EffectsBased Operations as a MultiCriteria Decision Analysis Problem

Effects-Based Operations as a Multi-Criteria Decision Analysis Problem Jouni Pousi, Kai Virtanen, and Raimo P. Hämäläinen Systems Analysis Laboratory Helsinki University of Technology jouni. pousi@hut. fi, kai. virtanen@hut. fi, raimo@hut. fi S ystems Analysis Laboratory Helsinki University of Technology 1

Introduction Effects-based operations (EBO) (e. g. , Davis, P. K. , “Effects-Based Operations: A Grand Challenge for the Analytical Community, ” RAND, 2001) – Concept for planning and executing military operations – Systems perspective - e. g. , country, electrical grid, transportation network – Effects: Consequences of actions in a system § - Effects: e. g. , change opinion of the populace, isolate a country, disable the electrical grid - Actions: e. g. , launch a missile, jam communications, apply economic sanctions EBO planning process allowing the application of multi- criteria decision analysis (MCDA) § Utilization of multi-criteria influence diagrams in the process Transparent and tractable way to plan EBO S ystems Analysis Laboratory Helsinki University of Technology 2

Effects-based operations (EBO) § The term ”EBO” originated during the Gulf War in 1991 § Historically military planning focused on actions § EBO shifts the focus to effects in a system § Military operations aimed for effects prior to EBO – Inconsistent approaches – Lack of analytical methods – Not called EBO § Practical experiences with EBO led to extensive debate on aspects § Multi-criteria approach not focused on in unclassified literature S ystems Analysis Laboratory Helsinki University of Technology 3

Aspects of EBO Actions § Cause effects in a system (e. g. , bomb railways, jam communications) – Through actions – Such that higher level objectives fulfilled § Systems in EBO – Uncertain – Dynamic – High level of interdependency § Multiple effects caused by even a single action ð MCDA is a necessity when planning EBO! System Effects (e. g. , transportation network) (e. g. , isolate a country) Higher-level objectives (e. g. , stabilize a region) S ystems Analysis Laboratory Helsinki University of Technology 4

EBO planning process ? Actions ? Best action System ? Effects ? Decision maker (DM) Goal of the process: Support the decision maker (DM) to find the best action causing the desired effects in the system S ystems Analysis Laboratory Helsinki University of Technology 5

EBO planning process: Step 1/9 Effects Undesired Unidentified Desired Insignificant Higher-level objectives Step 1: Identify effects from higher-level objectives S ystems Analysis Laboratory Helsinki University of Technology 6

EBO planning process: Step 1/9 Effects Ensure cooperation of country D on issue X Undesired Unidentified Desired Insignificant Higher-level objectives Change popular opinion of country D about government of country D Step 1: Identify effects from higher-level objectives S ystems Analysis Laboratory Helsinki University of Technology 7

EBO planning process: Step 2/9 System Effects Subsystem Element Step 2: Define the system consisting of subsystems and elements S ystems Analysis Laboratory Helsinki University of Technology 8

EBO planning process: Step 2/9 Country D Power grid System Effects Subsystem Element Power plant Step 2: Define the system consisting of subsystems and elements Civilian population S ystems Analysis Laboratory Helsinki University of Technology 9

EBO planning process: Step 3/9 System Effects Subsystem related to effect Step 3: Identify subsystems related to each desired and undesired effect S ystems Analysis Laboratory Helsinki University of Technology 10

EBO planning process: Step 3/9 System Civilian population Effects Change popular opinion of country D about government of country D Subsystem related to effect Step 3: Identify subsystems related to each desired and undesired effect S ystems Analysis Laboratory Helsinki University of Technology 11

EBO planning process: Step 4/9 Actions System Effects Feasible actions Step 4: Identify feasible actions having an impact on subsystems related to each effect S ystems Analysis Laboratory Helsinki University of Technology 12

EBO planning process: Step 4/9 Actions Effects System Bomb electrical power plants Feasible actions Civilian population Powerplant Step 4: Identify feasible actions having an impact on subsystems related to each effect S ystems Analysis Laboratory Helsinki University of Technology 13

EBO planning process: Step 5/9 Actions System Effects Other effects Step 5: Identify other effects caused by the feasible actions S ystems Analysis Laboratory Helsinki University of Technology 14

EBO planning process: Step 5/9 Actions Effects System Bomb electrical power plants Step 5: Identify other effects caused by the feasible actions Undesired effect: Infrastructure degradation Other effects S ystems Analysis Laboratory Helsinki University of Technology 15

EBO planning process: Step 6/9 Actions System Effects Indicator Step 6: Choose indicators for all the effects S ystems Analysis Laboratory Helsinki University of Technology 16

EBO planning process: Step 6/9 Actions System Effects Indicator Powerplant Undesired effect: Infrastructure degradation Step 6: Choose indicators for all the effects S ystems Analysis Laboratory Helsinki University of Technology 17

EBO planning process: Step 7/9 Actions Effects System Step 7: Construct criteria for effects from the indicators Criterion S ystems Analysis Laboratory Helsinki University of Technology Criteria 18

EBO planning process: Step 7/9 Actions Effects System Undesired effect: Infrastructure degradation Reconstruction costs Step 7: Construct criteria for effects from the indicators S ystems Analysis Laboratory Helsinki University of Technology Criterion Criteria 19

EBO planning process: Step 8/9 Actions Feasible actions System Effects Efficient actions Step 8: Find efficient actions based on the criteria S ystems Analysis Laboratory Helsinki University of Technology Criteria 20

EBO planning process: Step 9/9 Actions DM’s preference information System Effects Multi-criteria evaluation of efficient actions Best action 9: Select the best action based on DM’s preference SStep ystems Analysis Laboratory information Helsinki University of Technology Criteria 21

Summary of EBO planning process II. Feasible actions identified based on their impact on the system System I. Effects connected with the system • Subsystems • Elements • Indicators Actions Effects III. Criteria for effects constructed from indicators Multi-criteria evaluation of efficient actions Criteria Best action IV. Best action selected based on DM’s preference information S ystems Analysis Laboratory Helsinki University of Technology 22

EBO planning process: Interpretation as a multi-criteria decision problem EBO planning Decision analysis Define system and effects Structure the decision problem Identify feasible actions and other effects Assess impact of alternatives Choose indicators, construct criteria Determine DM’s preferences Find efficient actions and the best action Evaluate and compare decision alternatives S ystems Analysis Laboratory Helsinki University of Technology 23

EBO planning process: Interpretation as a multi-criteria decision problem Decision analysis EBO planning Define system and effects Identify feasible actions and other effects Choose indicators, construct criteria Find efficient actions and the best action Structure the decision problem MCDA methods can be applied for supporting EBO planning! Assess impact of alternatives Determine DM’s preferences Evaluate and compare decision alternatives S ystems Analysis Laboratory Helsinki University of Technology 24

Requirements for methods applied in EBO planning process § System modeling – Interdependencies – Uncertainty – Dynamics § Multi-criteria evaluation of a discrete set of actions Multi-criteria influence diagrams (MCIDs) - a potential method! – MCDA – Sensitivity analysis S ystems Analysis Laboratory Helsinki University of Technology 25

EBOLATOR § Decision support tool for EBO planning – Ge. NIe for MCID – Matlab-based GUI § Use of Ge. NIe – Construction of MCID graphically – Definition of MCID parameters – Calculation of criteria probability distributions § GUI – Visualization of actions – Evaluation of actions • Decision rule for criteria (e. g. , expected value) • Direct weighting • Constraints – Sensitivity analysis with respect to • Weights • Constraints • MCID parameters S ystems Analysis Laboratory Helsinki University of Technology 26

Example mission Defensive mission: Determining aircraft positioning and air combat tactics against air-to-ground Higher level objective: Defend the sovereignity of the country § Attack towards civil and military infrastructure – Power plants, electrical substations, water supply, radar sites, anti-air sites § Attacker assumed to conduct two air-to-ground operations § Uncertainties on mission planning – Enemy actions – Weather – System behaviour S ystems Analysis Laboratory Helsinki University of Technology 27

Results of EBO planning process Actions System - Aircraft positions Infrastructure (46 MCID nodes, 4404 parameters) - Air combat tactics (729 actions) Effects - Maintain functioning of civil infrastructure - Maintain functioning of air defense Criteria Multi-criteria evaluation of actions - Air combat losses - Functioning of air surveillance - Functioning of power grid - etc. (10 criteria) S ystems Analysis Laboratory Helsinki University of Technology 28

EBO planning process: Multi-criteria evaluation of actions § The best action according to the authors’ preferences – Protect power plants – Less aggressive air combat tactics during the second attack operation § Sensitivity analyses for best action – Robust with regards the weigths and constraints of the criteria – Can be executed in all weather conditions S ystems Analysis Laboratory Helsinki University of Technology 29

Conclusions § Multi-criteria perspective essential in EBO planning – Previous elaborations could be improved by inclusion of MCDA § New EBO planning process utilizing multi-criteria decision analysis § EBOLATOR using influence diagram methodology – Answers challenges of system modeling – Enables multi-criteria evaluation - Sensitivity analyses – Large system => Issues on model size and determination of parameters § Future research – – Alternative MCDA methods Enemy behaviour – game theory Improved dynamics Other application areas – medical, marketing S ystems Analysis Laboratory Helsinki University of Technology 30