Friday, October 14, 2016

Evolving Optimal and Diversified Military Operational Plans for Computational Red Teaming

Summary:

This journal article uses multi-objective evolutionary algorithms (MOEAs) on Computational Red Teaming (CRT) to increase diversity of near-optimal alternative strategies in the decision variable space. CRT is the computerization of Red Teaming through the use of agent-based models for application toward a given battlefield. The Red team tries to break the Blue’s defensive strategy, and for their experimentation the scientists used two case study approaches that use diversity-enhancement schemes (DES) to increase the diversification of MOEAs.

1. Covered in the first part of the study the scientists expressed the recent literature of CRT, which is an agent-based system (ABS) that mimics and simulates intricate models of warfare. Similar programs include ISAAC/EINSTein, WISDOM, and MANA. The advantage of using these models the scientists point out is that their low-resolution in their simulation models can produce many various simulations with many properties. That can then be searched by analysts for particular techniques and models that interest them. Included in this section are the plethora of mathematical operations and equations included in MOEAs to carry and develop the results from population-based stochastic search heuristics created from real phenomena in nature.

2. Following the literature review came the methodology where the researchers used an evolutionary framework that included 1) a model generator; 2) a simulation engine that usually generates 30 repetitions for each simulation model; and 3) an evolutionary algorithm that generates results for model specification files. Added to these steps for this research paper was the diversity enhancement scheme (DES) which focuses the evolutionary search focuses, “1) to preserve and promote the non-dominated solutions to exploit Pareto optimal solutions, and 2) similarly, the solutions which contribute to the aggregated, in both decision and objective, space diversity are also preserved/promoted” (Zeng, Decraene, Low, Zhou, & Cai 2012).

3.  By the experimentation point, the scientists did two case studies where Red strikes Blues defensive strategies in an urban ops scenario and a maritime anchorage protection scenario. Included in the measuring of these case studies was the DES and MOEAs to assess similarities and differences in the selected MOEAs of DES, Niching, HypeE, NSGAII, and SPEA2.

4. It was found that DES and Niching were the best MOEAs showing significant indicators of Pareto optimal solutions and solution diversification, with DES taking the lead of the two.

Concluding it all the researchers in their conclusion found that most research conducted using MOEAs has used Pareto-optimality as the focus and neglect the area of diversification of solutions within the decision variable space. Thus limiting choices a decision maker has when applied to real world problems. Through the DES research applied in this research study, the researchers give the ability to diversify the decision space without compromising the Pareto-optimality. Thus allowing more room and information for a decision maker to make a more informed decision and conclusion that is likely to be positive than negative.

Critique:

The usage of red teaming can be a useful modifier when it comes to needing to produce more realistic analysis and give a decision maker a more rounded brief that he can make the best decision for. Yet, in the modern day as this study on using MOEAs on CRT, we face a more technologic based model that is more reliant on technology, particularly when it comes to the cyber domain. Therefore, the application of CRTs gives access to the ability to run an exponential amount of simulations that analysts can use to produce more informed briefs for a decision maker, and speed up the intelligence process. Though this exercise in the research study was more of a War Gaming view, it none the less covers the notion that it is no longer analysts that work on teams or red teams solely. It also puts the computer programs that aid in the ability to produce data to also part take in similar exercises increasing the number of simulations and decreasing the likelihood of a potential estimates failing. Of course, intelligence is still an imperfect profession and art, but the use of red teaming can reduce, particularly through the use of CRT and technology, the uncertainty for a decision maker.

Sources

Evolving Optimal and Diversified Military Operational Plans for Computational Red Teaming

Zeng, F., Decraene, J., Low, M. Y. H., Zhou, S., & Cai, W. (2012). Evolving optimal and diversified military operational plans for computational red teaming. IEEE Systems Journal, 6(3), 499-509. <https://www.researchgate.net/profile/Suiping_Zhou/publication/258655766_Evolving_Optimal_and_Diversified_Military_Operational_Plans_for_Computational_Red_Teaming/links/5536541d0cf268fd00171fd6.pdf>.