Sea Control - Sea Control 260 - The Best Defense is a Good Offense with ENS Kara Dowling

Episode Date: June 20, 2021

Links1. "Cost-Effectiveness of Offense as a Ship Self-Defense Tactic," by ENS Kara Dowling, USNI Proceedings, April 2021.2. "A Multiagent Stochastic Control Model for Adversarial Planning in Naval Operations: Cost Effectiveness of the Offense as a Ship Self-Defense Tactic," ENS Kara Dowling, Phalanx 53, no. 3 (2020): 34-39.

Hey folks, Jared here. I've talked about our feed. I'm going to keep talking about it. There are two existing Seek Control feeds out there. So the one you want to be listening to is labeled simply Seek Control and includes the phrase Simsex Flagship Podcast in its description. We are working on a new logo and once that's finished, we'll have a better differentiator for you. But for right now, it's Seek Control Simsex Flagship Podcast. That's where you want to be. So please subscribe, rate, and review the new feed. Today I'm joined, this is going to be a shorter episode, but I'm joined by Ensign Cara Dowling.

She did her undergraduate work at Princeton, following in the footsteps of the late Captain Wayne Hughes, completed an analysis of missile combat for her thesis. So new voice. I think we're going to be hearing a lot from her. This episode was edited and produced by Taylor Fairless. Don't miss everything going on over at the website, simsec.org. Final call for articles for Project Trident just went live. We're partnering with Lockheed Martin and the Naval Warfare Studies Institute at the

Naval Postgraduate School for Emerging Technologies. We've also partnered with Transcom for a series of articles on strategic sea lift. We're eagerly anticipating that we had originally planned a week for that starting next week, the Monday after this airs, but it's actually going to go a full two weeks based on the quality and the quantity of the articles we've received. And there'll be an accompanying podcast with we'll just call them a mystery

guest for now to follow in the few weeks after that. So that's all over at simsec.org. Finally, I want to take the opportunity to recommend our partners in the SimSec Podcast Network, The Bilge Pumps. You can find Alex, Jamie, Drack, and a pile of Iron Brew bottles wherever you download your podcasts. On that note, I'll turn it over to Kimbersman. You're listening to Sea Control, hosted by the Center for International Maritime Security. Aloha, shipmates, and welcome back aboard Sea Control. Today, my guest is Kara Dowling, author of the recent proceedings article,

Cost Effectiveness of Offense as a Ship Self-Defense Tactic. Kara, welcome. As a reminder, all views expressed today are our own and not reflective of any organization with which we are otherwise associated. Kara, would you mind starting by telling us a little bit about yourself, please? Aloha, I'm Kara Dowling, a 2020 graduate of Princeton University and a newly commissioned ensign in the Navy, currently stationed in Japan. Well, thanks so much. Carrie, you've written for Phalanx already, Proceedings, and then this was also the subject of your undergraduate thesis at Princeton. So what was the genesis for your interest in this topic?

Well, actually, everyone at Princeton has to write a senior thesis in order to graduate. My major was operations research and financial engineering. And since I knew that operations research came about because of the military in the World War II era, I wanted to connect my degree with my military interests. So my executive officer at my NROTC unit actually connected me to NPS and I was connected with Captain Klein there. He has a lot of background in operations research and as a naval officer, he hypothesized that using offensive measures is more cost effective than traditional defensive measures as a method of sustained lethality. My Princeton thesis advisor and I, Professor Powell, agreed that this will be fascinating and a challenging

topic and thus began the research process to model a surface-to-surface missile engagement and prove the hypothesis. So your study obviously calls to mind the work of Wayne Hughes in the various versions of fleet tactics and what I think is the most important cornerstone, fire effectively first. If it's not the most important, it's at least the most fun. But are you deliberately setting out to test whether Hughes' assertion is correct? And then what do you think that you learned about naval warfare as a result of the study? The hypothesis that I was trying to test was whether offensive measures were more cost-effective than traditional defensive measures as a method of sustained lethality, which is directly related to Captain Hughes's cornerstone

saying fire effectively first. One of my main sources for the research was actually the book Fleet Tactics and Hughes's salvo model. My aim was to build upon his work and test the theory. When I spent a week at the Naval Post Graduate School, I had the opportunity to meet with Captain Hughes and discuss my research. He helped me edit my model and gave me some feedback. So it was a true honor to meet him. And I'm really fortunate that he was a part of this. Yeah, that is an incredible opportunity. And I didn't know that part of the story. I communicated with him a little bit while I was stationed over in Germany teaching over there. And he said he was no longer traveling. So it's incredible that you got to meet him and that he

actually provide feedback on your work. So what were some of the assumptions necessary to complete your study? Because you mentioned in more than a few places, and you and I both know, naval warfare is very complex. There are only two actors in the model, the blue ship and the red ship. Each actor has offensive missiles, defensive missiles, and point defensive measures. So evolved sea sparrow missiles, sea rolling airframe missiles, and a close-in weapon system. Each of the missiles has generic specs and limitations. For instance, the C-RAMs were launched at incoming enemy offensive missiles between one and three nautical miles from the ship. Each missile or weapon has a probability of success that determines if its purpose, offensive or defensive, was successful in that instance.

So if you're launching an offensive missile, did it actually take out the other ship if it gets there? We assume that each ship costs $2 billion and all the other weapons also have their own rough but set costs. The missiles fly at 0.8 Mach. The ships do not start in the engagement more than 300 nautical miles apart. Also, the blue and red have access to radar, satellites, electronic surveillance equipment, past acoustic sensors, unmanned aerial vehicles, unmanned surface vehicles for tracking and diverting of incoming missiles. Every time there's a simulation, the engagement begins when the first ship fires a salvo of missiles. And if either ship is hit, or if both ships run out of missiles, the simulation is over. You ultimately settled on a model of a ship versus ship encounter.

How realistic do you think that is as you sort of examine potential adversaries? I think that a ship versus ship encounter is more simplistic than an actual encounter would be. At the very least, aircraft and multiple ships would be in the picture. And my model was more of a closed system. You used Princeton professor's five-component mathematical framework. Can you explain that framework and then why you chose it? I used my thesis advisor, Princeton professor Warren Powell's five-component mathematical framework because it can be used to model any sequential decision-making process, which is exactly what I was trying to do. The framework creates a structure that updates during time steps in a simulation

to show the decision process playing out over time. I use the model to iterate over the decisions of the people and the machines operating a destroyer and a Navy surface missile engagement. So the five component mathematical frameworks consists of state variables, decision variables, exogenous information, transition functions, and an objective function. State variables contain all of the information and only the information that is needed to model the system from that point in time onward. So in this case, the state variable contained all of the information about the ships and the missiles on the ships.

This includes the locations of the ships, whether or not each ship has been hit, how many missiles of each type have been fired and their targets, the ship and missile speeds, and each ship's utilization of ISR equipment. The decision variables encompass all of the decisions Blue is able to make during each time step of the simulation for example, moving the ship or shooting various missiles. The exogenous information is the external information being introduced into the system, and in this case consists of the information that comes from radar and or satellite communications about what the red ship is doing, any incoming missiles, and success of missiles at their destination. The transition function

takes all of the decisions and the exogenous information and move us forward in time to the next step in the simulation. The objective function is the goal of the model. In this case, this was to avoid loss of the ship and life and minimize cost to Blue over the simulation, which includes cost of the ship, cost of the missiles, and other resources expended. How did you develop the simulation that you ultimately use? Developing the simulation I used was actually quite a long process that began during my junior your year. My thesis advisor, Professor Powell, started me off by having me create a mathematical model using his framework, and each week I'd go back to Professor Powell for revision and help on

the model. Then came the initial coding process. I again enlisted someone's help, my computer science lecturer at Princeton, Miss Donna Gabai. She spent countless hours with me helping me dissect my mathematical model and turn it into a Python program that produced results. Each variable in the model had to be directly translated into code. Then once I had my model coded up, the pieces of the puzzle started coming together. Throughout the process, I got feedback from Captain Klein, but I also went out for a week at NPS once I had my initial code in order to get some in-person feedback. Then I spent my senior year at Princeton majorly editing the model and running simulations. So what missiles did you choose to use and why?

So the way that the model was created, I actually have an imported Excel spreadsheet that can take changing inputs for the type of the missile. So the goal of the research was not to do work with exact specs of a missile, but rather to provide a broader conceptual point about surface-to-surface missile engagements. The inputs that I used in my model were pretty generic, and my model had offensive missiles, defensive missiles, and the point defensive measures that I mentioned before. This allowed me to frame on a basic level the offensive and defensive measures present in a missile engagement, and I worked with altering the inputs for the offensive missiles, such as the missile range and the missile success probability during the simulations to show trends in the data while keeping the specs of the other things the same.

I would be using different offensive missiles for different runs of the simulation to show if like you have missiles that were able to be longer range, how that would affect your success in the engagement. Did you do any wargaming to test your hypothesis in addition to the simulation that you're running? My main form of testing this was the simulation. When I first made the simulation, I was running single iterations of the model to see what happened if I changed certain variables. for instance, the salvo size. However, I needed to run a lot of simulations in order to test my hypothesis. And then can you explain how you incorporated satellite capability and use into the study? Because I think that it begins to address the ISR battle and what Hughes would call scouting and counter-scouting that precedes any engagement. The use of the ISR assets was

definitely something that can be expanded upon in future study. But in this model, if blue is using satellite capabilities, then it always gets the information about red used for targeting in the next time period of the simulation. However, if it's not using satellite capabilities, but utilizing radars or electronic surveillance equipment, USVs or UAVs, then I created a probability distribution that determines if blue was able to see any new information about red during that particular time period. So if blue is able to see information, then it was able to make a new decision during that time period. And if not, then and wasn't able to make that decision during that time period. You also discussed randomizing which player gets to make decisions

first. How and why did you do that? So originally, I didn't think it would make a big difference. But then when I put equal inputs for blue and for red, blue was still winning significantly more. So I realized that's wrong. And I fixed this issue by randomly selecting with equal probability which player, blue or red, made decisions first during each time step. So half of the time periods, blue chose first and the other half, red chose first. And this brings us back to Captain Hughes's statement, fire effectively first, and solidifies the point that even milliseconds make a difference. And the ability to make decisions first is essential to success in a missile engagement. This same concept is pretty well known to high frequency traders on Wall Street.

Why did you opt for 15 second time periods? I guess, one, what are the 15 second time periods represent? And then why did you opt for those periods? So the 15 second time periods are the time between each state of the model. And so I opted for these time periods simply because they were a balance frequent enough to show a rapidly evolving situation, but far enough apart that someone could potentially make a decision during the time period. If I said half a second for each one, that wouldn't leave enough time for maybe a thought process to happen. And then a final question is, if you're going to expand this research in the future, and we've hinted at this a couple of times, but how would you do it? Well, to expand this research in the future, I would definitely make the model more complex. I would change more inputs into the model. For instance, more research could be done regarding the ISR assets and the number of missiles and ships inventories could be changed and success probabilities of the defensive measures could be changed. This model only incorporates two ships on a one-dimensional scale because it's like, oh, how far away are these ships from each other? And to expand the model, further research could incorporate multiple ships, aircraft, and shore-based military facilities,

all as a part of the engagement played out on a 2D or 3D scale. Also, a more complex model of uncertainty and Red's location and actions would expand upon this research and make a model closer to a real engagement. I'd like to thank my guest, Kira Dowling. Kira, where can we find you online and what's your next project? Thank you so much for having me, Jared. You can find me on LinkedIn. And my next project is working on my Surface Warfare Officer pin.

Excellent. Thank you for joining us. For the listeners, thanks for tuning in. We'll see you next time. Put it on the shimbo now So help me, Boban Put it in the alley Way, hey, put it in the alley

Help me, Boban Put it in the alley Put it on the shimbo now A pot of rum and a pot of gin, oh Way, hey, put it in the alley A pot of wine, both white and red, oh Put it on the shimbo now Oh, help me, Bob, I'm pulling me out of this.

Why are you pulling me out of this? Help me, Bob, I'm pulling me out of this. Pulling out of this shit, Bob, now.