors. These planning behaviors will make significant use of geospatial reasoning and course of action analysis via simulation. It will also consider civilian behaviors as they relate to supporting either the insurgents or the government security forces. Armed with these simulation capabilities, the military services will be able to make better decisions about the proper mix of forces, equipment, and tactics to provide for use in the different theaters of operation.