Artificial Intelligence in Air Defence : Opportunities and Challenges

One of the buzzwords used nowadays while discussing future warfare is Artificial Intelligence (AI) and its application in combat operations. With the future warfare expected to be far more intense, complex and destructive, AI is seen at times to be the real game changer and any Army that plans to stay ahead of the curve is investing heavily in the same. In February 2018, the defence ministry had set up a task force for Strategic Implementation of Artificial Intelligence and Defence followed by creating an institutional framework for policy implementation, and laying out a vision for capacity-building. Indian Army hopes to induct constructive and destructive artificial intelligence soon with mechanised forces being the first to use it primarily in areas of constructive AI to  ‘help a commander in making his decision’.

The use of AI will be expanded and  introduced as per prioritisation of specific areas by the policy makers but one area where AI can be introduced and exploited is in the field of air defence.  But before discussing the applications of AI, it would be appropriate to understand what does AI stand for.

What is AI?

The term AI was coined in 1956 by John McCarthy and was first used to describe any task performed by a program or a machine that, if a human carried out the same activity, would have required application of intelligence.

AI, over time, came to be defined in varied manners with most definitions focussing on it being a sub-field of computer science and how machines can imitate human intelligence. It is defined by Encyclopaedia Britannica as “the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings” while the Merriam-Webster defines it as ‘a branch of computer science dealing with the simulation of intelligent behaviour in computers OR The capability of a machine to imitate intelligent human behaviour.’ Intelligent behaviour refers to the capacity to adapt to change.

Developments in AI and its applications have not necessarily been to imitate or replace intelligent human beings but to assist humans in decision making. If the various sub-fields of AI are seen, they can be grouped under three categories:

  • Replace humans by intelligent machines
  • Systems that can work without humans but without the need to apply human reasoning or logic
  • Assist in decision making using human reasoning but not by replacing it.

From Mimesis to Symbiosis. The use of AI ranges from Mimesis i.e. mimicking human behaviour wherein machines replace humans to perform tasks that normally requires human intelligence to Symbiosis wherein humans and machines work closely together, to transform work processes and organization. Both these are mutually non-exclusive and the usage of AI depends on the type of function it is meant to perform. 

Advantages of Using AI

AI affords a low error rate compared to humans, if coded properly. They would have incredible precision, accuracy, and speed even when handling large volumes of data. AI remains unaffected by emotions, fear or by working in a hostile environment. It can think logically without emotions, making rational decisions with less or no mistakes as compared to humans. Other advantages include the capability to work without breaks, for long hours and do not need rest as humans do. 

Applications in Combat Operations

AI allows combat and combat support systems to carry out regular, repetitive tasks requiring data processing so that the humans can focus on more demanding and complex tasks. One of the fields where AI is majorly used is in surveillancewhere large amount of data is required to be captured and analysed. AI  augments human analysis and decision-makingby capturing information that can be re-applied in critical situations. It also helps in making target decisions, more of  “who to target,” and not “how to target” though 

The other areas where AI can be used is in support systems like logistics

AI and Air Defence 

Air Defence is always a race against time with the biggest challenge being the processing of a very large volume of data in very short time to arrive at the correct decision. The reaction time available for most systems is in seconds with the information and orders needed to be passed to, and within, an air defence system in near real time. 

As the air threat evolves, the reaction time will further decrease and there will be a greater need arrive at the right decision faster, to neutralise all threats using the right weapon system. It is in this that AI can help.

Surveillance. With multiple threats being omnipresent, continuous 3600 surveillance by multi-layered surveillance platforms is essential but this presents a challenge as these platforms generate a very large volume of data that needs to be processed. The data also needs to be stored for later use so as to help in analysis of trends and patterns that can assist in decision making. The challenge is not collecting the data, but processing it, and this is where AI could be of use.

Target Selection. In case of multiple threats emanating at the same time, correct decision of “who to target” is important as the most threatening target needs to be engaged first. This decision making can be assisted by AI as it analyses the air situation and categorizes targets as individual and collective, type of threat it poses and the threat level. An air defence automatic control system (ACS) using AI was recently tested by the Russian Aerospace Forces, that can be integrated with S-300 and S-400 air defence systems, as well as Pantsir gun-and-missile anti-aircraft system.

Weapon Designation. One of the important functions of Control & Reporting System is weapons designation and it is in this domain that AI can help in decision making- depending on giving the recommendations based on readiness of different weapons, kill probabilities, ammunition availability, manning states and other related factors. This assumes greater importance in case of a multi-tiered air defence where a host of weapons systems may be available but it equally critical to select the right weapon system to ensure an effective engagement – and also have an economy of effort so that weapons systems that do not have a reasonable degree of assurance in engaging the target are not activated but are kept ready to take on other targets.   

Missile Defence. With reduced reaction time and more manoeuvrable missiles, one of the most promising fields for application of AI is missile defence. As most missile defence systems mostly operate on semi-automatic or automatic modes with human oversight, these systems already have AI imbedded in their command and control systems. The next step would be to step-up  the application of AI to help in decision making. United States is actively exploring the 

Autonomous Weapon Systems. Directed Energy Weapons (DEW) are one of the options being considered to neutralise small drones and swarms. One of the methods AI can be used is to deploy autonomous DEW systems to take on drones at installations/assets prone to such attacks and where conventional AD systems cannot be deployed due to certain constraints like proximity to border and reduced reaction time. These autonomous systems can carry out all Ad functions using AI and provide an effective cover to such installations. The need, of course, would be to keep such areas ‘out of bounds’ for own platforms for safety.

Soft Kill. One of the methods to neutralise the smart, unmanned aerial platforms is to jam the communications between them but this can only be effective if these platforms actively use communications to relay instructions and orders. In case the platforms function on autonomous, pre-programmed mode with no communications between them, this method would obviously will be ineffective against them.

Logistics. With diversified air defence weapons AI has the capability to allow for more efficient, data-backed logistics and maintenance of equipment.  An interfaces that provides  diverse variety of information related to the air defence systems can help provisioning of ammunition at optimal time. This is important due to the variety of ammunition especially missiles that need to be stored, handled and moved with very specific conditions. Similarly, better management of information on maintenance and repair of air defence platforms can help provide better repair cover. A similar system has already been developed by Lockheed Martin for maintenance of their craft. 

Challenges

Use of AI is not going to be limited to defensive weapons systems alone but will cut across the entire spectrum of combat systems- offensive, defensive and support systems as well. If the defensive weapons systems incorporate AI for better response, the same enhanced capabilities would  be done with available with the offensive weapons as well making them that much harder to be engaged and destroyed. This is particularly applicable to drones and autonomous systems. Besides, availability of data or lack of it will determine the way the system works.

Working in Information Voids. One of the distinctive features of human intellect is instinct and the ability of working in information voids or with incomplete data. There will be occasions when all data or information required to make an informed decision is not available. It is in such a scenario that AI finds itself at a loss of decision making, as yet. It is possible that the next generation machines may be able to work with less or incomplete information but present studies show that AI is more cautious than humans about its conclusions in situations when data is limited. As it would err to be more defensive and reluctant in identifying the enemy, it would limit the options of actually engaging the hostile platforms.

False Target Identification. The target identification and classification as hostile is dependant on the algorithms used and it is here that one of the biggest challenges remain. Very often it has been observed that the AI wrongly classifies friendlies and unknowns as hostile, with disastrous results at times. Two examples should suffice. First, the targeting of a Royal Air Force Tornado by United States Patriot missile during the Gulf War and 

Improvements in Offensive Weapon Systems. Missile defences may improve, and may include lasers for point defence in some places. But because such laser weapons inevitably fall off rapidly in power (as the square of the distance between the weapon and its target), it will be challenging for missile defences to provide area protection. Thus, while it is at least conceivable that ports and airfields could become much better protected, it is hard to escape the prediction that rail lines, road networks involving large numbers of bridges, tunnels, or elevated routes, and large concentrations of supplies in depots or warehouses will be at least as vulnerable in 2040 as they are today. To be sure, missile defences will improve but so will the missiles they have to counter, in terms of their speed and ability to manoeuvre warheads, along with the use of multispectral sensors or seekers.

Drone Swarms. Drone Swarms, and hypersonics,  represent the most serious threat to air defences today. With smarter drone swarms, attack on air bases and other assets is a distinct possibility. While terminal defences could destroy some of the drones, the swarm could adapt and try to overwhelm the defences with a saturation attack or deploy nearby to attack any 

Loitering aerial platforms.  These weapon systems, akin to the sensor fused weapon (SFW), were first used for Suppression of Enemy Air Defences(SEAD) and have matured to more generic roles. Fitting into the niche between cruise missiles and UAV, there aerial platforms can also be employed as swarms. The threat in future may well manifest itself as shape changing swarms that adapt to the defences encountered and adopts new routes and flight profiles to evade the defences and strike the target(s). Going further, an interconnected network of UAVs could be created that is more adapt to evasion and thus more difficult to detect and neutralise.

Cyber threats. As the weapon systems would be networked, will be prone to cyber attacks. Even though they may have developed adequate protection against the present generation cyber attacks like hacking, jamming and spoofing, there would have emerged new threats like  advanced persistent threat which aims to penetrate the  computer systems employing automated capabilities with massive raw computational power that continuously adjusts itself as per defences encountered.

Conclusion

AI can help in decision making but challenges remain in harnessing it and  using the information provided to a Commander to make the final call. The decision making remains an art, not a science, and this is where the test of AI lies – to provide trustworthy, reliable data to the Commanders. 

While GIGO(Garbage In, Garbage Out) may sound outdated and it is assumed that the new generation machines are smarter, it is worth remembering that minor variations in input, at times less than 1 percent, alters the data generated by AI machines. It will get worse if the data provided to the machines is incomplete, unreliable or deliberately false. But with the air defence operations of future giving lesser reaction time to Commanders to decide, the reliance on AI will have to increase. How the AI develops and is exploited for air defence is a real challenge worth working for.