Skip to main content
NCBI home page
MethodsX logoLink to MethodsX
. 2025 Feb 28;14:103250. doi: 10.1016/j.mex.2025.103250

A new method and information system based on artificial intelligence for black flight identification

Arwin Datumaya Wahyudi Sumari a,c,⁎, Rosa Andrie Asmara b, Ika Noer Syamsiana a
PMCID: PMC11930189  PMID: 40124329

Abstract

Identification of aircraft entering a country's sovereign airspace if it shuts down its identification system, either the Identification Friend or Foe system and/or the Automatic Dependent Surveillance Broadcast system, has long been a challenge for the National Air Operations Command. Aircraft that do not want their identities to be revealed are called black flights and generally have certain missions that can interfere with the sovereignty of a country's airspace. Military radar units that have the task of monitoring airspace are generally equipped with Primary Surveillance Radar that detects the presence of aircraft in their operating area and Secondary Surveillance Radar which functions to identify the aircraft. In the case of black flight, data from the radar in the form of airspeed, altitude, and position are not able to help identify the identity of the black flight. The contributions of this research are:

  • •

    a new method of black flight identification that combines air speed data and altitude with Radar Cross Section (RCS) data using machine learning,

  • •

    a new information system that combines the display of the Plan Position Indicator (PPI) of military radar and ADS-B to accelerate decision-making on black flight,

  • •

    a new approach to national air defense procedures.

Keywords: Air speed, Altitude, Artificial intelligence, Black flight identification, Machine learning, Radar cross section, Recommender System

Method name: Machine Learning with Aircraft's RCS, Altitude, and Air Speed

Graphical abstract

Image, graphical abstract

Open in a new tab

Specifications table

Subject area: Computer Science
More specific subject area: Artificial Intelligence
Name of your method: Machine Learning with Aircraft's RCS, Altitude, and Air Speed
Name and reference of the original method:
  • 1.

    T. K. Ho, “Random decision forests,” in Proceedings of 3rd International Conference on Document Analysis and Recognition, 1995, vol. 1, pp. 278–282 vol.1. DOI: 10.1109/ICDAR.1995.598994.

  • 2.

    C. Cortes, V. Vapnik, and L. Saitta, “Support-vector networks,” Machine Learning 1995 20:3, vol. 20, no. 3, pp. 273–297, Sep. 1995, DOI: 10.1007/BF00994018.
  • 3.

    E. Fix and J. L. Hodges, “Discriminatory Analysis. Nonparametric Discrimination: Consistency Properties,” International Statistical Review / Revue Internationale de Statistique, vol. 57, no. 3, pp. 238–247, 1989.

  • 4.

    J. Ross Quinlan, M. Kaufmann Publishers, and S. L. Salzberg, “C4.5: Programs for Machine Learning by J. Ross Quinlan. Morgan Kaufmann Publishers, Inc., 1993,” Machine Learning 1994 16:3, vol. 16, no. 3, pp. 235–240, Sep. 1994, DOI: 10.1007/BF00993309.
Resource availability:
Open in a new tab

Background

Air Defense Radar (ADR), especially Primary Surveillance Radar (PSR) or Radar for short, which functions as Early Warning Radar (EW), is a sensor that has a very vital role in a National Air Defense System and serves as a detector of the presence of human-crewed or uncrewed aerial vehicles or aircraft entering the sovereign airspace of a country [1]. Radar is generally strengthened by Secondary Surveillance Radar (SSR), which functions as an Identification Friend or Foe (IFF) system that is operated to obtain the identity of aircraft that the PSR has detected. The two types of radars are ideal for a Radar Unit, the frontline guard of a country's airspace sovereignty. The Radar will be installed at one height in strategic locations to maximize the detection range and protect the country's vital objects. The data obtained from the detection of the presence of aircraft generated by Radar is Bearing, Range, Altitude (BRA), and Speed [2,3], or abbreviated as BRAS, while the data obtained by SSR is an IFF code in the form of a squawk number which is used as a means of identifying the identity and type of aircraft [4] detected by Radar.

Black flight is a situation where the aircraft deliberately does not activate the IFF system to reveal its identity. On the other hand, the data from the Radar cannot be used to reveal the identity. The National Air Operations Command must dispatch one fighter aircraft flight to conduct aerial visual observations. However, the decision to dispatch could be a high risk to both pilots and fighters if the black flight turns out to be a more technologically advanced fighter than the fighter aircraft assigned to carry out aerial visual observations. Moreover, if the black flight is equipped with Beyond Visual Range (BVR) weapons, it can be fired with the help of its navigation system. For this reason, the certainty of the identity of black flights that have been detected is a critical factor in making decisions on fighter dispatch flights, and a new technique is needed for identification.

By the rules issued by the Federal Aviation Administration (FAA) of the United States of America (USA), all aircraft, both commercial and military aircraft, can use a flight monitoring system and record flight information of one aircraft called Automatic Dependent Surveillance-Broadcast (ADS-B) [5]. Specifically for military aircraft, the FAA provides the option of whether military aircraft may be used because they have specific secret missions, such as military aircraft surveillance and reconnaissance. The International Civil Aviation Organization (ICAO) has also issued guidelines on using ADS-B on military aircraft with the option to turn off the ADS-B transmitter if necessary [6]. However, in the future, all aircraft should be equipped with ADS-B, which is equipped with an optional feature to share information. On the other hand, non-military aircraft, and even commercial aircraft, are given unique mission payloads to carry out reconnaissance activities. In the latter case, the aircraft will not activate their IFF and ADS-B systems, so their identities cannot be revealed.

Aircraft that deliberately turn off all their identification systems, both IFF and ADS-B systems, can be assured that they are carrying missions that are indicated to endanger state sovereignty in airspace. The situation can become even more dangerous when the aircraft that turns off its identification system is not just one fighter aircraft but one or more fighter flights. However, the aircraft that has been detected in the air but does not want its identity to be revealed cannot yet be known who it really is, whether it is a non-civilian aircraft, a commercial aircraft, an armed non-combat military aircraft, or an armed combat aircraft. The aircraft's identity cannot be disclosed using only standard data from the Radar. For this reason, in this study, a new technique for aircraft identification is proposed that utilizes the predictive capabilities of Artificial Intelligence technology, in this case, machine learning, by utilizing Radar data that has never been used, namely Radar Cross-Section (RCS) and the air speed of the objects.

Method details

The proposed method is a new approach that combines three data that have never been used to identify air objects that cross airspace. The three data sets are RCS, altitude, and the flight speed of air objects. All data sets are inputted for machine learning to identify air objects. In addition to the new identification method, a new information system was built in this study that can display flight data from two different systems, namely the Plan Position Indicator (PPI) on the military Radar and ADS-B on the same display. With this integrated view, the military Radar Unit can monitor all flights in the airspace under its responsibility and quickly identify black flights as a basis for decision-making to dispatch fighter aircraft to act against such air objects. In this research, the focus is on the aircraft, whether military or civilian/commercial.

Method validation

Validation is carried out using the machine learning method and the new information system. The new information system is validated directly with prospective users from the Indonesian Air Force personnel in the Radar Unit and the commands above it.

Introduction

Detecting, recognizing, and identifying air objects flying through sovereign airspace is one of the duties and responsibilities of the National Air Operations Command, especially the ranks of Radar Units under the Sector Command. Each Radar Unit is generally strengthened by a Primary Surveillance Radar (PSR), or Radar for short, which also acts as an Early Warning Radar (EWR), and some of them are equipped with a Ground Control Interceptor (GCI) function, which is responsible for detecting the presence of air objects in the airspace within its scope. Because of its role as an early warning provider, the Radar typically has a long range of up to 400 NM, such as Thomson TRS 2215 D [7]. The detection data is in the form of data on airspeed, altitude, and the object's position in the air against the PSR antenna or called bearings. Identification of the identity of the air object detected by the PSR is carried out by the Secondary Surveillance Radar (SSR) [8]. This identification process is called Identification Friend or Foe (IFF) [9,10], where every air object, especially an aircraft, has been equipped with unique codes that show its identity. The SSR will generally transmit an interrogator signal to the aircraft detected by the Radar. After receiving the signal, the aircraft's IFF system will send a responder signal to answer the interrogation from the SSR.

In certain circumstances, aircraft whose presence has been detected in airspace cannot be identified. There are several possibilities for this to happen; namely, the IFF system on the aircraft suddenly inactivates, there is an error in the IFF system, or it is deliberately turned off so that its identity is not revealed. The highest threat comes from the last possibility. Suppose the turn-off of the IFF responder is due to a deliberate factor to avoid identification from the Radar Unit where the aircraft is flying. One situation where an aircraft deliberately turns off an IFF responder so that his or her identity is not revealed is called black flight [11]. Flights like this are one of the indications of the improper intention of the aircraft. The standard procedure for identifying black flight is to send a fighter plane flight for visual identification to know its identity [1], which includes the type of aircraft, country of origin, whether it carries weapons, and the purpose of the flight.

There are three challenges in identifying black flights using the current procedures. First, visual observation can be at high risk if the detected aircraft turns out to be armed and has higher capabilities than the fighter aircraft flown to make the air visual observation. Second, based on the results of field visits, the Radar Unit needs to have the authority to identify aircraft. The authority is one level above the command unit, namely the Sector Command. With this mechanism, there will be a time delay between the time when the presence of the aircraft is detected by the local Radar Unit and the decision to dispatch the fighter flight for visual observation due to the identification process carried out by the Sector Command. Moreover, the decision to dispatch fighter aircraft is the authority of the command above the Sector Command, namely the Air Operations Command, which oversees it. Third, there is a high risk that can occur for the flight of the fighter aircraft to make visual observations when it is found in the air that the black flight turns out to be a hostile fighter aircraft equipped with more sophisticated weaponry [12].

Many studies leading to the recognition and identification of air objects have been carried out, including those based on Artificial Intelligence technology, especially using methods in machine learning. However, research closely related to the use of RCS for aircraft recognition and or identification is very rare and almost non-existent. Very rare research in this field is very possible because RCS data is confidential data, especially RCS military aircraft However, research closely related to the use of RCS for aircraft recognition and or identification is very rare and almost non-existent. Very rare research in this field is very possible because RCS data is confidential data, especially military aircraft RCS for combat missions and reconnaissance or surveillance. Two studies related to RCS in weapon systems are how to calculate RCS for fighter aircraft and Ballistic Missile Defense System (BDMS) using computational electromagnetics using available data [13,14]. On the other hand, the use of machine learning for recognition and identification including utilizing RCS data is more widely used on Unmanned Aerial Vehicle (UAV) objects, while recognition and identification in the military and defense sectors are almost unfounded. The only research that discusses RCS for the military field is to discuss stealth technology as well as the principles of making aircraft difficult to detect by radar [15]. The method proposed in this study is a new and different approach from the existing one, which combines RCS, altitude, and air speed data as input to machine learning for the recognition and identification process. Through this research, there are four things that we contribute as follows:

  • 1.

    Artificial Intelligence-based black flight identification method using RCS, altitude, and airspeed data,

  • 2.

    A new information system that combines the ADS-B view of the public and the PPI view of military Radar,

  • 3.

    A new information system that displays the prediction of identification results as the basis for decision-making of fighter dispatch flights to make visual observations,

  • 4.

    A new information system that can reduce the risk of fighter flights will make visual observations.

Research method

The research methods carried out to build a black flight identification method include collecting RCS, altitude, and airspeed data, processing these three data so that they can be processed by Artificial Intelligence technology, especially the selected machine learning methods, and testing methods. On the other hand, research methods to build new information systems include creating user specifications, Use-Case Diagrams, coding applications, and testing applications. The following research method integrates the black flight identification method with the information system and conducts integrated application testing involving potential users.

Data collected by radar system from detected aerial object

Radar is a system used to detect the presence of one or more aerial objects within its observation range. In the context of airspace, Radar is used to detect the presence of one or more objects flying in the airspace under its surveillance coverage. In its operation, the Radar will routinely sweep the airspace periodically with the rotation of the radar antenna, which has been adjusted according to needs. When an object is exposed to a radar signal, the object's body will reflect the signal in the form of an echo back to the Radar, which will be received by the Radar receiving antenna. The reflected signal from the object is processed into four primary data, namely airspeed, altitude, range or distance, and the object's location in the air against the radar antenna called bearing [3].

The data received by the radar antenna will be displayed on the display screen to show the detected object's location. Five types of displays can be used according to the needs: A-Scope, B-Scope, C-Scope, P-Scope or PPI, and Range-Height Indicator (RHI) [15]. PPI is a circular display screen that displays signals reflected horizontally. The data displayed is range or distance, and the azimuth angle is displayed as polar coordinates representing the object's bearing relative to the Radar's position. Unfortunately, those data cannot be used to recognize and even identify the detected air objects. According to the current procedures, the detection results are reported to the Sector Command for identification to identify the detected object's identity. The SSR identifies air objects, transmits interrogator signals to the object, and waits for a reply in the form of a responder signal containing its identity data. The challenge arises when the aerial object does not respond to the interrogation signal sent by SSR.

Based on the results of the visit to several radar units, one more datum can be obtained from the Radar: the reflected signal from the object that represents the cross-section of the detected air object. However, the cross-section signal of this object cannot be processed due to the absence of equipment for this need because it is adjusted to the task of the Radar Unit, which only detects the presence of objects rather than further processing the reflected signal from the object. With the failure of the SSR to identify aerial objects, cross-section data could be a solution for military radars for aerial object identification. This problem also motivates identifying air objects in the radar unit to speed up decision-making time, especially for air objects categorized as black flight. This study does not discuss the manufacture of hardware or software that processes the reflected signal to obtain cross-section objects but focuses on processing RCS data obtained from various sources and combining it with airspeed data to recognize and identify air objects.

ADS-B

Automatic Dependent Surveillance-Broadcast (ADS-B) is a surveillance system used to monitor the movement of air objects, especially aircraft, starting from the condition of stopping at the apron at the departure airport before carrying out the flight until the aircraft lands at the arrival airport to stop at the airport apron. ADS-B utilizes information from the Air Traffic Controller (ATC), avionics equipment on the aircraft, space facilities, and facilities on the ground, including information from other aircraft. With ADS-B, the aircraft continuously transmits flight information from start to finish after arriving at its destination. The advantage of ADS-B is that most of the aircraft information can be obtained by anyone who operates the equipment to obtain flight data from ADS-B. Another advantage that makes it a prospective system for the aviation world is the possibility of replacing SSR, which must transmit an interrogator signal first to identify the aircraft and not necessarily be responded to by the aircraft.

The ADS-B system consists of transmitters and responders (transponders) installed in aircraft and receivers installed at ground stations. The ADS-B system on the aircraft receives aircraft position data in the form of latitude and longitude locations as well as the time when the aircraft is in that position from the Global Positioning System (GPS) or Global Navigation Satellite System (GNSS) receiver according to the position data sent by GPS/GNSS satellites. The data is transmitted to ADS-B stations on the ground as ADS-B Out signals or sent to nearby aircraft and received as ADS-B In signals. ADS-B position data sent from aircraft to ADS-B ground stations is sent to the Air Traffic Controller (ATC) as flight information to complete the data that has been owned. The ADS-B transponder must be activated for commercial and non-military aircraft, while the ADS-B transponder is optional for military aircraft. Given that the aircraft position data from ADS-B is critical, in this study, the information from ADS-B is combined into the military Radar PPI display to provide recognition and identification to speed up the decision-making. How ADS-B works, and its relationship to other systems is shown in Fig. 1.

Fig. 1.

Fig 1

Open in a new tab

Illustration of how ADS-B works.

Radar cross-section and air speed

The Radar Cross-Section (RCS) of a single object is the area that a radar perceives. The imaginary area catches that level of power, which, when spread out evenly in every direction, creates a reflection on the Radar like that from the object. RCS is data currently not used for recognizing and identifying air objects in Radar Units and the above command. On the one hand, the possibility of further processing of Radar reflected signals to generate RCS data with a sensor requires advanced technology. On the other hand, RCS data is not easily obtained through public sources because it is confidential data, and it is related to the ability of an aircraft not to be detected by Radar. In the study, RCS data from various commercial, non-military, and military aircraft was obtained from various sources, both from scientific articles and websites that have been validated for their existence and continuity in providing information online. RCS is measured in square meters (m2) or milliwatts (dBm). For example, The RCS of the F-16 fighter aircraft is estimated to be 5 m2.

As mentioned in the previous section, one of the data obtained by Radar is the flight speed of the aircraft detected by it. There are several types of aircraft speeds, namely True Air Speed (TAS), Indicated Air Speed (IAS), Ground Speed (GS), and Calibrated Air Speed (CAS). TAS is the speed of the aircraft relative to the air it passes, IAS is the speed of the aircraft as indicated by the avionic Air Speed Indicator (IAS) instrument in the cockpit of the aircraft, GS is the speed of the aircraft relative to the earth, and CAS is the IAS corrected due to an error in the instrument and position. This study uses aircraft flight speed data in TAS because it is the actual flight speed in the air. For example, the F-16 fighter aircraft is 480 knots (kts) in cruising conditions. The main reason for using aircraft airspeed data is that it is highly unlikely that the aircraft will fly below its minimum speed because it will have a stall impact or loss of lift, which will impact the loss of altitude and fall to the earth.

RCS, altitude, and TAS data are obtained from many sources, both from journal articles [16] and proceedings [17] as well as from online sources [[18], [19], [20], [21], [22]], that can be accounted for validity. The data collected are shown in Table 1. RCS data is classified as very confidential and sensitive because it is closely related to the self-protection of military aircraft, especially fighters, and attackers, and the ability to infiltrate without being detected by Radar. The RCS data presented here are data that are disclosed to the public either from research results or studies from sources that can be validated. One of the aerospace technology competitions between developed countries is the advantage of producing fighters, attackers, and bombers with high stealth capabilities, the leading indicator of which is the size of the RCS. Table I shows RCSs for three living things compared to the aircraft and missile's RCSs.

Table 1.

RCS, Altitude, and TAS.

No. Object Name RCS (m2) TAS
 Altitude (feet) Role
Kts Mach
Fighter
1. Eurofighter Typhoon 0.1 1147 2 55,000 Multi-Role
2. F-4 Phantom 6–10 1280 2.23 59,000 Fighter-Bomber
3. F-15 Eagle 10–25 1630 2.84 60,000 Air Superiority
4. F-16A 5 1147 2 50,000 Multi-Role
5. F-16C (with reduced RCS) 1.2 800 1.4 50,000 Multi-Role
6. F-16IN Super Viper 0.1 687 1.2 50,000 Multi-Role
7. F/A-18C/D Hornet 1–3 1200 2.09 52,300 Multi-Role
8. F/A-18E/F Super Hornet 0.1 1067 1.86 52,300 Multi-Role
9. F-22 Raptor 0.0001–0.0005 1303 2.27 65,000 Air Superiority
10. F-35 Lightning II 0.0015–0.005 1067 1.86 50,000 Multi-Role
11. F-117A Nighthawk 0.025 594 1.04 67,000 Stealth Fighter
12. Hawk Mk209 N/A 560 0.98 50,000 Attack
13. Mig-21 3 1175 2.05 50,000 Multi-Role
14. Mig-29 Fulcrum 3–5 1330 2.32 60,000 Air Superiority
15. Mirage 2000 1–2 1453 2.54 59,000 Multi-Role
16. Rafale 0.1 1200 2.09 55,000 Multi-Role
17. Su FGFA (Su-57 Version) 0.5 1347 2.35 65,000 Multi-Role
18. Su-27 Flanker 10–15 1350 2.36 60,000 Multi-Role
19. Su-30 MKK/MK2 4–12 1145 2 60,000 Air Superiority
Bomber
20. B-1B Lancer 0.75–1 825 1.44 60,000 Bomber
21. B-2 Spirit 0.1 550 0.96 50,000 Stealth Bomber
22. Β−52 Stratofortress 100–125 560 0.98 50,000 Bomber
Cargo/Military Transport
23. C-130 Hercules 80 365 0.64 30,000 Cargo
Passenger Plane
24. Boeing 737–200 1–150 418 0,73 37,000 Passenger
25. Boeing 737–900ER 1–150 453 0,79 41,000 Passenger
26. Boeing 737–800 1–150 453 0,79 41,000 Passenger
Trainer
27. T-6 Texan II 0.4–6 316 0.55 22,500 Trainer
28. T-38 Talon 1 745 1.3 53,000 Trainer
Surveillance
29. A-12/SR-71 0.01 18,909 3.3 85,000 Reconnaissance
Helicopter
30. AH-6 Little Bird 1–2 126 0.22 20,000 Light Attack/ Transport
UAV
31. Small Drone 0.01- 0.1 11,5–28,6 0,01–0,05 400 Various
Missile
32. AGM-86 ALCM 0.1–0.5 435.6 0.76 12,7 Air-Launched
33. Exocet 0.1 533 0.93 98–229 Anti-Ship Missile
34. Harpoon 0.1 487 0.85 12,6 Anti-Ship Missile
35. MIM-23 Hawk 1.0 1547 2.7 65,000 Medium-size Missile
36. MIM-72 G Chaparral 0.1 1318 2.3 10,000 Small-size Missile
37. MIM-104 Patriot 1.6 1719 3 60,000 Air Defense Missile
38. Tomahawk SLCM 0.5 429.75 0.75 98–164 Submarine-launched
39. Tomahawk LAM 0.5 475.60 0.19 98–164 Land-Attack
Living Entity
40. bird 0.01 – – –
41. insect 0.001 – – –
42. man 1 – – –
Non-Aircraft
43. automobile 100 – – –
44. cabin cruiser 10 – – –
45. truck 200 – – –
Open in a new tab

Machine learning

Machine learning is a subset of Artificial Intelligence and has a variety of prospective methods to solve many problems humans face. There is no ultimate method that can solve all problems. Each method has advantages that can be empowered to provide solutions to specific problems. Machine learning is aimed at imitating certain human intelligence, such as making predictions about phenomena in their environment. Like humans, machine learning needs data to build knowledge; from that knowledge, it can make predictions. Methods in machine learning build knowledge by learning from past data or experience. Several learning paradigms for building knowledge are supervised, unsupervised, semi-supervised, and reinforcement learning.

In this study, identifying air objects indicated as black flight using one machine learning method is considered to perform best. Five machine learning methods are tested for recognition and identification: Decision Tree (DT) [23], Random Forest (RF) [24], k-nearest Neighbor (kNN) [25], Support Vector Machine (SVM) [26], and Neural Network (NN) with a backpropagation mechanism or BPNN [27].

Result and discussion

This section will convey two main things: a new method for black flight identification based on Artificial Intelligence, especially machine learning, and a new information system for black flight recognition and identification that combines the ADS-B and PPI displays from military Radar.

Finding and validating the most optimal method for black flight identification

The training will be conducted on the five machine learning models presented in the previous section to obtain the optimal model for recognition and identification. We produced a synthetic dataset based on references from different sources, as shown in Table 2. A 600-sample dataset comprising RCS (m2), TAS (knots or Mach), and altitude (feet) data was generated for seven distinct aircraft roles during this simulation stage. As indicated in Table 2, we only display 20 data samples from this dataset. These 30 aircraft types are classified into seven groups: military fighter, bomber, cargo or military transport, commercial or passenger plane, trainer, surveillance, and helicopter. They serve three distinct purposes: tactical, multirole, and transport. The F-16C and Su-30MKK/MK2 are fighter aircraft types with sample heights ranging from 5000 feet to maximum altitude. Between 23,000 and 41,000 feet is the detected altitude for commercial and cargo airplanes.

Table 2.

Synthetic RCS, Altitude, and TAS dataset.

No. Type RCS (m2) TAS
 Altitude (feet) Role
Kts Mach
Fighter
1. F-16C (with reduced RCS) 0.627797 905 1.58 38,639 Multi-Role
2. F-16C (with reduced RCS) 1.106826 1031 1.8 42,434 Multi-Role
3. F-16C (with reduced RCS) 0.958878 1083 1.89 41,977 Multi-Role
4. F-16C (with reduced RCS) 1.151872 940 1.64 35,861 Multi-Role
5. Su-30 MKK/MK2 7.605235 980 1.71 45,043 Air Superiority
6. Su-30 MKK/MK2 7.210643 928 1.62 45,670 Air Superiority
7. Su-30 MKK/MK2 8.607113 1014 1.77 45,815 Air Superiority
8. Su-30 MKK/MK2 7.210643 1031 1.8 43,847 Air Superiority
9. Boeing 737–200 8.607113 464 0.81 39,297 Passenger
10. Boeing 737–200 5.71384 470 0.82 39,130 Passenger
11. Boeing 737–900ER 135.0688 465 0.82 39,565 Passenger
12. Boeing 737–900ER 142.408 475 0.82 39,218 Passenger
13. C-130 Hercules 83 309 0.54 31,005 Cargo
14. C-130 Hercules 95 304 0.53 32,436 Cargo
15. C-130 Hercules 85 309 0.53 30,835 Cargo
16. C-130 Hercules 87 323 0.54 32,379 Cargo
Open in a new tab

The synthetic dataset for aircraft identification was categorized according to their kinds. Table 3 displays the findings utilizing several pre-existing classifier models with conventional parameters. With an accuracy of 82 %, the RandomForest model achieves the maximum accuracy. However, with an accuracy of 32 %, the SVM model achieves the lowest accuracy. While BPNN and K-NN with K = 1 both have an accuracy of 68 %, Decision Tree's accuracy of 79 % is higher than those of the other two models. Due to its lowest accuracy rating of 32 %, the SVM result is worth discussing. SVM is a better non-linear classifier; hence, this outcome should not occur. It was discovered that the F-16 and Su-30 military aircraft data were confused in practically every group of aircraft types in the confusion matrix results displayed in Table 4. Similarly, 32 % of the Hercules C-130 military transport aircraft data were classified as military fighter aircraft, whether Su-30MKK/MK2 or F-16C. These findings demonstrate how difficult it is to identify an aircraft accurately when the IFF system is disabled. Compared to other machine learning methods, the RandomForest algorithm demonstrates its superiority. Its structure, which includes multiple trees to lessen the chance, may contribute to its strength.

Table 3.

Comparison of several machine learning methods’ accuracy, precision, and recall for aircraft recognition and identification.

Decision Tree K-NN Random Forest SVM BPNN
Accuracy (%) 79 68 82 32 68
Precision
F-16C 1.00 0.68 1.00 0.27 0.75
Su-30MKK/ MK2 0.95 0.67 1.00 0.29 0.67
Boeing 737–200 0.50 0.52 0.55 0.25 0.50
Boeing 737–900 0.00 0.54 0.56 0.35 0.50
Hercules C-130 1.00 1.00 1.00 0.33 1.00
Average Precision Recall 0.69 0.68 0.82 0.30 0.68
F-16C 0.95 0.65 1.00 0.20 0.60
Su-30–30MKK/ MK2 1.00 0.70 1.00 0.20 0.80
Boeing 737–200 1.00 0.70 0.60 0.05 0.60
Boeing 737–900 0.00 0.35 0.50 0.65 0.40
Hercules C-130 0.79 1.00 1.00 0.50 1.00
Average Recall 0.75 0.68 0.82 0.32 0.68
Open in a new tab

Table 4.

Views and inputs from questionnaire answers.

No. View Input
1. The application can assist in implementing air defense operations by displaying the air situation as airspace and capturing both civilian and military aircraft within the coverage area. Further developed applications can utilize the frequency database of electronic equipment civilian and military aircraft use for identification purposes.
2. The application can provide information about targets captured by the Radar but does not yet have identification or black flight. With AI technology, the application can provide information on possible types of black flight aircraft, making it easier for the National Air Operation Commander-in-Chief to determine interception fighters to carry out scramble orders or interception missions. The aircraft data label can include information such as the Squawk Number [SN], the modes used in modes 1, 2, 3A, and S, and target range and bearing information from Radar or a point on the map.
3. The current application is not able to display black flight. Applications can be developed to display any Radar source or Radar that captures the same target: civilian and military Radar.
4. This web-based application makes it easier for operator personnel to monitor flights passing through the airspace under their supervision. Applications need to be developed to integrate active radar capture and passive Radar so that the information displayed is complete.
5. This application makes it easier for leaders to determine further policies, especially regarding interception actions.
Open in a new tab

Air surveillance and recommender information system

In order to implement an information system that displays the dynamics of airspace in the sovereign territory and can display aircraft detected by Radar and flight data by ADS-B, an air surveillance system is needed that can display aircraft detected by Radar as well as aircraft whose flight data is known by ADS-B to an air surveillance and recommender system. This information system is not found in the current military radar layer view or in computer applications that can track aircraft travel by utilizing ADS-B data. A flow chart, a User-Case Diagram, and an Activity Diagram have been created to build this information system application. The flow chart shows the working mechanism of the information system from the time the system is run to receive and process the data to display the prediction results that lead to the recommendation of a fighter dispatch flight to make visual aerial observations. On the other hand, User-Case Diagrams are compiled to show the interaction between information system users and supported facilities, along with the authorization level to access the information system. An activity diagram is used to show the stages of the mechanism used in this information system application. The three diagrams are shown in Fig. 2, Fig. 3, Fig. 4.

Fig. 2.

Fig 2

Open in a new tab

Information system flowchart.

Fig. 3.

Fig 3

Open in a new tab

Use-case diagram [28].

Fig. 4.

Fig 4

Open in a new tab

Activity diagram [28].

The User-Case Diagram showed that the actors who had access to the system were high-ranking officials in the National Air Operations Command [28]. They will be able to access the system and learn the name and location of the Radar Unit, which is represented by icons on the Earth map. Furthermore, the system offers tools to identify air objects that travel across national airspace, including aircraft codes, departure and destination flights, flight duration, and real-time air object location coordinates. Suppose the identified aerial object fails to activate the transponder and is subsequently classified as being in black flight. In that case, the system will identify and recognize the object using Radar data to predict its identity. The air object's type, name, and capabilities will be predicted based on the prediction's results. Based on this forecast's outcome, the National Air Operations Command will conduct one fighter mission to perform an aerial intercept.

The following actors utilize the system: the Radar Units Commandant, the Commandant of the Sector Command that oversees a group of Radar Units, the Commander-in-Chief of the National Air operating Command, and the Commander-in-Chief of the Air Operation in each operating region. Briefly said, the Sector Command, which in turn reports to the Commander-in-Chief of the National Air Operation Command, will receive reports from the Radar Unit that detects unknown aircraft or aerial objects. After that, he makes choices and gives the relevant Region Air Operation Command instructions to send out a fighter flight with three fighter aircraft so they can intercept or bring down the unseen aircraft or aerial objects.

The authorized user of the National Air Operation Command accesses the air surveillance and recommender information system and enters the administrator-specified user and password. To access the system, a user who is not enrolled must register with the system administrator. If users enter their password or account username incorrectly, they will receive an error message and be prompted to enter the proper information. Following successful login, the Commander-in-Chief of the National Air Operation Command or other authorized users will see the system's main screen, which by default displays an Indonesian map that may be zoomed in and out and moved in all directions. Additionally, the maps allow for global map viewing. Choose the ``Surveillance'' icon to access the radar system interface's primary display. The ADS-B and radar facilities are accessed by icons at the top center of the main system display, representing the two primary system characteristics.

ADS-B will automatically activate by default and show all aerial objects on the map that are highlighted. The main display will show every identified aerial object that ADS-B has captured. The information presented for the airborne objects comprises the kind of aircraft, the airports of origin and destination, the flight speed and altitude, and the direction from the radar station. If the Radar function is enabled, all aerial objects within the range will be shown in real-time on the PPI and the main display screen. Black flights, also known as unexplained aerial objects, are aerial objects that disable IFF and ADS-B. AI technology will identify black flights by analyzing Radar data containing crucial information known as RCS. The system will notify the National Air Operation Commander-in-Chief, the decision-maker, of future decisions and actions if the black flight poses a threat.

In the Activity Diagram [28], to signify the initial status, initial action, or beginning point of activity in the Radar system interface with an AI-based black flight identification system, the process begins at the start point. The National Air Operations Command must authorize the system before accessing the surveillance menu. This menu has two information menus: Radar Information and ADS-B Information. All radar data is shown in the radar menu, while details about captured objects in the air are shown in the ADS-B menu. A registered item will show some information (such as the code, airline name, city of origin, destination city, altitude, and speed), and an AI algorithm will be used to identify unregistered objects detected by Radar. The identification method will show data output on suggested aircraft to intercept and object predictions. This information output will assist the National Air Operations Command determine whether to launch a fighter aircraft flight to perform an airborne intercept.

Operating air surveillance and recommender information system

Referring to the flow chart in Fig. 2, the operation of the Surveillance and Recommender information system begins with logging into the system using the username and password that the system has recorded. Access to the system is adjusted to the user's authorization level, the highest level of which is the National Air Operation Commander-in-Chief. The Commanders of the Air Operations Command in each area of operation, the Commanders of the Sector Command, and the Commanders of the Radar Unit are also given access to the system according to their respective authorization levels. After the user enters the system, the initial display that appears or the dashboard that displays the earth map from Indonesia and surrounding countries is accompanied by submenus that can be selected as needed. Select the ``Surveillance'' submenu. The Radar antenna icon indicates that the aerial observations should be displayed. Fig. 5 shows the main dashboard. The Surveillance and Recommender information system is a subsystem of the primary information system that integrates several essential capabilities to support military and defense activities and support national interests. The dashboard page will automatically display the aircraft icons based on real-time data from the ADS-B equipment. The data is obtained through access to the Flightradar24 application with a paid Application Programming Interface (API).

Fig. 5.

Fig 5

Open in a new tab

The system's main dashboard contains aircraft data from ADS-B.

The option on the ``Radar'' feature instructs the system to display a map of the Indonesian earth and several Radar Units at random locations. This system is intended to show its superiority in black flight identification compared to the current information system. The display of the screen is shown in Fig. 6. To find out which aircraft are under the surveillance of a particular Radar Unit, click on one of the Radar icons. After clicking on the icon, a combined screen will be displayed, which is a fusion of the military Radar PPI and aircraft according to data from ADS-B, which is in the surveillance area of the Radar Unit. The fusion display of these two displays is shown in Fig. 7.

Fig. 6.

Fig 6

Open in a new tab

Dashboard features ``Radar" on the'' submenu “Surveillance.''.

Fig. 7.

Fig 7

Open in a new tab

Fusion display of PPI military radar and ADS-B.

For example, clicking the 234th Radar Unit icon will display a circle showing the Radar coverage area and a green simulated PPI on the bottom left. In addition, aircraft in the radar coverage area are displayed based on data from ADS-B. This display makes the surveillance of aircraft in the radar coverage area more focused because only specific aircraft are displayed. Interestingly, all aircraft displayed based on ADS-B data are not displayed in the simulated PPI screen even though the Radar also detects them. That means that if the simulated PPI does not display any points, no black flights will fly within the coverage area. Conversely, if the simulated PPI raises one or more points, the Radar detects the black flight, as shown in Fig. 8.

Fig. 8.

Fig 8

Open in a new tab

Simulated PPI with black flight dots; see the red circle.

The display in the simulated PPI informs about the detection of two black flights, where one aircraft appears on the main screen while one aircraft is not visible because it is covered by other aircraft from the ADS-B data. Even though it is closed, the existence of the black flight has been detected, and at a particular second, it will appear on the main screen. To find out the identity of the black flight that appears on the main screen, which is the one given a red circle, click on the object, and its identity prediction will be displayed on the main screen, as shown in Fig. 9. Based on the knowledge that AI has, the black flight is predicted to be the F-117A Nighthawk, which is the best prediction result for the F-35 Lightning II fighter alternative. From this estimate of the identity of the black flight, the system recommended that the Commander of the National Air Operations Command order the dispatch of one fighter flight to act on the black flight. The recommendations include the type of fighter aircraft with combat capabilities equivalent to the black flight and the nearest air base for the fighter's dispatch flight. Recommendations for the type and nearest airbase that has the recommended fighter aircraft are essential requirements at the speed of time in intercept black flight.

Fig. 9.

Fig 9

Open in a new tab

Identity of the black flight in the red circle.

Validation of air surveillance information systems and recommenders

Validation of the air surveillance information system and recommendations are carried out directly to Radar users, namely Air Force Officers who have served in the Radar Unit and handle the Radar system. Validation was carried out through a simple questionnaire but directly led to experience while serving in the Radar Unit, as follows:

  • 1.

    What do they think about operating the air surveillance information system, and what recommenders are associated with its benefits for the Indonesian Air Force?

  • 2.

    What are their suggestions for improving the air surveillance information system and recommenders?

The questionnaire was submitted to two senior officers, the rank of Colonel, from the radar branch who have experience handling the operation and maintenance of the radar system. Table 4 summarizes the essence of the answers to the questionnaire.

Advantages of the proposed air surveillance and recommender information systems

Based on the results of operating this information system, there is a significant difference between the standard or existing information system used in the Radar system currently operated in the Radar Units and the air surveillance information system and recommendations proposed in this study. The differences between the two information systems are summarized in Table 5.

Table 5.

Difference between the current radar information system and the proposed information system.

No. Current Information Systems Proposed Information System
1. Recognition and identification of black flights without AI Black flight recognition and identification has used AI
2. Identification of black flights is highly dependent on SSRs Identification of black flights does not depend on SSRs
3. Identification of black flights has not utilized RCS data. Black flight identification has leveraged RCS data.
4. Identification of black flights cannot be done after they are detected. Identification of black flight can be done as soon as it is detected.
5. The black flight identification procedure takes time due to the lengthy procedure. The procedure for identifying black flights is shortened at the Radar Unit, and decisions can be made immediately.
6. It is not equipped with a recommender for the basis of decision-making. It has been equipped with a recommender for the basis of decision-making.
7. The radar system is not equipped with black flight identification capability. The Radar system is equipped with black flight identification capability.
8. Separate PPI and ADS-B display. Fused PPI and ADS-B display.
9. They are not equipped with simulated PPI facilities. A simulated PPI is provided to display detected black flights.
10. The Radar system cannot identify black flights.
The Radar system		 can identify black flight.
11. There is a long delay time between black flight detection and identification. Delay time between detection and identity identification is shortened.
12. Fighter jet dispatch flights for observation from the air can be high risk. Dispatch flights of fighter aircraft observation from the air have been considered a risk factor.
Open in a new tab

Conclusion

Identification of black flights has long been a challenge for the National Air Operations Command, which oversees the Air Operations Commands, as well as the Sector Commands, which oversees the Radar Units, which are tasked with safeguarding the sovereignty of national airspace from any air infiltration attempts. The radar detection data in the form of airspeed, range or distance, altitude, and bearing cannot reveal the identity of the detected black flight. On the other hand, the length of the procedure in identifying black flights from the Radar Unit to the command above has an impact on decision-making faced with the movement of black flights in the air and dispatch flights of fighter aircraft to conduct direct observations from the air can be at high risk if it turns out that black flights are fighters armed with higher combat capabilities. In the study, a new method for black flight identification was proposed that combines RCS, altitude, and airspeed data and is equipped with an air surveillance information system and recommenders. The results of the tests on the methods and information systems show that the proposed new methods and information systems have positive prospects for development and implementation in the current military radar systems. We name our system as Intelligent Air Defense Recommender System (IADRS).

Limitations

The proposed new technique's limitation is the actual RCS data of the aerial objects listed in Table 1. However, this new technique fills a gap in the identification of black flights, which is still a challenge for the National Air Operations Command.

Ethics statements

The authors confirm that no experiments involving human subjects or animals were conducted in connection with the present work and that no data from social media platforms were used.

Credit author statement

Arwin Datumaya Wahyudi Sumari: Conceptualization, Formal Analysis, Funding Acquisition, Methodology, Project Administration, Supervision, Validation, Writing – original draft, Rosa Andrie Asmara: Software, Visualization, Ika Noer Syamsiana: Writing-Review & Editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This research is supported by the BIMA Vocational Product Research grant from the Ministry of Education, Culture, Research, and Technology of the Republic of Indonesia for 2022–2024. Thanks for the support from the Center for Research and Community Service, State Polytechnic of Malang, Indonesia.

Footnotes

Related research article: None.

For a published article: None.

Contributor Information

Arwin Datumaya Wahyudi Sumari, Email: arwin.sumari@polinema.ac.id, arwin.sumari@tni-au.mil.id.

Rosa Andrie Asmara, Email: rosa.andrie@polinema.ac.id.

Ika Noer Syamsiana, Email: ikanoersyamsiana@polinema.ac.id.

Data availability

Data will be made available on request.

References

  • 1.National Air Defense Command . Komando Pertahanan Udara Nasional; Jakarta: 2017. Decree of the Commander of the National Air Defense Command Number Kep/79/XII/2017 Regarding Permanent Procedures For Air Defense Operations; p. 105. [Google Scholar]
  • 2.Rahman H. CRC Press; 2019. Fundamental Principles of Radar. [Google Scholar]
  • 3.Skolnik M.I. McGraw-Hill; 2008. Radar Handbook. [Google Scholar]
  • 4.Choudhary M., Chauhan J. Identification of friend or foe radar. Int. J. Adv. Res. Sci. Eng. 2013;8354(2):88–92. [Google Scholar]
  • 5.Federal Aviation Administration, “Automatic dependent surveillance-broadcast (ADS-B) aviation rulemaking committee (ARC),” 2007. [Online]. Available: https://employees.faa.gov/tools_resources/orders_notices/.
  • 6.International Civil Aviation Organization, ADS-B implementation and operations guidance document, no. September. 2022, pp. 1–143.
  • 7.Des Mathurin R. Thomson CSF; 1985. Long Range 3-D Mobile Fixed Radar Thomson TRS 2215. [Google Scholar]
  • 8.F.D. Messina, “Overview of secondary surveillance radar (SSR) and identification friend/Foe (IFF) systems”, 2021. [Online]. Available: https://ieee.li/pdf/viewgraphs/overview-of-ssr-and-iff-systems.pdf.
  • 9.Moustafa A.F.K.H. 5th International Conference on Electrical Engineering (ICEENG) 2006. 2006. A survey of IFF systems; pp. 1–11. [Google Scholar]
  • 10.BAE Systems, “What are IFF technologies?” https://www.baesystems.com/en-us/definition/what-are-iff-technologies.
  • 11.Sumari A.D.W., Asmara R.A., Risman H., Syamsiana I.N., Handayani A.N., Arai K. Proceedings - IEIT 2022: 2022 International Conference on Electrical and Information Technology. IEEE; Sep. 2022. Black flight identification using radar cross section (RCS), speed, and altitude from RADAR data using supervised machine learning; pp. 377–382. [DOI] [Google Scholar]
  • 12.Affandi L., Sumari A.D.W., Abdulloh R.Wakhidah, Addin I.M.I., Kirom M.A. An artificial intelligence-based application for recognizing and identifying aerial objects based on voice input. Procedia Comput. Sci. 2024;234:19–27. doi: 10.1016/j.procs.2024.02.148. [DOI] [Google Scholar]
  • 13.Han L., Feng C., Hu X., He S., Xu X. Ballistic target recognition based on multiple data representations and deep-learning algorithms. Chin. J. Aeronaut. 2024;37(6):167–181. doi: 10.1016/j.cja.2024.01.029. Jun. [DOI] [Google Scholar]
  • 14.Touzopoulos P., Zikidis K.C. Physical optics radar cross section predictions for an anti-ship cruise missile. J. Def. Model. Simul. 2024;21(3):301–312. doi: 10.1177/15485129211033039. Jul. [DOI] [Google Scholar]
  • 15.Li H.J., Kiang Y.W. The Electrical Engineering Handbook. Elsevier; 2005. Radar and inverse scattering; pp. 671–690. [DOI] [Google Scholar]
  • 16.Zikidis K., Skondras A. Low observable principles, Stealth Aircraft and Anti-Stealth technologies introduction – Historical background of stealth aircraft. J. Comput. Modell. 2014;4(1):129–165. [Google Scholar]
  • 17.Nopriansyah A.I.Wuryandari, Sumari A.D.W., Andaruna S. SEMINAR RADAR NASIONAL 2008 Prosiding. 2008. Sistem identification friend, foe, or neutral radar menggunakan radar cross section dan kecepatan pesawat berbasis jaringan syaraf tiruan adaptive resonance theory 1 dan Fusi Informasi; pp. 81–86. [Google Scholar]
  • 18.Aero Database, “Aircraft models & Manufacturers database,” Rev. Avia. Ltd. 2024. [Online]. Available: http://www.aero-database.com/.
  • 19.GlobalSecurity.org, “Radar cross section (RCS),” GlobalSecurity.org. [Online]. Available: https://www.globalsecurity.org/military/world/stealth-aircraft-rcs.htm#google_vignette.
  • 20.U. Specs, “Aircraft Technical Data,” Ultimate Specs. [Online]. Available: https://www.ultimatespecs.com/aircraft-specs.
  • 21.Federal Aviation Administration, “Aircraft,” U.S. Department of Transportation. 2024. [Online]. Available: https://www.faa.gov/aircraft.
  • 22.EuroControl, “SKYbrary,” SKYbrary Aviat. Safet.. 2024. [Online]. Available: https://skybrary.aero/.
  • 23.J.R. Quinlan, “C4.5: programs for machine learning,” in Machine Learning, vol. 16, no. 3, S. L. Salzberg, Ed., Morgan Kaufmann Publishers, Inc., 1993, pp. 235–240. doi:10.1007/BF00993309.
  • 24.Ho Tin Kam. Proceedings of 3rd International Conference on Document Analysis and Recognition. IEEE Comput. Soc. Press; 1995. Random decision forests; pp. 278–282. [DOI] [Google Scholar]
  • 25.Fix E., Hodges J.L. Discriminatory analysis. Nonparametric discrimination: consistency properties. Int. Stat. Rev. 1989;57(3):238. doi: 10.2307/1403797. Dec. [DOI] [Google Scholar]
  • 26.Cortes C., Vapnik V. Support-vector networks. Mach. Learn. 1995;20(3):273–297. doi: 10.1007/BF00994018. Sep. [DOI] [Google Scholar]
  • 27.Hecht-Nielsen . International Joint Conference on Neural Networks. Vol. 1. IEEE; 1989. Theory of the backpropagation neural network; pp. 593–605. [DOI] [Google Scholar]
  • 28.Sumari A.D.W., Asmara R.A., Risman H., Syamsiana I.N., Putra D.R.H., Ayuningtyas A. In: Intelligent Informatics – Proceedings of Eight International Symposium on Intelligent Informatics (ISI 2023) Pal S.K., Thampi S.M., Abraham A., editors. Springer Nature; Bangalore: 2025. A handy simulated radar interface for black flight identification system; pp. 37–52. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data will be made available on request.

Articles from MethodsX are provided here courtesy of Elsevier

RESOURCES