Download PDF

Stealth military doctrine specifies the use of antidetection technologies for the clandestine movement of friendly forces into unfriendly environments. In this manner, the capability of surprise attacks with high probability of defeating the opponent is available. This doctrine covers both active emissions (low probability of signature of intercept, LPI) and passive structural aspects (very low observable) of the stealth unit.¹ Of course, the holistic signature of the stealth object, like its acoustic, visual, and infrared emissions, must be considered.²

 A great deal of effort has been given to aerial stealth since fighter aircraft and other airborne ordnance (missiles) are an efficient method to project force over land and under sea. The doctrine of fifth-generation or stealth aircraft dictated that the main antidetection principle must defeat radar because the electromagnetic method of detection has the highest early warning capability.³ Therefore, the design and construction of fifth-generation aircraft such as the F-117, F-22, F-35, or B-2, or even stealth warships like the Zumwalt-class destroyer, were based primarily on the reduction of the platform’s radar cross-section (RCS), which is the amount of electromagnetic backscattering that a target reflects to the radar.

By reducing the RCS, the detection range of a stealth platform by adversary sensors increases. For example, an F-35, equipped with an LPI radar and telecommunications system, could operate outside of the detection abilities of adversary sensors and therefore see first, shoot first, kill first, and finally evade unscathed.⁴

Since radar is the primary early warning sensor, it has the farthest detection range of all other sensor technologies. Detection range depends on the RCS of the aircraft, so extreme measures were taken to reduce the backscattering of fifth-generation fighter jets. Examples of these measures include radar-absorbent material, wave-trap structures inside the wings, sigma-shaped exhausts, convolutional body designs, smooth construction to avoid diffraction effects, and fuel tanks in a conformal shape. Considerable effort was also made to provide more capable radars on adversary interceptor aircraft and ground-based antiaircraft systems that could detect stealthy targets as soon as possible—that is, from the greatest possible distance.

Generally, many other measures were taken to defeat fifth-generation aircraft stealth, like the antiaccess/area-denial (A2/AD) concept.⁵ A relevant example is the concept of an advanced integrated air defense system to nullify the stealth aircraft’s ability to penetrate hostile airspace without detection. Nevertheless, it is impossible for any aircraft to remain undetected as it approaches an adversary air defense, even with active electronic warfare measures, because at a certain burn-through range, the target will register on the radar regardless of any RCS reduction and any other standoff auxiliary electronic warfare actions.

Aircraft backscattering differs by frequency band. For example, the B-2 bomber has a very low RCS to X-band radars but quite a considerable return to very high-frequency radars; the physical dimensions of the wings resonate better with the meter wave bands (wavelength) of the high-frequency, very high-frequency, and ultra-high-frequency radars.

Thus, fifth-generation stealth aircraft cannot act as deep penetrating forces into enemy air defense systems but instead can perform as standoff weapons carrier platforms in various theaters of operation. This is especially true for states that wish to maintain a global military presence, like the United States. In other words, fifth-generation aircraft, according to the U.S. school of thought, are a weapon(s) platform that will successfully operate from a great distance and altitude and, thus, far away from adversary sensors. The platform will have the advantage of surprising an enemy force by launching weapons first and then leaving the theater of operations as soon as possible. Additionally, efforts to operate from a greater altitude and distance are justified by the economic factor of lower fuel consumption and the strategic factor of being farther from an adversary electronic order of battle.

Paradoxically, electronic warfare capabilities, either onboard the stealth aircraft or provided by other standoff jammer units like the Boeing EA-18G Growler, are the major factor for a stealth aircraft’s survivability. For this reason, the edge obtained against adversary radars by combining manned fighter aircraft employing stealth and electronic warfare capabilities is negated by the more economical employment of fourth-generation aircraft equipped only with electronic warfare systems, such as the Dassault Rafale or unmanned platforms.⁶ The stealth concept that dictated fifth-generation fighters had several major technical, tactical, and operational disadvantages. These disad- vantages for stealth aircraft certainly did not mean a better game-changing performance from nonstealth aircraft.

On the contrary, the stealth concept was successfully applied in the design of missile systems. For example, the AGM- 158 joint air-to-surface standoff missile (JASSM) employs many elements of stealth design, making it a low-observable weapon. Therefore, the physical dimensions and external shaping of an offensive weapon are important to sustain an offensive mission with high probability of success. After all, the F-35 fighter was designed to carry the AGM-158 cruise missile for the increased probability of successfully completing a sortie from a great distance. The AGM-158 JASSM-ER, for example, has a range of 925 kilometers, and the JASSM-XR version has a range of 1,900 kilometers.

This article proposes a more effective way to employ the stealth concept in a multidomain theater of operations and argues that the future of stealth military doctrine will be influenced by artificial intelligence, swarm tactics, and the scaling down of a platform’s physical dimensions. These concepts should allow unmanned aircraft vehicles (UAVs) and drones to form a stand-in variable speed multiple-domain (SiVSMD) concept. In this manner, UAVs and drones will be used as deep penetration capabilities for invading an adversary sector with a high probability of surprise due to security via obscurity. Specifically, this article stresses that the main point of the future development of stealth military doctrine is based not on surprise by speed but on surprise by obfuscation through artificial intelligence, swarm tactics, and autonomous systems.

Conventional Stealth for Aerial Operations

Historically, air superiority, and ideally, air supremacy, have been the major elements of victory in every confrontation. Numerous examples from World War II show that as long as one army has control of the skies, all other land and naval forces could advance, conquer, and prevail on the battlefield. This produced a credible reason for the aerial stealth concept to become so analyzed and then realized in the form of many aircraft, especially by the U.S. Air Force. Unfortunately, the stealth concept is the most difficult to realize in the skies. As far as radar is concerned, there is no serious volume clutter to inhibit the radar-detection process.

As mentioned, radar is the sensor with the largest detection range; therefore, the radar cross-section of elements and shapes initially was the major concern. Both the United States and Russia, from 1950 to 1970, specifically researched and produced materials that could dissipate electromagnetic energy from radar or the backscattering of geometrical shapes that would reflect the electromagnetic energy away from the radar.⁷

Although emphasis was placed on radar echo, the complete signature of an aircraft—such as infrared radiation, optical, and acoustics—was important to the stealth design to produce an offensive weapon with a high probability of survivability through reduced detectability. Stealth is defined as a concept for aiding the production of offensive weapons and concealing military force mobilization.⁸ Stealth tries to minimize the effort needed to produce necessary surprise. More importantly, there was also a call to incorporate stealth not as a tactic but as a doctrine, or, in other words, a capability or methodology to conduct military operations.

Generally, as described in the Joint Operational Access Concept document, an offensive force attempts to establish an operational access/assured access (OA/AA) situation. Especially for aircraft, stealth encompasses both active (low probability of intercept) and passive (low observability) measures of a platform. The active section uses the concept of LPI to hide the emissions of the stealth platform. For example, the type of radar wavelengths used in stealth have low mean power and are crafted in ways that are not easily intercepted by adversary radars. The passive section uses the concept of low observability to absorb or redirect all reflective backscattering from the platform to adversary radar. Straightaway, aerial stealth is described by the concept of very low observability that tries to reduce the signatures of the stealth platform. These signatures are illuminated or seen by a threat sensor, and the concept of LPI is to try to reduce the detectability of the active emissions of the stealth platform.

Stealth aircraft, such as the F-22 and F-35, were designed long ago. The F-22 was born out of the advanced tactical fighter demonstration and validation program in the early 1980s, and the F-35 first flew in late 2006. Moreover, although many stealth measures were in- corporated into their designs, these were large conventional planes that included a pilot and supporting infrastructure (oxygen, emergency seat). The concept of aerial stealth is to try to reorient the beams or dissipate the energy of adversary radar to produce a cone of silence mainly in the front of the stealth aircraft. In this manner, the reaction time of the related air defense system or interceptor aircraft would be minimized in that direc- tion. Furthermore, stealth aircraft could not penetrate deep into an integrated multilayered and multiaspect (bistatic or polystatic) defense since the aircraft sides would emanate a strong RCS, and there- fore its presence would be detected.

Unfortunately, as shown during operational tests, stealth aircraft still required extensive electronic warfare support to bring their survivability to a satisfactory level. Therefore, stealth measures alone did not provide a game-changing capabil- ity. For example, the F-35’s AN/APG-81 active electronically scanned array radar still was called on to act as a narrow-band jammer against engagement radars once the fighter was deep inside enemy territory. The chance nature of backscattering from a stealth platform could not ensure that the F-35 could act as a real stealth target—and electronic warfare was still necessary. Of course, any jammer could be located by its emissions, and accordingly home-on-jam missiles could be fired to counter this threat.

Conventional aerial stealth doctrine emphasized surprise attacks by two methods. For short-distance attacks using terrain-following radar, the aircraft tried to hide behind natural objects like hills or to fly low over the sea surface to close in for an attack on its targets. This approach could be achieved by conventional planes, such as the Dassault Mirage 2000, Dassault-Breguet Super Étendard, or F-16, but required pilot skill and consumed a lot of fuel. Furthermore, this approach was operationally upgraded using terrain-following missiles, such as the BGM-109 Tomahawk Land Attack Missile, which flies 30 to 90 meters above sea or at ground level. On the other hand, for more distant attacks, the high- altitude radar approach called for the realizations of stealth aircraft such as the F-22 and F-35 since the aircraft would be exposed to open skies, which provide minimal cover to either radar or infrared detection sensors.

Future of Stealth in Air Combat

Theoretically, the stealth entropy concept states that the correlation of the environment and the properties of the invading stealth object define the amount of distance that can be traveled without being realized as a true threat by adversary detection sensors. Furthermore, the stealth platform must approach its adversary at as high a speed as possible to allow for limited reaction time. In this concept, the speed of the stealth object may be low or even drop to zero so as not to let its stealth properties be compromised. The main challenge is how to change the existing stealth concept and manned fighter realizations to a more effective, smaller sized unmanned offensive OA/ AA capability. The most promising way to answer this problem is to employ a synergy of artificial intelligence (AI), swarm tactics, and miniaturization of military platforms.

Artificial Intelligence. AI is a computer-based process that tries to reproduce human intelligence processes. Simply put, AI is a feedback control process in which the inputs are continually assessed for their degrees of similarity against a previously determined correct behavior or pattern. Characteristic examples of the AI computer process are the realization of expert systems, natural language processing (performing a task after a verbal command), speech recognition (converting speech to text), and vision capabilities. Another example is the use of AI to recognize adversary telecommunications and radar signals by comparing their current reception to a previously acquired signal of the same kind.

For military purposes, an agent is used to realize AI applications. An agent is a standalone computer system that can perceive its environment through its sensors and accordingly act on this environment in a desired manner through its actuators. When the task is more complex, such as an air defense task or an offensive mission, a multiagent system is used, in which a collection of agents exhibit synergy—that is, their collective efforts are more effective than that of the capabilities of a single agent. This agent can be in one unit or can span through several units.⁹ Here, a characteristic example is the concept of mosaic warfare, where an agent is only fitted with surveillance abilities and other agents are tasked with delivering payloads or acting as decoys. Thus, in mosaic warfare, a collection of actors that possess different capabilities, like obstacle avoidance and formation control, come together to achieve synergy in the same task, either offensive or defensive.¹⁰

It is important to acquire physical geometrical shape measurements together with infrared, sound, and RCS signatures of opponent naval units for training the AI system of the autonomous mosaic warfare drones. In a real battlefield situation, to avoid hitting friendly assets, an additional safety measure would be an identification friend or foe signal between the drones and the rest of the friendly forces that would prevent a friendly fire incident. Autonomous systems operational risk has been studied.¹¹

Swarm and Guerrilla Tactics With Stealthy UAVs. Military swarming is a battlefield tactic designed to overwhelm target defenses.¹² In this tactic, many units conduct a simultaneous convergent attack on a target from one or multiple offensive vectors to maximize the probability of a successful hit. Using a cost-benefit analysis, the advantage of employing swarm tactics is the ability to attack much more expensive and strategically important targets, such as warships and aircraft carriers, while employing less costly offensive actors, such as missiles or drones.

Miniaturization. The miniaturization process, which is the manufacturing of ever smaller mechanical, optical, and electronics products, has revolutionized all industries. Especially for defense, miniaturization has provided novel capabilities, from sensor systems to complete autonomous military platforms. In the case of unmanned systems, miniaturization eradicates the requirement to sustain human life conditions inside the military airborne platform. The offensive weapon can have small dimensions and thus present a small physical form to the enemy. This fact, when coupled with the use of plastic materials, diminishes the RCS of the unmanned drone and produces a weapon actor that has enhanced penetration OA/AA abilities against adversary air defenses.

The SiVSMD Offensive Concept

The stealth concept will be heavily influenced by AI, swarm tactics, and miniaturization to compose the stand-in variable speed multiple-domain offensive concept. Its main advantage is that it will be easier for the attacking force to create a successful OA/AA operations wedge through heavily defended A2/ AD zones because it will be difficult for the defending force to recognize the infiltrating threat. This capability will allow offensive platforms to act autonomously through AI, to carry out swarm tactics in a noncontinuous time frame, and to be of small physical dimensions. In this manner, the stealth concept is enhanced because the security by obfuscation level is increased and the ability to surprise via a sudden attack will be based on a timewise interrupted attack process and high camouflage.

Mosaic warfare represents a transformative approach to military operations, significantly enhancing the concept of stealth by leveraging distributed, decentralized, and dynamic capabilities. By overwhelming enemy detection and engagement systems, employing electronic warfare and cyber tactics, and integrating autonomous systems, mosaic warfare creates an environment where the element of surprise is maintained and operational effectiveness is maximized. As the military landscape continues to evolve, embracing the principles of mosaic warfare will be crucial for maintaining a strategic advantage and achieving mission success in increasingly complex and contested environments.

Stealthy tesserae make a stealth SiVSMD mosaic. Mosaic warfare enhances stealth in military operations through several key mechanisms:

• Dilution of detection and engagement capabilities: Traditional stealth relies on reducing the RCS and other signatures of individual platforms. In contrast, mosaic warfare leverages the principle of dilution. By operating numerous small and diverse platforms, the enemy’s detection and engagement systems are overwhelmed. Each platform may be easier to detect, but the sheer number and dispersed nature of these assets make it challenging for the enemy to track and target all of them effectively. This dilutes the enemy’s situational awareness and response capabilities, enhancing overall stealth.
• Decentralized and dynamic operations: Mosaic warfare facilitates decentralized operations, where each unit can operate independently or as part of a coordinated swarm. Decentralization makes it difficult for adversaries to predict and counteract movements, as there is no single point of coordination that, if com- promised, could jeopardize the operation. The dynamic nature of these operations, with platforms constantly moving, adapting, and reconfiguring, further complicates enemy detection and targeting efforts.
• Electronic warfare and cyber capabilities: Integrated electronic warfare and cyber capabilities within mosaic warfare can create additional layers of stealth. These systems can jam, spoof, or deceive enemy sensors and communication networks, effectively masking the true nature and intent of operations. Cyberattacks can disable or degrade enemy surveillance and reconnaissance systems, further enhancing the surprise element. By incorporating electronic warfare and cyber tools, mosaic warfare not only hides physical signatures but also disrupts the enemy’s informational and electronic situational awareness.
• Decoys and deception tactics: Mosaic warfare’s reliance on numerous platforms allows for the extensive use of decoys and deception tactics. Some platforms can be specifically designed to mimic the signatures of more critical assets, drawing enemy attention and resources away from the main operational focus. This deception creates uncertainty and misdirection, essential components of stealth. The ability to deploy and coordinate a mix of real and decoy units adds a layer of complexity to enemy defense planning and response.
• Integration of autonomous systems: The integration of autonomous systems into mosaic warfare enhances stealth by reducing the need for human intervention, which can be a vulnerability. Autonomous platforms can execute preprogrammed mis- sions with high precision, operating silently and efficiently in contested environments. Their ability to perform synchronized maneuvers without emitting communication signals minimizes the risk of detection. Additionally, the autonomous decisionmaking process can outpace human reaction times, maintaining the element of surprise.

SiVSMD Impact on Stealth

An undersea mosaic fleet of unmanned subsystems can achieve unprecedented stealth by adopting a novel attack vector strategy, diverging from the traditional high-speed, direct-course torpedo approach. Instead, these autonomous subsystems mimic the behavior of a pod of dolphins, employing a slow, coordinated approach that blends seamlessly with natural marine environments. This tactic involves dispersed, low-speed movements that mask their presence, creating an illusion of harmless marine activity. As these elements converge stealthily toward their targets, such as enemy ships or port facilities, the element of surprise is maintained until the final moment of engagement. This dolphin-like maneuvering not only evades detection systems but also exploits the enemy’s expectations, producing a considerable tactical advantage and a considerable probability of operational success.

The SiVSMD concept is also multi- domain, as shown in the figure. In case A, a conventional stealth attack is shown where mainly a weapon (missile) is using terrain-following or sea-skimming approaches to its target. Guided missiles were the first step toward unmanned weapon systems. The main principle was speed and concealment through terrain- following or sea-skimming methods. In contrast, the respective technique for manned platforms (that is, fighter aircraft) is threading-the-needle, where the aircraft rises and exposes itself to the adversary sensors for a short time and then dips again below the line of sight of enemy radars.

In case B, the fifth-generation stealth aircraft concept is shown, where the main dogma is shoot-and-scoot, or the weapon is released from a great distance and the aircraft subsequently withdraws swiftly from the reach of the adversary air defenses to increase survivability.

In case C, the SiVSMD concept is shown, where fully autonomous unmanned platforms use a multidomain (land-sea-undersea-air) approach to enemy sectors to obfuscate their identity. In this example, the offensive drone, initially airborne, goes underwater as it approaches the perimeter of the A2/ AD sector. When it is safe, the drone becomes airborne again, and as it ap- proaches the adversary installation, it decides whether it has been detected. If so, the drone lands, thus reducing its speed to zero. Finally, when the drone decides that enemy activity has lessened, it again becomes airborne and attacks the enemy units.

Figure. Future Stealth Doctrine

Financial Concerns

There is a justified concern about budget overrun issues in the investment for research and development of novel high-technology weapons like stealth platforms.¹³ A detailed report advises that to avoid economic collapse and strategic retrenchment, commercial-off-the-shelf equipment must be utilized. Undoubtedly there is an inherent uncertainty in the cost of developing novel weapon systems. One important effort of the designers and manufacturers is to reduce this uncertainty by using decision trees bearing appropriate risk probabilities and leading to all possible outcomes. There is seldom any correct prediction about the real cost for the development of a weapon or weapon platform. One way to alleviate this fact is to aggressively look for national and export sales of the system. Nevertheless, there are some basic factors that can be used to estimate the financial performance of a new weapon type.

Weapon-to-Platform Cost. Generally, the attacker-to-defender financial cost ratio must be as little as possible—that is, lesser value weapons should be able to destroy or render unusable targets of much higher value. In the case of unmanned vehicles, there are two major types by cost. UAVs are expensive platforms that need to return to base because they contain expensive equipment, like remote sensing and surveillance sensors. There are many land, sea, air, and under- water unmanned platforms that belong to this category, such as the Northrop Grumman RQ-4 Global Hawk that costs around $275 million. Drones, on the other hand, are expendable platforms, and their return to base is not required. Loitering munitions, which cost around $100,000, belong in this category.

Research and Development Costs and Risks. Conventional stealth aircraft, or fifth-generation fighter aircraft, area a mature design. Many countries have developed such designs, but the main criterion for this homologation is a design that has minimum diffraction returns to the enemy radar, such as the F-22 and the F-35. Nevertheless, such a design must incorporate life support and functions, thus increasing its overall size and limiting its range capabilities.

Based on the elimination of life-support requirements, unmanned platform research and development costs are considerably less. Of course, there will be resistance and reluctance to approve spending on research and development that can produce novel weapon systems. It is more practical to simply issue new equipment made from old designs. Therefore, the dilemma is either to keep older weapons that have proven capabilities or scrap a portion of this older stockpile to find funding for the research of other innovative weapons.

Operational Costs and Risks. The main idea of creating a stealth aircraft was higher survivability in an unfriendly environment based on no detection, thus creating chances for surprise attacks. Moreover, the stealth dogma dictated that a smaller number of stealth aircraft could be used instead of the numerous nonstealth fighters previously required for a mission. Unmanned systems approach this dogma with a much better
solution. For example, a predominantly stand-in unmanned wingman, like the Boeing MQ-28 Ghost Airpower Teaming System, can provide many new capabilities to an offensive air force, from reconnaissance to electronic warfare. Another capability can be provided using fully autonomous systems that may be remotely guided to a waypoint by a conventional fighter aircraft or be deployed by a cargo airplane, such as the Hercules C-130, to conduct their offensive operations in a fully autonomous manner.

Considering all the factors mentioned and conducting a preliminary cost capability analysis that considers the advantages of increased stealth and extended endurance, it can be argued that the introduction of unmanned systems for offensive operations significantly reduces both operational costs and associated risks.¹⁴

Discussion

In World War II, a need emerged for precision-guided weapons. Due to the lack of technological advancement, the relevant solutions were prone to countermeasures, such as the Fritz X antiship glide bomb, or did not have a high degree of precision, such as the V-2 rocket. Moreover, in Japan, the approach was to use human pilots (kamikazes) to inflict a high degree of damage to an opponent with a much smaller offensive unit, such as the Nakajima Ki-115 Tsurugi or the Mitsubishi A6M “Zero” aircrafts. As technology advanced, fighter aircraft still needed a human pilot, but the missile and bombing capabilities were highly upgraded. Nevertheless, all military systems were mainly designed to accommodate and support the human presence; therefore, all necessary amenities and space had to be included in the platform design.

In any case, even today an offensive operation is mainly based on the speed of the attacking weapon. To support this argument, the latest development in missile technology is supersonic and hypersonic capabilities. Toward this attack-vector approach, necessary A2/AD systems were designed to be able to fend off saturation attacks from high-velocity intruders. For example, the S-500 Prometheus air defense system can track targets that are 600 kilometers away and have speeds up to 18,000 kilometers per hour. But with the advancement of unmanned technol- ogy, a new dogma was proposed that shifted the threat-based approach to a capabilities-based one.¹⁵ Analytically, the focus is now on the way a confrontation is conducted and is not based on the identity of the adversary or the specific theater of operations. An unmanned army of autonomous systems guided by AI can perform swarm attacks while its physical dimensions remain small and can blend with the environment. This can lead to enhanced stealth, that is, having an enriched amount of surprise against an opponent’s defenses.

Advantages. This article proposes a dogma shift for offensive operations where a need for high speed to reduce an adversary’s response abilities is not a basic requirement. The center of gravity for the offensive operation is the ability to reduce the signature of the offensive platform through a multidomain path and adjustment of speed to maximize security by obscurity and therefore augment the element of surprise.

Challenges. The most important challenge is that target identification is now required rather than simple target categorization. There is no human in the loop, and some considerations must be taken regarding correct target homing. For example, rules of engagement could be valid only for certain areas where adversary forces are known to exist. These barriers could be defined by global positioning system coordinates or strategically placed buoys that mark the beginning of hostile areas.

Autonomous unmanned vehicles using novel stealth concepts will have a pivotal role in future conflicts. The stealth concept will be heavily influenced by AI, swarm tactics, and miniaturization to compose the stand-in variable speed multiple-domain offensive concept. Its main advantage is that it will be easier for the attacking force to create successful OA/ AA operations through heavily defended A2/AD zones because it will be difficult for the defending force to recognize the infiltrating threat. This platform of small physical dimensions, autonomous via AI and carrying out swarm tactics, will use a stealth concept based on low speeds and high camouflage. JFQ

Notes

¹ Theodoros G. Kostis et al., “Stealth Aircraft Tactical Assessment Using Stealth Entropy and Digital Steganography,” Journal of Applied Mathematics & Bioinformatics 3, no. 1 (2013), 99–121, https://www.researchgate.net/publication/280572920_Stealth_Aircraft_Tactical_ Assessment_using_Stealth_Entropy_and_Digital_Steganography.

² Hao Licai et al., “Research Development of Thermal Infrared Camouflage Textiles,” Journal of Textile Research 35, no. 7 (July 2014), 158–164.

³ Konstantinos Zikidis, Alexios Skon- dras, and Charisios Tokas, “Low Observable Principles, Stealth Aircraft, and Anti-Stealth Technologies,” Journal of Computations & Modelling 4, no. 1 (April 2013), 129–165, https://www.scienpress.com/Upload/JCM/ Vol%204_1_9.pdf.

⁴ Thomas W. Hampton, “The Quest for Air Dominance: F-22—Cost Versus Capability” (master’s thesis, Air Command and Staff College, April 1998), https://apps.dtic.mil/sti/tr/pdf/ADA398765.pdf.

⁵ Joint Operational Access Concept, Version 1.0 (Washington, DC: Department of Defense, January 17, 2012), https://dod.defense.gov/ Portals/1/Documents/pubs/JOAC_Jan%20 2012_Signed.pdf.

⁶ “Document: Department of the Navy Unmanned Campaign Framework,” USNI News, March 16, 2021, https://news.usni.org/2021/03/16/document-department- of-the-navy-unmanned-campaign-framework; George Allison, “Royal Navy Looking at Fixed-Wing Carrier Based Drone for Airborne Early Warning,” UK Defence Journal, April 2, 2021, https://ukdefencejournal.org.uk/royal-navy- looking-at-fixed-wing-carrier-based-drone-for-aew/; George Allison, “UK Considering Carrier Based Drones for Aerial Refuelling,” UK Defence Journal, April 2, 2021, https://ukdefencejournal.org.uk/uk-considering-carri- er-based-drones-for-aerial-refuelling/; Rebecca Grant, Return of the Bomber: The Future of Long-Range Strike (Arlington, VA: Air Force Association, February 2007), https://secure. afa.org/Mitchell/reports/0207bombers.pdf; Rebecca Grant, The Radar Game: Under- standing Stealth and Aircraft Survivability (Arlington, VA: Mitchell Institute for Airpower Studies, September 2010), https://secure.afa. org/Mitchell/Reports/MS_RadarGame_0910.pdf; Carlo Kopp, “Stealth in Strike Warfare,” Air Power International 6, no. 1 (February 2000), https://www.ausairpower.net/API- VLO-Strike.html.

⁷ W.F. Bahret, “The Beginnings of Stealth Technology,” IEEE Transactions on Aerospace and Electronic Systems 29, no. 4 (October 1993), 1377–1385, https://ieeexplore.ieee. org/document/259548; P. Ya. Ufimtsev, Method of Edge Waves in the Physical Theory of Diffraction (Wright-Patterson AFB, OH: For- eign Technology Division, Air Force Systems Command, 1971), https://apps.dtic.mil/sti/ pdfs/AD0733203.pdf; Translated from P. Ya. Ufimtsev, Metod krayevykh Voln v Fizicheskoy Teorii Difraktsii (Moscow: Izd-Vo “Sovetskoye Radio,” 1962).

⁸ Todd Fisk et al., “Integrating Stealth,” Air & Space Power Journal 29, no. 6 (Novem- ber–December 2015), 9, https://www.airuniversity.af.edu/Portals/10/ASPJ/journals/ Volume-29_Issue-6/Integrating_Stealth.pdf.

⁹ Stuart Russel and Peter Norvig, Arti- ficial Intelligence: A Modern Approach, 3rd ed. (London: Pearson, 2019); Kwang-Kyo Oh, Myoung-Chul Park, and Hyo-Sung Ahn, “A Survey of Multi-Agent Formation Control,” Automatica 53 (March 2015), 424–440, https://doi.org/10.1016/j. automatica.2014.10.022; Jorge Rocha, Inês Boavida-Portugal, and Eduardo Gomes, “Introductory Chapter: Multi-Agent Systems,” in Multi-Agent Systems, ed. Jorge Rocha (London: IntechOpen, 2017), https://www.intechopen. com/books/5996; Eleni Kelasidi et al., “Path Following, Obstacle Detection, and Obstacle Avoidance for Thrusted Underwater Snake Robots,” Frontiers in Robotics and AI 6, no. 57 (July 2019), 1–15, https://doi.org/10.3389/frobt.2019.00057.

¹⁰ National Security Commission on Artificial Intelligence: Final Report (Washington, DC: National Security Commission on Artificial Intelligence, 2021), https://nwcfoundation.org/wp-content/uploads/2021/04/NSCAI- Final-Report-AI-Accelerated-Competition-and-Conflict.pdf.

¹¹ Paul Scharre, Autonomous Weapons and Operational Risk (Washington, DC: Center for a New American Security, February 2016), https://www.cnas.org/publications/reports/ autonomous-weapons-and-operational-risk; Charles Clark and Connor McLemore, “The Devil You Know: Trust in Military Applications of Artificial Intelligence,” War on the Rocks, September 23, 2019, https://warontherocks.com/2019/09/the-devil-you-know-trust-in-military-applications-of-artificial-intelligence/.

¹² E. Emary and Hossam M. Zawbaa, “Impact of Chaos Functions on Modern Swarm Optimizers,” PLOS ONE 11, no. 7 (July 2016), https://doi.org/10.1371/journal.pone.0158738; Seyedali Mirjalili, Seyed Mohammad Mirjalili, and Andrew Lewis, “Grey Wolf Optimizer,” Advances in Engineering Software 69 (March 2014), 46–61, http://dx.doi. org/10.1016/j.advengsoft.2013.12.007; Sean J.A. Edwards, “Swarming and the Future of Warfare” (Ph.D. diss., Pardee RAND Graduate School, 2004), https://www.rand.org/pubs/rgs_dissertations/RGSD189.html; Y. Shi and R.C. Eberhart, “A Modified Particle Swarm Optimizer,” 1998 IEEE International Confer- ence on Evolutionary Computation Proceedings (Anchorage, AK: IEEE, May 1998), 69–73.

¹³ Bryce A. Silver, Break the Kill Chain, not the Budget: How to Avoid U.S. Strategic Retrenchment (Norfolk, VA: Joint Advanced Warfighting School, 2016), https://apps.dtic.mil/sti/pdfs/AD1017793.pdf.

¹⁴ AcqNotes, s.v., “Cost Capability Analysis,” July 30, 2021, https://acqnotes.com/acqnote/careerfields/cost-capability-analysis-cca.

¹⁵ Stephen K. Walker, “Capabilities-Based Planning—How It Is Intended to Work and Challenges to Its Successful Implementation” (master’s thesis, U.S. Army War College, 2005), https://apps.dtic.mil/sti/tr/pdf/ ADA434864.pdf.