31 March 2017

Formless Warfare: An Innovative Concept to Gather More Information, Analyze it Faster, and Strike Harder

by Michael Kim and Charles Schultz

“If I determine the enemy's disposition of forces while I have no perceptible form, I can concentrate my forces while the enemy is fragmented. The pinnacle of military deployment approaches the formless: if it is formless, then even the deepest spy cannot discern it nor the wise make plans against it.”

-- Sun Tzsu, Art of War

On March 21, 2013, in a conference hosted by the Carnegie Endowment for International Peace, LTG H.R. McMaster stated that there are four main continuities in war and warfare: 1. War is an extension of politics, 2. War is a profoundly human endeavor, 3. War is uncertain, and 4. War is a contest of wills.[i] These thoughts reflect those established by Clausewitz in the early 19th century and continue to reverberate throughout the US Army’s development of doctrine and capabilities. Although it is fair to say that war will continue to be a human endeavor, history has shown that the instruments used to carry out this “contest of wills” can drastically change. This paper presents a vision of future warfare by extrapolating technological trends and uniting them under an operational concept that, when followed, allows the US to remain the sole dominant military force in 2030-2050.

Extrapolating Technological Trends: Assumptions

In order to create a vision for future warfare, it is important to analyze technological trends and establish assumptions. Although these are conjectures into the future, they are necessities when developing future strategies and capabilities. This section analyzes artificial intelligence, autonomous systems, and the proliferation of technologies to provide underlying assumptions from which the proposal is built upon. 

Trend #1: Artificial Intelligence (AI). When AlphaGo beat Lee Sedol, the top Go player in the world, in March of 2016, it resounded throughout the AI community.[ii] Go, a game that originated in China more than 2,500 years ago, has a googol (10100) more possible positions than chess and was thought to be played through intuition and feel – a depth that could be only captured through the human imagination.[iii] AlphaGo established a landmark in AI, not only because it beat the Go world champion, but developed a new way for these systems to learn. AlphaGo created a breakthrough in AI game play when it combined Monte-Carlo tree searches with deep neural networks. These neural networks not only trained on 30 million moves from games played by human experts but played thousands of games between itself (self-play), adjusting connections using a trial-and-error process (reinforcement learning) and discovering new strategies.[iv]

A more powerful example is IBM’s Watson, a cognitive technology (AI) that mimics human thinking through understanding, learning, reasoning, and interacting. In August of 2016, the University of Tokyo reported that Watson correctly diagnosed a rare form of leukemia in a 60-year old Japanese woman that stumped leading Japanese oncologists.[v] In India, Watson correctly diagnosed a rare and aggressive form of breast cancer, suggesting a number of alternative treatments in just 60 seconds based on the patient’s genetic data and medical records.[vi] The US Department of Veterans Affairs has enlisted Watson to help diagnose and treat its patients.

Trend #2: Autonomous Systems. The Teal Group published in 2015 that Unmanned Aerial Vehicles (UAVs) continue as the most dynamic growth sector of the world aerospace industry in this decade. They estimate that UAV production will soar from $4 billion annually to $14 billion, totaling $93 billion in the next ten years. They believe military UAV research spending will add another $30 billion over the decade.[vii] Furthermore, production for UAV payloads, electro-optic/Infrared sensors (EO/IR), Synthetic Aperture Radars (SARs), signal intelligence (SIGINT), electronic warfare (EW) systems, and Command, Control, Communications, Computers, and Intelligence (C4I) systems, forecast to double from $3.1 billion in FY15 to $6.4 billion in FY24.[viii] The military has allocated approximately $4.61 billion for drone-related spending in the FY17 budget with a growing emphasis on unmanned undersea vehicles and unmanned ground vehicles. The budget allocation from FY16 to FY17 showed a transition from unmanned aerial vehicles (due to major acquisition program targets being met) to unmanned sea and ground vehicles.[ix]

Trend #3: Global Drone Proliferation. The accessibility of drone technology is increasing at an alarming pace. According to New America’s International Security Program drone database 19 countries have armed drones and eight countries have used armed drones in combat: the United States, Israel, the United Kingdom, Pakistan, Iraq, Nigeria, Iran and Turkey.[x] Pakistan’s military unveiled two domestically produced drones which experts say appear to be based on China’s popular CH-3 drone.[xi] On February 3, Nigeria announced its first successful drone strike against the militant group Boko Haram, using the Chinese CH-3 model UAV.[xii] The concern does not reside only with state actors purchasing armed drones. Militant groups such as Hezbollah, “which have no responsibility to adhere to international regulations, treaties, and Geneva Conventions,” also has used armed drones in combat.[xiii] Hezbollah acquired Iranian-built Ababil drones, capable of carrying an 88-pound warhead and in September 2014, reportedly used drones to bomb a building occupied by al-Husra. Not all drones used by non-state actors are sophisticated or state of the art. On March 17, 2015, US military officials stated that an airstrike destroyed a small commercially available drone operated by ISIS around Fallujah. In April of 2016, an Iraqi reported shooting down an ISIS surveillance drone confirming the terror network’s use of cheap, commercial products like the DJI Phantom 3, produced by a Chinese commercial drone company.[xiv]

Having assessed technological trends relevant to future warfare, the following assumptions are established: 

The capabilities of AI, through neural networks and future capacities, develop exponentially and are an integral part of future warfare. It is assumed that the US moves past the debate on the regulation of AI and focuses on complete integration into future force capabilities.
 
Autonomous systems are fully integrated into the joint force and act both in conjunction with formations and independently. These autonomous systems are used across the spectrum of operations from logistics to maneuver warfare. 

The proliferation of UAV technology continues to increase capabilities of state and non-state actors. Any militant group or terror network possess easy access to autonomous systems, and the ease of acquisition and low cost of armed drones make it the norm across the battlefield. Cheap and insignificant off-the-shelf drones armed with explosives prove to be a combat multiplier similar to the effects of improvised explosive devices in Iraq and Afghanistan. 

It is clear that future warfare will include AI and drones. Although warfare remains a human endeavor, the use of autonomous platforms that can make decisions independently given the parameters set by humans is inevitable. Having analyzed several instruments of future warfare, the following section examines the required capabilities of these platforms and networks.

Vision for the Future: Required Capabilities 

As state and non-state competitors continue to invest in drones, AI, and counter-battle network capabilities such as cyber, EW, and counter-space, the joint force may face opponents with rough parity. In November of 2016, Pentagon officials presented five questions, that when answered, mitigate future operational challenges. [xv] Based on those questions, the paper presents five required capabilities for future operational concepts: 

U.S military force must be able to quickly deploy, and adapt to uncertain environments while being independent of vulnerable logistical and command nodes that can be targeted by the enemy. 

Battle networks must be robust and redundant to enable survivability in High Intensity conflict. U.S. Networks must also be covert to minimize enemy exploitation. 

US power projection must be 1. decentralized with no center or critical points of failure (negates mobile precision attacks) 2. redundant and plentiful (negates massing and swarming) and 3. must target enemy battle networks, while our battle networks remain obscured from the enemy. 

The US must operate in ways that minimize indicators that can be pieced together to determine our battle plan. We must win the ISR battle by employing more information gathering assets than are available to our enemy, and we must more efficiently aggregate and analyze the incoming data to develop an operational picture. 

Decision-making must be augmented by sensors that enhance human decision making by providing real-time updates. Systems must be developed to automatically and instantly sort incoming information by relevance and urgency to prevent information overload for Commanders. Battle networks must be self-contained to counter cyber threats. 

Innovative Concept: Formless Warfare

Written in the 5th Century, Sun Tzsu’s maxim that the “pinnacle of military deployment approaches the formless” is more relevant than ever. Based on the technological trends, assumptions and required capabilities above, the paper presents an innovative concept – Formless Warfare:

A drone network consisting of platforms (air, sea, land) that are rapidly deployable, expendable, free of vulnerable nodes, self-contained, and operable in teams specifically tailored to mission sets. These networks are governed by Artificial Intelligence (AI) that work together under a HIVE framework: Highly-Intelligent, Independent, Versatile, and Erudite.

The very nature of these drone networks are formless. Systems that autonomously coordinate their efforts to concentrate sensors and firepower at the critical time and place. There are progressions in AI and drone network development that must be tracked. The following sections provide a roadmap and a vignette that display the capabilities of formless warfare.

Road to Development

Development of Intuitive AI. While traditional AI is reactive (every action needed to be anticipated and scripted by human programmers), the newest versions of AI can use past experience to be predictive and proactive. Deep learning AI correlates data to draw conclusions about new events in a method strikingly similar to human intuition. The key to deep learning is the collection of significant data. It is important for the military to immediately aggregate data in a format that is readable and understandable by the AI’s neural networks. It is deep learning that will truly elevate AI capabilities to a profound level of military effectiveness. While traditional AI excels at certain straightforward tasks (like being able to form search patterns, or using image recognition software to spot enemy positions and vehicles), it is deep learning that allows the AI HIVE system to execute abstract tasks, such as determining the enemy's formation of battle, enemy’s plan of operations, and optimal counterattack strategies.

Development of Drone Platforms. Formless Warfare presents a paradigm shift in the way drones are perceived. Current concepts for drone deployment employs the platform like their human operated counterparts which fails to leverage the advantages unique to an autonomous platform. Formless Warfare views drones not like vehicular platforms (expensive, durable, highly capable systems) but munitions (comparatively low cost, single use, specialized platforms). An example of this approach is the experimental Sense and Destroy Armor round (SADARM) that when fired is capable of acquiring and guiding itself to an armored target. If a drone can be built with off-the-shelf consumer electronics then it can be developed cheap enough to be considered expendable. To further decrease costs, these drones should be highly specialized (see Figure 1 for drone examples). While a reconnaissance drone today may be equipped with high-resolution visual spectrum imaging, near infrared imaging, and high frequency communication gear, this single drone could be replaced with three lower cost and smaller platforms that have greater efficiency.



Figure 1. Land, Air, and Sea Support Drone Examples

Integrating HIVE AI and the Drone Network into a New Warfighting System. Formless Warfare provides an innovative concept that takes current technological trends, AI and drones, and combines them into a frighteningly compelling method of determining the shape of the battlefield and concentrating effects at the critical time and place. For any given mission, a specifically tailored drone package is deployed to meet the demands of the operating environment and specific battlefield challenges. The drone package could be minimal, consisting of a handful of inexpensive disposable drones for reconnaissance, or as large as hundreds of linked platforms for complex missions (see Figure 2).



Figure 2. Drone Network Composition Example

Highly-Intelligent. No discussion of the current operational environment is complete without a discussion of the perils of “information overload” – the challenge of analyzing increasing amounts of data to establish a clear common operating picture (COP). The natural advantages of HIVE AI in this category are obvious. Not only can HIVE AI sift through data at inhuman speeds, but the very nature of collecting incredible amounts of information into a database increases its capabilities and effectiveness. HIVE AI turns an inherent detriment (information overload) and turns it into a strength (increasing data for deep learning). The HIVE AI has a continual feedback loop that allows the AI to improve the COP as it continually orients its data collection assets to gain pertinent data.

Independent. The incredible robustness of the drone network provides an enormous advantage in any High Intensity conflict (HIC) in which it is deployed. The computational power required for the AI to administer the drone network can be offloaded to all available drone platforms giving the network as a whole no single points of failure or critical command and control nodes. HIVE AI has no physical location in the drone network, enabling it to disperse processes throughout the system for minimal interruption in network performance. If a drone or multiple drones are destroyed, HIVE AI can send the processes to other platforms and continue the mission. HIVE AI and the drone network combats other domains of future warfare, such as cyber and electronic warfare, because it is self-contained. Hackers cannot take control of the network while it is deployed because there is no active data link, except for radio communications with the governing AI. An electromagnetic pulse (EMP), such as those generated by atomic weapon detonations at extremely high altitudes, may seem like a simple solution to a drone network, however, the simplicity of each individual platform would work to minimize the effects of an EMP. In order to neutralize the network, every last drone must be destroyed. This creates a robust network that is incredibly difficult to destroy.

Versatile. HIVE AI manages the drones, providing flexibility and a knowledge base to react to any operational challenge the drone network encounters. The term “Formless warfare” is generated from the opaqueness to enemy observation that makes the drone network difficult for enemy commanders to define, anticipate, and react. Although each individual drone platform is specialized, the drones share a common base design making it difficult to visually distinguish between drone variants. When the enemy sees a drone network coming to their position, they will not know if it is a package that is designed to attack them, or if it is reconnaissance drones trying to reconnoiter their position. Just as a flock of birds, or a school of fish use huge formations of closely maneuvering individuals to confuse predators, the complexity of the drone networks deployment and identifies would confuse enemies to the networks size, composition, and mission. This makes it impossible for the enemy to anticipate the actions of a drone swarm, or distinguish the main effort of the swarm from various supporting platforms. This advantage allows friendly forces to seize the initiative and set conditions for continuing operations.

Erudite. Modern warfare is an extremely complex, rapidly changing, and multifaceted environment. The dilemma is the ability to create a force structure that is effective across the spectrum of operations (from low to high intensity conflicts). Terms such as “atrophied skills” and “back to basics” are tossed around to denote the difficulty in maintaining units prepared for operations across the range of operations. HIVE AI mitigates this challenge as it has access to tremendous amounts data on all types of warfare. Not only will the AI be an expert on employing the platforms under its command in accordance with all current US doctrine, it will also have access to all battle reports and AARs of all friendly forces across the world. While a commander in a given situation may find a past experience to draw from, the HIVE AI possesses the experiences of multiple commanders aggregated into a single data set. When Deep Blue defeated Gary Kasparov, the reigning world chess champion, in 1997 it was said that he did not lose to a computer but to the “ghosts of grandmasters past.”[xvi] In theory, HIVE AI aggregates the experiences of past commanders to best inform decisions. Even more significant is the ability for HIVE AI to self-learn, whether through simulations or training scenarios (drone networks against drone networks) to develop new strategies and techniques. Like AlphGo, HIVE AI learns from simulations of future mission sets to develop branch courses of actions for multiple scenarios and alternatives.

Perhaps the most important variable in the development of HIVE AI and the drone network is parity. How does the US establish a competitive advantage that off-sets them against near peer state and non-state actors? No doctrine or weapon system should be used against an enemy without a full understanding of imitation and comparative capability. Formless Warfare counters imitation by leveraging uniquely American advantages. American corporations are the pioneers of deep learning computer software. US companies developed Deep Blue (IBM), Watson (IBM), and AlphaGO (Google), which provides an immediate head-start and advantage in developing a comprehensive battlefield AI system. The capabilities of AI and drone networks inevitably develops an arms race. Because of the very nature of self-learning AI, the country that develops warfighting AI systems first possess significant advantages against its competitors. If the US develops HIVE AI now, it has a time and experience advantage against its adversaries. Even when facing similar systems HIVE AI will retain the advantage in employing the assets available to it in a more effective formation.

Vignette

To counter enemy aggression against friendly nations, U.S. Joint Forces, complemented by the new drone technology, deploys to a region. Each component commander possesses a drone package tailored to their specific mission set: land, sea, and air (see Figure 3).

Figure 3. Drone Deployment

Land. To provide early warning of attack or enemy infiltration attempts, several land based surveillance drones are employed to cover routes over the international boundary. These drones are inexpensive with only a low resolution camera and radio transmitter. Many of these drones can be emplaced and camouflaged to ensure that all possible routes are covered. When several of these surveillance drones detect movement and the rudimentary AI in these systems identify irregularities, Joint Force HQ decides to send a reconnaissance focused drone package consisting of 44 networked drones: 10x high resolution camera (High res) drones, 10x near infrared (I.R.) spectrum drones, 6x radio direction finding drones (RD), 6x long range communication drones (L.R.C.), and escorted by 6x hellfire attack drones and 6x gunship drones to investigate the disturbance (see Figure 4).



Figure 4. HIVE AI Massing Drones on Suspected Enemy Locations

The HIVE AI governing these drones analyzes the terrain, determining trafficability and using reverse line of sight analysis to estimate key terrain. The AI then develops an optimal search pattern for the operating environment. One of the high resolution drones quickly spots a clump of vegetation that HIVE AI determines does not match the growth pattern of its surroundings. HIVE AI determines the need for multi-spectral analysis of the area in question (possible artificial camouflage), and moves it higher in an I.R. drone’s priority queue. The I.R. drone maneuvers into position and confirms the use of artificial camouflage, which triggers the A.I. to mark it as an enemy Observation Post.

A Long Range Communication drone radios (via U.S. satellite network) back to the commander requesting permission to engage. The commander denies the request as he is still hoping for a diplomatic solution to the conflict.

HIVE AI uses the presence and orientation of the enemy OP to estimate the enemy’s situational template, and employs the network to reconnoiter the refined areas of interest. The drones soon encounter another enemy formation invading friendly territory. Image recognition software identifies the enemy’s decisive effort consisting of main battle tanks. HIVE AI reports the armored column’s location and direction, and requests permission to engage. The armored column begins to engage the drones, knocking two I.R. drones, and one L.R.C. drone out of the air. The AI governing the drones adjusts formation to compensate for the lost asset and moves the remaining drones to safety while it awaits direction. The commander, given confirmation of the enemy’s hostile intent, chooses to retaliate. The AI uses the enemy doctrine template to estimate the location of key leaders while simultaneously using triangulation of enemy radio transmissions to identify the critical nodes of the enemy’s battle network.

The A.I. prioritizes key vulnerabilities and directs 50 hellfire attack drones to destroy the targets. After expending all ordnance, the A.I. reports battle damage assessment and continues to track the armored column. Real time intelligence, along with the disorder caused by the drone network’s precision strikes, severely degrades the enemy and ensures overwhelming supremacy of US conventional forces.

Sea. As the ground campaign continues, HIVE AI detects that the enemy has coordinated a hypersonic missile strike against the regional U.S. carrier battle group. The carrier deploys its drone network consisting of visual and audio platforms. Although the missile is able to avoid air defense radar by skimming just meters above the surface of the ocean, the missile gives a plain visual and audio signal that is detected by the carrier group drone network. The advanced warning allows countermeasures to be employed, and defeats the missile attack. The Carrier Battle Group maintains freedom of maneuver and continues supporting operations in the region.

Air. The outbreak of hostilities triggers the joint force air supremacy campaign. Phase I deploys another mission tailored drone package over enemy airspace. Like the famous Israeli drone decoy swarm in the Six Day War, these drones draw anti-aircraft fire with the intention of exposing and weakening the enemy's air defense network. Unlike the rudimentary drones used in the Israeli conflict, these contemporary drones are designed to force a response from the enemy. The air mission tailored drone package consists of platforms with precision payloads or anti-radiation missiles. The enemy now faces a dilemma: either suffer constant precision targeting or activate their air defense network. Either decision provides an advantage for the joint force. In this instance, the enemy responds to the strikes by bringing their air defense networks online. Drones overhead immediately fire their armament of anti-radiation missiles to degrade the enemy's air defense radar. In response, the enemy fires anti-air missiles to shoot down several US drones. Given the redundancy in the drone package, the losses do not significantly degrade its capabilities. At the conclusion of Phase I, the air component command deploys conventional aircraft to neutralize the enemy’s air defense network. The drone network’s preparatory reconnaissance and softening of the air defense networks sets conditions for the conventional air force’s mission to destroy the enemy and gain air superiority.

This vignette provides a clear picture of how drone packages can augment the joint force and provide lethal advantages in combat operations. The joint force must take its unique strengths and implement a systematic process to develop drone networks.

Conclusion

The implications of integrating AI into drone networks is vast. The concept of Formless Warfare in 2030-2050 is developed with the presupposition that the concept is inevitable: competitors will integrate A.I. into drones. The inherent benefits of integrating drones and A.I. make this innovation a foregone conclusion. The US must get ahead to gain and maintain a competitive advantage in this field.

In an environment of fiscal constraints, the decisions made today will have a profound impact on the outcome of future operations. Given the nature of the military profession and the resources provided by our nation to execute combat, the responsibility to properly allocate resources, direct training and develop force structure is great. Formless Warfare applies an ancient maxim that has stood the test of time and presents an innovative concept that allows the US to remain the sole dominant military force for the foreseeable future. It is with fervent discipline, focus and creativity that our military and civilian leaders must consider the future of the Army.

End Notes

[i] America’s Second-Longest War: Taking Stock Geopolitical Lessons, Transcript, Thursday, March 21, 2013 , Washington, D.C. LTG HR McMaster, Carnegie Endowment for International Peace

[ii] Alpha Go, Deep Mind, accessed at https://deepmind.com/research/alphago/

[iii] Ibid.,

[iv] Ibid.,

[v] NG, Alfred, “IBM’s Watson gives proper diagnosis for Japanese Leukemia Patient after Doctors were Stumped for Months,” Daily News, August 07, 2016, http://www.nydailynews.com/news/world/ibm-watson-proper-diagnosis-doctor...

[vi] “Tech Impact-How IBM’s Watson is Greatly Helping Indian Oncologists,” Daily Rounds, accessed http://www.dailyrounds.org/blog/tech-impact-how-ibms-watson-is-greatly-h...

[vii] Finnegan, Phil, “Press Release: UAV Production Will Total $93 Billion,” August 19, 2015, http://tealgroup.com/index.php/teal-group-news-media/item/press-release-...

[viii] Ibid.,

[ix] “Drone Spending in the FY17 Defense Budget,” February 15, 2016, accessed at http://dronecenter.bard.edu/drone-spending-in-the-fy17-defense-budget/

[x] “World of Drones: Military,” International Security, accessed at http://securitydata.newamerica.net/world-drones.html

[xi] Sims, Alyssa, “The Consequences of Global Armed Drone Proliferation, July 09, 2016, accessed at http://thediplomat.com/2016/07/the-consequences-of-global-armed-drone-pr...

[xii] Ibid.,

[xiii] “World of Drones: Military,” International Security, accessed at http://securitydata.newamerica.net/world-drones.html

[xiv] Sims, Alyssa, “The Consequences of Global Armed Drone Proliferation, July 09, 2016, accessed at http://thediplomat.com/2016/07/the-consequences-of-global-armed-drone-pr...

[xv] Pellerin, Cheryl, “DoD Crowdsourcing Effort Produces Innovative Operational Approaches,” US Department of Defense, December 21, 2016, accessed at https://www.defense.gov/News/Article/Article/ 1035881/dod-crowdsourcing-effort-produces-innovative-operational-approaches

[xvi] “Frequently Asked Questions: Deep Blue,” accessed at https://www.research.ibm.com/deepblue/ meet/html/d.3.3a.shtml

No comments: