A Vision for the Future of Missile Defense

The CSIS International Security Program is pleased to welcome Vice Admiral Jon Hill, Director of the Missile Defense Agency, to speak on his vision and intent for the Agency. The week of the event also marks the fifteenth anniversary of the initial activation of the Ground-based Midcourse Defense (GMD) system.


This event is made possible through general support to CSIS.
---------------------------------------------------------------------

[stevengarcia]

I swear Admiral Hill is a true patriot and the union should praise him for the safety, ingenuity and dam right military fortitude he possesses. And his mission is clear and one of the most important this year to date

[Franklin-pcxd]

It's frightening to think how vulnerable we are with corrupted ghouls like this involved in our military industrial complex.

[danchak]

Deploy space laser beam to intercept DF-41 and DF-17, does US have that technology at this moment and in the near future?

[timothydavid]

QUESTION: critical; what is the 'production rate' of all missiles? i.e., if Ukraine showed us that they would use 80% of EU/US missiles during this limited war... Taiwan will see a 'multi-fold' use/need of missiles all types. Question 2: have you 'proven' at higher hit rates, and producing in 'volumes' missile intradictions?

[Slavic_Goblin]

To protect Europe from Iran? Oh, yeah totally.
LOL, what's next? To protect Europe from miniature giant space hamsters?

[silversurfergw]

so many threats.. imagine other countries with less resources, money printing abilities having to deal with these threats..

[tonysu]

I don't know why there should be any discussion about what a future missile defense shield should be...
To me, it's pretty obvious what has to be developed and there really aren't any options what it has to consist of.

It starts with understanding the direction of the threat that is developing, which is hyipersonics, which is defined as an incoming attack that is at least 5x the speed of sound, and maneuverable at any phase of its flight.

This means that the threat must be recognized as soon as possible after launch, and launch can be from ground, air, sea or even space. The launch point could be close, just off the shores of any American state or territory, or from directly above. The propulsion of the threat could be chemical reaction, electromagnetic, electrical or simply gravity which means that the threat may silent and does not necessarily emit tell tale signs.

So, particularly to detect and identify threats as soon as possible after launch, the following are needed...
Sensors that point to everywhere on the surface of the earth, perhaps below the surface in the oceans, objects traveling at various subsonic, supersonic and hypersonic speeds in the air and in space that's larger than a specified size that would be considered a threat. Would a nuclear device the size of a shoebox be a threat? I don't know, but minds better than mine know the amount of threat a shoebox sized object could pose.

Assuming that the US has the current and future technological sensor capability to detect objects, an AI system will need to be built that can differentiate threats from the noise in the background. Humans and ordinary non-AI computing will not be able to perform this task for 2 reasons, the massive volume of data and the likelihood that threats can appear very much like ordinary objects. Only AI will be able to aggregate all the relevant sensor data for every object. We actually have the AI and hardware technology today that can support the beginnings of this kind of AI capability, all that would be necessary is to build the virtual environment the AI would learn from and that is not particularly difficult to do.

Once a threat has been detected and identified by a powerful AI, at this point it's likely that some day humans will also allow AI to make the decision to intercept and destroy the incoming threat because every split second loss by inserting human decision making into the entire kill chain exponentially increases the chances of failure. But until humans are willing to turn over the entire decision making process to machines, humans will be part of the decision that could mean extinction of humans on earth.

At least one if not many interceptors would be selected to attack the incoming threat. In general interceptors will likely need to be more maneuverable than the incoming threat and possibly faster. That's a difficult combination of critically needed factors and can be satisfied only by using energy based weapons. Kinetic and chemical propulsion can never be fast enough except against first generation hypersonic weapons. There are two types of energy based weapons, light (LASER) and electromagnetic (EM). The US currently has a significant technological lead in LASER technology, but light is greatly affected by the atmosphere, moisture in the atmosphere which include clouds and boundary layers. For this reason LASER weapons will likely need to be deployed mainly in space and only in close proximity to threats. EM weapons (including rail guns) has more promise but the US has had difficulties developing this technology so currently has no active officially supported research. EM weapons might be subject to powerful EM countermeasures, but in general is able to cut through the environment either projecting a powerful stream of radiation or propel a projectile at hypersonic muzzle speeds, and along with its advantages over LASER weapons within the US atmosphere can also perform well in space. The Chinese continue their research and possible technological lead in EM, possibly motivated by their use of EM in their HSR network. Note also that EM technology is critical to both China's and the US' latest CATOBAR aircraft carriers, the Fujian and Gerald Ford.

So, in summary of the critical technologies described above,
The US has a long road to developing necessary sensors that operate everywhere in the oceans, across land, across air spaces and in space, not limited by territorial claims. Literally every cubic meter of ocean, atmosphere and near earth space and beyond needs to be monitored by sensors that detect in many ways including visual, passive and active, and detecting various emissions.
The US and likely only one adversary China have the AI capability today to develop an AI driven command and control for future missile defense but that doesn't mean other countries including Russia will be left too far behind. Russia and today's lesser military powers like North Korea, Iran and Turkey will likely either develop less capable AI command and control of their own or purchase or obtain through theft what they need.
Interceptor technology might be considered in 2 phases, what is technologically possible in the near future using deployable technologies and what will likely be developed. The US today heavily relies on an upgraded SM-2 which of course is untested in a real combat situation to protect naval assets. The US has THAAD which likely would be ineffective against highly maneuverable hypersonics that either operate outside its domain(like low altitude) or change ballistic trajectory.
Today's interceptors like THAAD, Patriot and the various naval missile interceptors likely will be inadequate against next generation hypersonic and maneuverable threats so even faster interceptors will need to be developed, especially energy weapons. China and the US will likely lead the way in R&D, and each country is on completely different paths which suggests a likely future when one country or the other will obtain a decisive advantage over the other.

Sprinkled in to these technological requirements are time honored basics...
There is always an advantage of look-down positions from high ground to minimize all types of interference and obstacles.
The more data that can be aggregated and sorted in a timely manner, the better decisions will be but data overload can degrade the ability to make proper decisions.
Today, AI is in a Golden Age where in 2017 for the first time hardware became powerful enough to support AI applications like Google's AlphaZero to suddenly become practical and possible, implementing computationally intensive methods like machine learning and the Monte Carlo technique which was conceived over 20 years ago. No longer is AI learning strictly a near sighted iterative task, better AI learning is now based on longer evaluations that can uncover better results that might start with less obvious initially poorer choices. Since AlphaZero was initially published as FOSS and in academic papers, its approach now exists both as a closed, proprietary technology controlled by Google/Alphabet and publicly as an open source technology. The public technology which is open to all has made remarkable and unexpected progress, having now supplanted humans in playing Chess and Go by performing in a way emulating human creativeness that was considered impossible 10 years ago to where even human champions don't expect to win a match and would consider winning even one game would be an accomplishment. ComputerVision AI and applications in the Medical field have demonstrated that AI can be built that equal the very best human experts that can be educated and trained, achieving at least a 97% accuracy and percentile in skill. The important thing to note though is that although it's now fairly easy and cheap(less than $1M to build a world class algorithm that can do almost anything within a year) to build an AI capable of World Class decisions equal to the very best of humankind, those decisions are often different than human results. That's an important concept and something to understand that AI cannot be guaranteed to make the same decisions as humans, while equalling the quality of decisions and for that reason maybe we humans might not want to turn every decision over to an AI.

[peterwilson]

Missile defense, so the Russians use long range artillery. Or slow drones. Ukraine is just the latest test zone for competing weapon systems.

[giorgospapadopoulos]

Putin offered to USA the best missile defence system human ever made in a bargain price. Why to go Mr generals.

[lorenzoo]

So what is this on trick pony is the best option. lets say idk the one trick pony is the only pony able to fire off, then the pony with one trick has the only move the other pony's are not even in the filed. @Aearojet Rocketdyne

[peterwilson]

They should worry about what places Ukraine is reselling what was given to it. And where Zelensky and his Fascist buddies are spending the 84.2 billion dollars.