Flight Data Recorders Past Present and Future

This article has been accepted for publication in IEEE Aerospace and Electronic Systems Magazine.
This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/MAES.2023.3335333
IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 1
Flight Data Recorders: Past, Present, and Future

Waleed Raza Graduate Student Member, IEEE, Justus Renkhoff , Olayemi Ogirimah , Richard S. Stansbury ,
and Houbing Song Fellow, IEEE
Abstract—Devices that record data generated by an aircraft On various occasions, they are referred to by different names,
and its crew during operation are known as flight data recorders such as black boxes, digital FDRs (DFDR), solid-state FDRs
(FDR) and cockpit voice recorders (CVR). Data recorded by these (SSFDR), or accident data recorders (ADR) [4].
devices enable the analysis of aircraft state and behavior from
take-off to landing. The information provided by these devices is This study of FDR serves a variety of purposes, including
used to conduct inspections, investigations, training, and other new aircraft development, training assessment, maintenance,
measures by aviation experts following hazardous events or and diagnosing accident causation. Installing an FDR in a UAV
accidents. As of today, multiple instruction manuals covering is becoming increasingly difficult, given the ever-shrinking
the specifications of FDRs and CVRs have been presented by size of the aircraft design, power, and other constraints. In
various organizations, experts, and scientists within the aviation
industry. However, no consolidated research document that addition, unmanned aerial vehicle (UAV) technologies have
brings all this information together for regulatory and technical evolved over time, and the strong interest in their use indicates
purposes has been published yet. Our objective is to carry out that these technologies will be the future of the aviation
an in-depth study of FDRs for both manned and unmanned industry for military and commercial purposes. The rapid
systems. We aim to examine the requirements used in the past, growth of UAV technology exceeded the pace of regulators,
current technological advancements, and future intelligent FDR
approaches, intending to advance FDR technology. Starting with standards development organizations, and researchers to de-
manned aviation systems, we provide a comprehensive overview velop standards, requirements, and other regulatory policies
of general FDRs and then look at new challenges and the for FDRs.
resulting requirements for unmanned aerial vehicles (UAVs). This paper contributes to the research community by identi-
After reviewing all relevant resources, we present a roadmap fying research gaps that can be solved in the future to improve
for the next generation of FDRs.
UAV FDR technologies and help regulators set standards
Index Terms—Flight data recorder, International civil aviation for UAV FDRs, their integration, operations, and analysis.
organization, Federal aviation administration, Unmanned aerial A literature review collected most of the legal requirements
vehicles, Intelligent FDRs, IoT networks
and technical aspects of FDRs, from general technologies to
artificial intelligence (AI)-based approaches for UAVs. In this
I. I NTRODUCTION way, we consider FDR solutions that either send the recorded
data with real-time monitoring and data transmission to a
I N 2019, the chief executives of Boeing stated in a testimony
to the United States Congress discussing the B737 max, “If
back then, we knew everything we know now, we would have
ground station or record all the parameters on board. We’ve
created a comprehensive overview of FDRs, from manned to
made different decisions” [1]. Although today’s aircraft are unmanned systems, and associated it with the current AI-based
designed based on information uncovered from past accidents, revolution. Some opportunities through current technologies
there is inadequate data to determine the complex events that are presented, as well as challenges that come along with them.
lead to an accident [2]. The safety of a flight depends not This paper bridges between regulatory aspects and the
only on the aircraft’s design and the onboard systems’ features technical history of FDRs. More specifically, limited literature
but also on the support provided by the ground stations, the is available connecting the FDR devices from manned systems
skills of the pilot, and air traffic control (ATC)[3]. At the to unmanned systems. It becomes necessary to associate these
same time, flight data recorders (FDR) and cockpit voice aspects and understand the criticality of FDR devices from
recorders (CVR) also positively contribute to flight safety. small-scale evolution to current standards building a single
To determine the cause of aircraft accidents, both are used. document where the readers can opt for vital information
regarding FDR studies, including requirements of parameters,
Manuscript received January 31, 2023 weight limitations, size restrictions, power constraints, and
W. Raza is with the Department of Electrical Engineering and Computer
Science, Embry-Riddle Aeronautical University, Daytona Beach, FL 32114 communication protocols. These research gaps allow us to
USA (e-mail: [email protected]). contribute to the next generation of FDR by proposing UAV
J. Renkhoff is with the Department of Electrical Engineering and Computer architectural frameworks.
Science, Embry-Riddle Aeronautical University, Daytona Beach, FL 32114
USA (e-mail: [email protected]). The remainder of the paper is structured as follows: In the
O. Olayemi is with the Department of Electrical Engineering and Computer first section, the introduction and background comprehensively
Science, Embry-Riddle Aeronautical University, Daytona Beach, FL 32114 illustrate the scope of the work, some real application usage of
USA (e-mail: [email protected]).
R. Stansbury is with the Department of Electrical Engineering and Com- FDR, and the general overview of FDRs. The paper describes
puter Science, Embry-Riddle Aeronautical University, Daytona Beach, FL the emergence of FDR over time in the second section. Later,
32114 USA (e-mail: [email protected]). the FDR’s background is illustrated in which the data ac-
H. Song is with the Department of Information Systems, University
of Maryland, Baltimore County, Baltimore, MD 21250 USA (e-mail: quisition system concept addresses estimations of parameters,
[email protected]). input data, input sensor information, and their encoding and
Authorized licensed use limited to: Sapthagiri College of Engineering. Downloaded on June 15,2024 at 06:45:01 UTC from IEEE Xplore. Restrictions apply.
© 2023 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
This article has been accepted for publication in IEEE Aerospace and Electronic Systems Magazine. This is the author's version which has not been fully edited and
decoding strategies in the third chapter. Moreover, this section was utilized by the GM recorders. These FDRs can record
includes a recording operational checklist, which defines the parameters for up to 300 hours by engraving data into a metal
quality of recording and calibration checks on measured foil with a needle. The prototype of the ARL’s flight memory
parameters. In the next part “Fourth Section”, the general unit design was based on David Warren’s1 theory of using
standards and regulations of FDR are demonstrated, which the voice wire recorder aboard an aircraft, which combines a
covers a review of the regulatory considerations of FDRs cockpit voice and data recorder. This type of data recorder can
and a comparison regarding regulatory requirements, which record cockpit voice and eight different instrumental readings
highlight some of the challenging hurdles in the industry. The per second for four hours [9], [10]. The ARL in Fisherman’s
fifth section considers FDRs of UAVs, which defines FDR Bend, Australia discovered the first black box flight recorder
requirements for a UAV and challenges while implementing in 1953.
those requirements. Furthermore, this section analyzes report The first flight recorder used by a flight crew was based
accidents from various official accident information portals on ARL’s flight memory unit, which was also the first to be
of U.S. public safety departments, i.e., the National Trans- designed as a permanent instrument on an aircraft that can then
portation and Safety Board (NTSB), accident and Incidents by help identify accidents by investigative organizations [11]. At
the Federal Aviation Administration (FAA), and the Aviation the same time, while not the first FDR, the GM device was
Safety Reporting System (ASRS) portal from the National the first to emphasize the concept of impact protection and the
Aeronautics and Space Administration (NASA). In the sixth first to implement a rigorous program of destructive testing for
section, we present a roadmap for the researchers to execute scientific purposes. The devices invented by Ryan and Warren
research for the FDRs of UAVs that includes an intelligent were not simply flight recorders; these devices were designed
FDR model, which can record all the parameters of an aircraft and used primarily above all as accidental technologies (as-
smoothly and design a real-time data transfer system with an sisting in the investigation into aircraft accidents). Initially,
enhanced quality of recording followed by the conclusion part. the FDR could record only a few parameters; for example,
the first FDR could only engrave five parameters on a non-
II. E MERGENCE OF F LIGHT R ECORDERS reusable metal foil. The Czechoslovak military received some
Soviet-made aircraft that had the first famous K2-717 (Baro-
FDRs that conform to the latest version of the (ICAO) Speedo-Graph) recorder, which could record speed, time, and
Annex 6 ”Operation of Aircraft” are classified into two main barometric altitude by engraving it with a needle on a floating
categories, the first one is ”crash-protected flight recorders,” paper strip [12].
and the second category is ”lightweight flight recorders” [5]. In 1958, the aviation industry began to enforce basic
The first category is crash-protected because it fulfills the specifications for aircraft that would facilitate the conduct of
requirements of FDR survivability of a mishap situation and engineering studies. In addition, great efforts were made to
satisfies crash requirements. The second category is generally manufacture FDRs in such a way that they could retain their
considered non-crash FDRs, which are lightweight, compact original shape in the event of accidents, for example, in the
systems sometimes referred to as Quick Access Recorders event of a fire or other major damage to the aircraft [22].
(QAR). QAR is portable using removable media, such as a Beginning in the 1960s, the Boeing 707, Caravelle, and DC-
USB drive or microSD card, to make the data accessible. 8 were the first jet aircraft to use FDRs. These devices used
Since they are lightweight, they may not meet future regu- a mechanical stylus to engrave data into a metal foil; they
latory requirements similar to the first group. ICAO, Annex 6 are called metal foil and photo film recorders (MFPFR). In
describes crash-protected flight recorders as consisting of an these types of FDRs, the data always come directly from basic
FDR and the CVR, as well as an Airborne Image Recorder aircraft sensors, such as accelerometers and control tubes, and
(AIR) and a Data Link Recorder (DLR). On the other hand, the parameters are processed internally. At the same time,
a lightweight recorder consists of a data recording system another technology was introduced that used a photographic
(DRS), an aircraft cockpit audio recording system (CARS), film and a mechanical stylus with light beams to replace the
an AIR, and the DLR [6]. The CARS are specially used for metal sheet. They could only store five to six parameters
operational purposes, such as training, analysis of incidents or in total, which means the number of parameters including
accidents, and performance evaluation. On the other hand, the magnetic heading and airspeed was also limited [23].
CVR is employed by accident investigators to determine the By 1965, as the need for accident investigation increased
cause of an accident or incident. due to the number of airplane crashes, both metal foil and
By the mid-1950s, installing or incorporating FDR systems photographic film recorders could no longer meet the demands
on commercial flights became necessary due to the lack of of examinations and fell behind. To meet the requirements,
survivors or witnesses in several serious aircraft crashes [7]. magnetic tape recorders were introduced that not only record
Two models for FDRs were designed at that time, named conversations but also improve the number of parameters that
General Mill’s (GM) Ryan Flight Recorder (RFR), a data FDR can record. The new innovation of the aviation experts
recorder, and the Australian Research Laboratories (ARL) was that the parameters were sampled, digitized, and then
”Flight Memory Unit”, a voice recorder [4], [7]. For the multiplex coded in a long frame (1 second), which means
RFR, the patent published by Minnesota State University by
Prof. James J. Ryan provides a measurement of four data 1 Black Box Flight Recorder https://www.dst.defence.gov.au/innovation/bl
parameters, i.e., altitude, velocity, g-force, and time [8] which ack-box-flight-recorder, (accessed: 03/15/2023)
Fig. 1. FDR’s Technology Devices Evolution [32] [33]
that the parameters were no longer stored in a single stream. netic FDRs. This allows up to 100 parameters to be recorded,
In this type of recording, the magnetic tape helped to record although the memory card had been made smaller, the sam-
the digital parameters, which were encoded as binary digits. pling frequency had been increased, and useful recording
Later, these types of recorders were called DFDRs[24]. performance has been achieved by increasing the recording
In the 1970s, as technology evolved, these data recorders time of the models up to fifty hours or even more. For
were replaced by SARPP-12 recorders capable of recording six sound playback, not only has the possibility of digital sound
analog parameters and up to ten discrete parameters on pho- recording been improved but the recording time has also been
tographic film [13]. It can be assumed that the improvement extended by a few hours [27].
and design of FDRs related to the sensors, signal processing The choice of aircraft and flight parameters is an essential
techniques, and recording systems have always gone hand factor that depends on the type and size of the aircraft and the
in hand with technology development [13]. The innovations equipment used. Before the 1980s, flight recorder manufactur-
and improvements in recording technology helped the FDR to ers recorded different attribute sets. Later, ICAO recommended
record numerous parameters such as from analog to digital on a list of 32 parameters that must be recorded in every aircraft
tape and then to a solid state. An FDR must meet the following and on every flight [16]. The considered parameters were air
to be considered a viable device for detecting the causes of speed, anticipated flight speed, altitude barometric, vertical
aircraft accidents: length of the recording, a group of relevant flight speed, pitch angle, flight time/course, rolling angle, and
recorded parameters, the survival after an accident, i.e., no overloads in both axes i.e., longitudinal and perpendicular. The
damage to the recorded data once the accident occurred, and recorded parameters for on-board flights were also an engine
the quick recovery of the crashed aircraft’s data records [14], revolution, position of aileron and elevator, various pressure
[15]. measurements such as fuel, oil, and hydraulic system, some
In 1985, as digital technology developed, magnetic tape discrete signal measurements of aircraft settings including the
recorders were replaced by solid-state recorders, which use status of landing gear, landing flap, the cutoff status of power
solid-state memory cards for recording. The new types of electric generators, heat, or engine fire signals, etc. [17]. The
SSFDRs offered greater resilience and reliability than mag- length of the recording depends on the recording tape, the size
of the memory, and the customer’s requirements. However, for in the electronic equipment bay or in the cockpit. Other devices
assessing the cause of the accident, the recording of the last 30 that function in the same way as the FDRs are the Quick
minutes is often sufficient to extract the information needed Access Recorders (QARs). The data acquisition system serves
[18]. as a data dispenser for both the FDR and the QAR. Also, the
To accurately reflect the situation causing the accident, QARs have the advantage of input ports that meet the standard
quick recovery and ensuring the quality of data must be of the aircraft buses including (ARINC 429) hence they
assured [19]. It is also important to consider and calculate can receive more data [30], [31]. Using scalar measurement
the negative factors of the FDR system that may affect the and aggregated data in [73], the authors performed quantile
recording or recording resolution during an aircraft accident. regression exploring the influential features that affect the
Usually, these forces are of a mechanical nature, e.g., overload, maximum vertical acceleration during landing. They employed
the impact force when hitting the terrain, the heat generated a sufficient dimension method to evaluate the performance of
during a fire, the attack of sea salt during an accident in water, the model identifying nine important features and studying
and some attacks of the aircraft itself like different liquids as how they relate to hard landings. The authors proposed the
well as fire extinguishing agents. CurveCluster+ in [74] approach which automatically identifies
The survivability of an FDR during an aircraft crash in hard landing using the QAR data. The flight parameters are
the ocean or on land is an important factor in the devel- extracted in this method to perform analysis of the data and
opment of FDRs. The software framework and the quality then provide a two-level hierarchical classification for harsh
of the data are also of great importance to the overall FDR landing which reveals the reason for the incident.
system. Additionally, the size also influences other important There is another category of data recorder named Direct
factors of FDR design, including the size, power consumption, Access Recorder (DAR) which gets the data from data man-
communication protocol, and installation on an aircraft of agement units (DMU). These recorders are designed to allow
the FDRs. We described two models in the background part the selection of parameters. This means that they not only
which emerged in the mid-1950s and provided crystallization select the parameters to be recorded, but also the recording
of cultural impulse to FDR’s study into an object of scientific mode, periodic recording, and recording triggered by certain
and institutional analysis similar to the motivation provided accidents, e.g. exceeding a predefined threshold. Usually, these
by Greg Siegel[20]. types of recorders are used for operations such as maintenance,
research, and/or flight data monitoring [5].
III. BACKGROUND ON FDR S
As described in the section above, FDRs record all im- A. Data Acquisition Systems
portant parameters during the flight of an aircraft and are Data acquisition has a significant and critical role in col-
required by law [21]. In addition, CVRs are also an important lecting data for FDRs. The concept employs data acquisition
component that records audio flight crew’s voices, as well as computers that centralize and format the data input from
other sounds inside the cockpit during the flight. In this article, several sensors and other instruments. Afterward, the data is
the research is limited to FDRs as they are the core component transferred to the FDRs through an associated digital link such
necessary to draw technical conclusions about an incident. as ARINC 573 or 717 serial link [34]. The data acquisition
As technological innovation advances over time and de- system obtains the data from various sensors and avionic
mands also increase, it has become insufficient for FDRs to systems and then delivers the data to an FDR.
process parameters internally in a limited amount of memory. To obtain a continuous flow of data from the sensors to
In order to store all the information received from the sensors, the FDRs, data acquisition units collect the data samples, and
a data acquisition unit was constructed to gather all the param- conditions and then digitize analogs signal that represent the
eters before recording. Therefore, different types of devices aircraft functions. Particularly, the data acquisition unit (DAU)
have been developed, including Flight Data Acquisition Cards system involves the sampling of parameters and their digital
(FDAC), Flight Data Acquisition Units (FDAU), and Flight encoding from actual physical values to recorded values [35].
Data Interface Units (FDIU). They collect data from various The Data Frame Layout (DFL) depends on the recording
sensor sources and then store it in the FDR, whose purpose is method used in the data acquisition system, which entails the
limited to just recording data [22]. However, this type of data encoding method, parameter location, and bit numbers that
unit can only be used in large aircraft for public transport. In were used to encode parameters[36]. Its function enables the
smaller aircraft, the purpose of data acquisition is still fulfilled conversion process from recorded value to physical value. The
by the FDRs [25], [26]. conversion function runs through a checklist of measuring and
For flight data monitoring, data acquisition technologies processing channels. Then, the calibration control reports are
provided a huge benefit [28], [29], the FDRs were employed obtained in the DFL document to complete the process for
for recording the data, and that data may be utilized in the case essential parameters. Figure 1 represents a functional block
of an accident. Data can be routed through various other types diagram and a depiction of the memory organization, which
of recorders using a data acquisition unit, which is considered every aircraft needs to carry to fulfill the requirement of reg-
unprotected. The data is sometimes stored on a magnetic tape, ulators. The parameters output by the data acquisition system
optical disk, and/or a memory card constructed in a way so that are decoded and a sequenced binary file with images in four-
it can be removed or replaced quickly. Generally, it is located second intervals is obtained. Furthermore, these frames are
divided into four different one-second frames. In a one-second TABLE I

frame or “subframe,” the data is encrypted into 64, 128, 256, FACTORS AFFECTING THE Q UALITY OF M EASUREMENTS [27]
and 512 “words” of 12 bits each using the generation of FDR Calibration Factor Consequences and Explanation
technology described above [37] [38]. An example is shown Sensor Aging Sensors can be subjected to various
in Figure 2, obtained from a document from the European environmental constraints, such as tem-
perature and high-pressure variations
Union Aviation Safety Agency (EASA) which shows that the causing the system to drift from its
main element of the record is a “word” consisting of 12 bits. initial calibrations.
Several parameters can be stored in this word. A parameter is Connecting of Additional Potentiometers or Synchronization
Devices to Analog Input transmitters may alter the electrical
often recognized by the word and the binary and/or bit number characteristics of transmitted signals in
in which it was coded. Then, these words are categorized the context of amplitude and/or phase.
into sub-frames, each containing one second of data. These Assembling and Disassem- Sensor misalignment can occur as a
bly of Mechanical Elements result of significant overhauls or when
subframes are mostly sizes 64, 128, and 256, etc., and four undergoing an FDR systems retrofit,
sub-frames construct one frame[40]. leading to potential issues with sensor
calibration.
B. Operational Verification Checklist for Recording

To improve recording quality, some authentication checks C. Calibration Process
are performed to verify the recorded parameters and to check
As described earlier, the calibration of the measuring and
the calibration of the measuring and processing channels. To
processing channels is to obtain enhanced data quality, more-
verify the recorded parameters, evaluate the overall operation
over, the quality of data is obtained after evaluating the com-
of the recording system, and improve the quality of the
parison of the parameter’s value as measured by instruments,
DFL data, a review of the recording parameters in the FDRs
with the recorded value. The calibration process of parameters’
is performed. To get the authenticated model’s performance
measuring and processing channels is provided that involves
and for validation, these checks must be performed only for
the generation of the baseline values, which then enter a
FDRs, not for other devices such as QARs and DARs. The
sensor noting the output value of acquisition devices. If the
entire set of stored information can be copied and converted
data acquisition is performed by the recorders, it retrieves the
into engineering units for further analysis using a decoding
corresponding recorder value [27].
software platform that must be programmed according to
The methods that generate the baseline values are created
the DFL document. This process may involve the check of
using one of the following strategies [40]: 1) Activation of
recorded parameters; for instance, parameter locations must
Sensors: this process includes “sensor stimulation,” which
be set according to the data frame layouts, authentication
refers to the input and sensation received when the sensor is
of conversions functions of regulatory parameters which are
activated after getting physical inputs, for instance, “calibrated
under consideration and their value range of operation, consis-
pressure on pitot-static system port.” Also, it is responsible for
tency in terms of different phases of flight, also coherence of
reproducing the electrical signal input of a sensor, performed
parameter’s pattern in several phases of flight, verification of
when sensor stimulation encounters complex situations such
unreadable data for a big length or cyclical areas, and finally
as engine temperature or different alarms. 2) System’s Auto-
sequential integrity of data recording.
Test Activation: the Inertial Reference System (IRS), whose
The calibration of measuring and processing channels is to
functions internally adjust the parameters according to baseline
calculate the data quality evaluated by comparing the value
values. 3) Instruments Indication Check: assessing all the
of the parameter determined from the instruments and the
instrument indications when there is no availability of internal
recorded value. It is necessary to calibrate the measurement
test or if a sensor feeding recording system also feeds to
and processing of individual parameters for better data quality.
cockpit indicators. To check the obtained value from the data
The conversion functions given by the manufacturers are
acquisition system, a valid or compatible read-out system is
theoretical, and their test results performance is only calculated
connected for real-time value computation by utilizing the
in the context of prototypes; hence the calibration results may
conversion functions similar to that provided by the data frame
differ from those of actual aircraft [39]. Moreover, the factors
layout document. Note particular aircraft manufacturers only
that affect the quality of measurement are described in the
offer this calibration of measuring and processing channels
following Table I:
before delivery [41].
In the end, the sensors that save data for recorders might
be different from those that feed data to the various flight in-
struments and other aircraft systems. As a result, the recorded IV. G ENERAL S TANDARDS FOR FDR AND R EGULATIONS
value is likely to differ from the actual value. Suppose the Regulators and standards development organizations define
allowable difference between the recorded value and the actual the requirements, manuals, and other resources for FDR devel-
value is too large. In that case, two actions are initiated, opment, installation, and operation. In this section, we cover
depending on the nature of the problem: The first is to replace some major regulatory organizations that develop standards for
or repair the malfunctioning elements in the system. The the FDR, CVRs, and their operation. Several versions, editions,
second is to change the conversion functions in the DFL using and a few amendments in these texts can be found having
a calibration procedure. shared goals to set the standards of the FDRs. For example,
Fig. 2. Principle of Recording [42]
ICAO Annex 6 “Operation of Aircraft” Part-I, II, and III, (and organizations and regulatory bodies to share best practices and
its editions of flight data operation) also defines the FDR safety information and take part in worldwide initiatives to
requirements, general standards, procedures, etc. These are raise aviation safety standards.
considered the principal documents pertaining to FDRs. Table
II lists a number of organizations from various countries that
contribute to aeronautical and aerospace engineering research TABLE II
mainly assisting the aircraft operators in defining the require- T HE G OVERNING BODIES , P ROFESSIONAL AVIATION O RGANIZATIONS ,
S OCIETIES , AND A SSOCIATION
ments, and other resources for FDR development, installation,
and operation, including the design of data frame layouts. The Organization/Authority Name Country with Headquarters
Joint Aviation Requirements - Operations (JAR-OPS) read-out International Civil Aviation Organiza- United Nations, Montreal, CA
tion (ICAO)
manual is an excellent source to study FDR standards at the
Federal Aviation Administration (FAA) Washington, D.C. USA
European Level; which is a joint aviation requirement for the European Union Aviation, Safety European Union Cologne,
operation of commercial airplanes. Another critical institution Agency (EASA) Germany
that defines regulations for FDRs is EUROCAE; their manuals European Organisation for Civil Avia- Paris Region, Saint-Denis,
tion Equipment (EUROCAE) France.
are used to set the regulations in Europe [27]. In the U.S. the Australian Transport Safety Bureau Canberra, Australia
FDR regulation is prescribed in Federal Aviation Regulations, (ATSB)
Part-125, “Certification and Operations”. Civil Aviation Administration Of China Dongcheng District, Beijing,
(CAAC) China
Civil Aviation Authorities Various Aviation Bodies
Joint Aviation Authorities (JAA) European States
A. Brief Survey of Aviation-related Professional Societies, Experimental Aircraft, Association USA and Canada Based
Associations, and Organizations (EAA)
Flight Safety Foundation (FSA) USA and Canada Based
To cover the technical aspect of aircraft, such as operation, General Aviation Manufacturers Asso- USA and Canada Based
inspection, and safety-related standards, ICAO Annex 6 serves ciation (GAMA)
as a complete guide including recommended best practices. International Air Transport Association USA and Canada Based
(IATA)
There are more Annexes defined by the ICAO. In this section,
Society of Experimental Test Pilots USA and Canada Based
we shall specifically focus on Annex 6 as this mainly covers (SETP)
safe operating policies for aircraft operation. Additionally, European Civil Aviation Conference Council of Europe Neuilly-sur-
Table II outlines a number of organizations, societies, and (ECAC) Seine, France
European Organization for Safety of 41 Member States Brussels,
professional organizations that work for the sake of aviation Air Navigation (EUROCONTROL) Belgium
safety, flight operation, and other space-related activities [43].
These organizations collaborate with international aviation
B. Regulatory Requirements and Operations recordings, and transcripts are provided in the ICAO
Annex 6, Part-II, 10th Edition.
Section 2.4.1 of the ICAO Annex 6, Part-II, 10th edition,
states the regulatory requirements concerning FDR operational
guidelines and other similar devices such as CVR, AIR, and C. Current Standards for the Survivability of FDRs
DLR for the issuance of an airworthiness certificate. It must The FAA and the EASA have set the current standards for
be capable of recording parameters listed in Table A2.3- the survivability of FDRs in their documents TSO C123A
1 of Appendix 2.3 for various types of aircraft based on (CVR) and C124A (DFDR)2 . These standards mandate that
weight, size, and operations [44]. An aircraft with a maximum FDRs must be capable of withstanding extreme conditions,
take-off mass of 5700 kg must be equipped with an FDR, such as high-impact crashes, intense fires (high and low inten-
regardless of the airworthiness certificate it holds. This has sities), and deep-sea submersion, to ensure the preservation of
been defined in ICAO Annex 6, Part-I, and paragraph 6.3. vital data for accident investigations. These tests include a low-
There are specific recommendations and installation guides intensity fire test, where the FDR must endure temperatures
provided in attachment D to Annex 6, which describe the of 260°C for 10 hours, as well as an impact shock test, where
rules as follows: 1) the recording requirements, such as the the FDRs must withstand 3,400 Gs for 6.5 milliseconds in
number of parameters, data quality, etc., must be verified by all directions, and a static crush test, where the FDRs must
standards that are approved by certifying authorities. 2) the resist 5,000 pounds of pressure for 5 minutes on each axis.
manufacturer needs to provide operating instructions and pro- In addition, FDRs must undergo fluid immersion tests, where
cedure of installation, parameters source or origin 3) equations they are submerged in aircraft fluids, such as fuel and oil,
that relate counts to units of measurement followed by the for 24 hours, and water immersion tests, where they must
manufacturer’s test reports to the certifying authority. The remain submerged in seawater for 30 days. The FDRs must
equipment or instruments with their installation should be also demonstrate penetration resistance, as they must withstand
viable to the State of Registry (The State on whose register a 500-pound weight dropped from a height of 10 feet with a
the aircraft is entered). Among them, section 2.4.16 ”Flight 0.25-inch-diameter contact point. Lastly, FDRs must exhibit
Recorder” is dedicated to features and characteristics of FDR hydrostatic pressure resistance to remain operational at the
equipment, defining the essential information and providing a equivalent of depths up to 20,000 feet.
few notes for the FDRs [45]. For instance:
• The crash-protected flight recorders should consist of at D. Challenges
least one of them; a complete FDR, a cockpit voice Depending on the application, the characteristics of an
recorder, an airborne image recorder, and a data link aircraft change. New types, designs, weight classes, and areas
recorder. As per the Appendix provided in ICAO, Annex, of use are constantly emerging, all of which have to be taken
the images and information of data link possibly be into account in the requirements for FDRs. Nevertheless, the
recorded on the CVR either or the FDR. requirements should still be implementable and future-proof.
• Similar to that of Crash-protected FDR, lightweight New challenges are constantly arising, such as the use of
recorders must be comprised of one or more of an air- AI in aviation, urban aircraft mobility (UAM), or unmanned
craft data recorder system (ADRs), cockpit audio record- aerial vehicles (UAVs) in the national airspace, which must be
ing system (CARS), airborne image recording system taken into account and, if necessary, result in a reevaluation
(AIRS), and data link recorder system (DLRS). The data of the current requirements for FDRs. Since an aircraft is
link information can be recorded on either one of the sometimes used over long distances and also across national
following i.e., ADRs or CARS. borders, care must also be taken to ensure that the different
• Complete information regarding FDR guidelines and reg- requirements of the various jurisdictions and organizations do
ulatory requirements can be found in Appendix 2.3 of the not conflict as far as possible. A significant challenge exists
ICAO, Annex 6, Part II. with respect to data volumes becoming more extensive and
• The EUROCAE ED-112, EUROCAE, ED-56A, larger due to new technologies like artificial intelligence or
EUROCAE-ED-55, and minimum operational new sensor technologies. UAVs, in particular, however, are
performance specifications (MOPS) are the primary expected to become smaller and smaller. As a consequence,
documents illustrating the specifications that are it is a significant challenge to provide enough memory and
applicable to crash-protected flight recorders for those to use it efficiently. This includes, for example, optimizing
aircraft for which the certification application is the recording interval, and recording resolution of different
submitted to a Contracting State earlier than January 1st, sensors.
2016. The section EUROCAE ED-112A also describes
the same content for crash-protected flight recorders on V. F LIGHT DATA R ECORDER FOR A UAV
or after 1 January 2016.
This section explains the importance of a UAV and its
• The ED-155 of the EUROCAE section shows the spec-
benchmarks using its history, civil and commercial applica-
ifications regarding the lightweight flight recorder in the
tions, future requirements, and a review of reported accidents
MOPS or the equivalent document.
• After November 7, 2019, contains requirements for 2 Department of Transportation, Federal Aviation Administration Aircraft
States; concerning the usage of voice, images, data Certification Service-TSO C123A, C124A
in the United States from three different agencies to gather as a remotely controlled (RC) aircraft. However, it differs
data that will allow us to understand which flight parameters from an RC aircraft in its complexity, flight control system
and sensors are essential for lightweight FDR technologies. requirements, the use of onboard sensors, and, as mentioned
earlier, the ability to fly beyond line of sight (BLoS). UAVs are
A. Motivation and Importance Regarding FDRs for UAVs also considered flying robots due to the above characteristics
and have various levels of automation that conventional robots
Unmanned aerial vehicles (UAVs) are often known by
do not have [48]. Table III lists the applications of UAVs
various names, such as Unmanned Aircraft Systems (UAS)
for civil and commercial purposes that are needed in today’s
and/or Remotely Piloted Aerial Systems (RPAS). Over the
world.
past decade, UAVs developed by academia and industry have
The UAVs can be further divided into two subcategories:
added new system complexities and introduced more data
fixed and rotary. For military missions, fixed wings are often
sources to be captured and stored by an FDR or CVR. The
employed because they can carry a larger payload and have
question now arises here, How can we ensure the safe and
similar characteristics to that of traditional airplanes. The term
efficient management of new data while meeting all neces-
“rotary-wing aircraft” refers to aircraft that generate lift by
sary requirements and avoiding any negative consequences?
means of wings or blades rotating around the aircraft, attached
Traditional manned aviation already has existing standards for
to a mast. In the case of UAS, several of these rotors often
FDRs and CVRs as discussed above; however, these standards,
provide the necessary lift and stability. These aircraft often
under many contexts, are insufficient in enabling the ability
fulfill similar tasks as fixed-wing aircraft. However, they can
to diagnose the cause(s) of a UAV’s mishap or accident. A
also perform functions in urban or other areas where fixed-
considerable increase and advances in computer technology
wing aircraft have difficulties navigating. Fixed-wing UAVs
have allowed the manufacturer to design a novel kind of FDR
typically have higher speeds than rotary-wing UAVs, and like
that can record all the parameters per the recommendation of
their manned counterparts, have a more extended flight range.
the ICAO and others, despite the fact that the flight hours
Today’s UAV manufacturers offer a variety of sizes, pay-
have become longer. These new FDRs are often referred to as
load capacities, and ranges depending mainly on their area
”next-generation” or ”extended-range”.
of applications, mission length, wingspan, and a number of
A recent IEEE conference publication [46] focused solely
sensors. Figure 3 illustrates the classification of UAVs. We
on traditional aircraft raised the question for researchers and
classified them based on some parameters such as size, weight,
aviation experts: “which flying parameters are more helpful
application, mission, and user. Figure 4 shows the typical
for recording.” They provided an idea to construct two sets of
protocol of the UAV system obtained from the University of
parameters; the first set they regarded as a pilot awareness set,
York document [47].
and the second one is a traditional data set used to help experts
and investigators determine the cause(s) of a mishap, incident, UAV technology is increasingly needed for a variety of
or accident. The first set of parameters requires a great deal of applications, especially those where the environment is in-
effort to study past accidents and then build a model/system to accessible to humans and/or deemed unsafe. Assuming that
ensure that we have a sufficient number of recorded parameters human pilots in manned aircraft are exposed to danger re-
that could help identify the pilot’s current situation during gardless of the flight or mission, using UAVs in missions can
flight. For UAS, this is not the case, as there is no pilot onboard help reduce risk factors and keep human pilots out of harm’s
replaced by a remote pilot and/or automation. Looking at the way. Considering the safety of manned aviation systems,
growing number of UAVs and their applications, this will most these systems are being developed more carefully and at
likely also become a necessity for UAVs during the following a higher cost than UAVs [48]. Designs frequently seek to
decades. ensure UAV survivability, but this is not always practical for
Given that manned and unmanned aircraft are designed, small or medium-sized UAVs. There exists a trade-off between
employed, and maintained differently, a minimum standard unmanned and manned systems in terms of unmanned system
parameter set is required for UAVs that can contribute to size, range, cost, payloads, specifications, and mission scope.
flight diagnosis. While each attribute can be valuable toward Table III illustrates civilian uses in various situations.
determining the cause of a system failure, but recording
all features typically increases the size, weight, and power C. General FDR requirements of UAV and Parameter Classi-
requirements of the FDR, which would limit the platforms fication
capable of carrying it. To address the challenges, the following
The requirements for safe UAV operation are essential in ev-
few sections will provide further analysis of the needs of FDRs
ery application domain. The rapid development of many appli-
for UAVs.
cations requires specific guidelines, regulations, and standards
depending on the type of UAV and its use. The development of
B. Introduction of UAV Applications requirements for an FDR must take into account different prop-
UAVs have become very popular in today’s world as they erties, such as the type of UAV, the size of the UAV, computing
can be deployed in environments inaccessible to humans power, the quality of the data, the application area, the survival
and perform a variety of operations/missions The potential standards, and more. Comparing a conventional aircraft and
applications are limitless: UAVs can be used for defense and a UAV, the FDR in a manned system must differ from that
civilian purposes. We can consider a UAV in the broadest sense of a UAV. ICAO Annex 6, Operation of Aircraft Volume III
Fig. 3. Classification of UAVs[48]
outlines the requirements for traditional aircraft systems. The ure of this equipment, which includes electric, hydraulic,
question now arises whether the parameters are identical for navigation, etc.
UAVs. To answer this question, our study conducted a brief • Configuration Data: This category addresses the storage
literature review to determine the requirements for the FDR of of UAS configuration parameters such as aileron de-
a UAV. In addition, the ICAO Type I, Type II, and Type IIA flection, rudder, elevator, landing gear position, etc. For
Annexes describe that an aircraft must record parameters such example, if a fly-by-wire (FBW) control system is used,
as aircraft configuration, flight path, operation, altitude, engine the key value of each phase for every channel is recorded
power, and speed (for a list and aerodynamic drag devices, see as well, to show the information concerning the overall
Type, II, and Type II-A) to determine the flight characteristics response of the control system.
of the flight [43] [44]. • Peripheral Information: This category provides records
By regarding the selection of parameters for manned sys- and remarks about flight data such as GMT, flight index,
tems, aviation industry experts have enough data from accident and takeoff weight, etc.
reports. If maintenance experience, rationality, systematics, FDRs are crucial to be installed for training, maintenance,
and a complete set of parameters are available, then the accident investigation, and for the design of future FDR
selection of mandatory parameters for recording purposes is generations, including airworthiness certification. However,
achievable since there is real-time evidence [48]. According to the calibration process of FDRs for small UAVs is often
a few publications in the proceeding of the First Symposium neglected and there is a lack of standardization for parameter
on Aviation Maintenance and Management [50], we can divide sets.
these parameter sets into the six classes shown in table IV.
UAVs are not only utilized for military operations but also
Descriptions of these classes are as follows: for various civilian missions, leading to potential conflicts
• Aerodynamic Parameters: Environmental or surrounding with general aviation platforms. FDRs play an important
conditions can be obtained from aerodynamic parameters, role in this aspect, as they help to meet safety and main-
such as the motion of the flight. tenance requirements through the parameters they monitor.
• Motor/ Engine Data: We can deduce the operational Table IV provides a classification of these parameters (sample
status and the engine’s current status from this class, for parameters), allowing for monitoring of the UAV’s status and
instance, engine RPMs, engine temperatures, etc. equipment, as well as assisting with training and investigating
• Control Input Data: This category provides the total input accidents. However, not all of the parameters listed by the
from the pilot or UAV’s flight control system. ICAO’s Annex 6 apply directly to UAVs, and further research
• Status and Alarming of Airborne Equipment: The data is required to establish standards for UAV parameters, the
coming from this category deals with the status and fail- actual range of parameters, and recording intervals, etc., to
TABLE III TABLE IV

L IST OF UAV C IVIL AND C OMMERCIAL A PPLICATIONS [49] C LASSIFICATION OF FDR PARAMETERS IN S IX C ATEGORIES [50]
Category Applications Aero- Motor/ Control Status and Configur- Peripheral

Maps, Surveying, and Aerial Survey Structural Technologies, Real dynamic Engine Input Alarming ation Informa-
Photos Estate, Bathymetry, Archaeology, and Parame- Data Data of Air Data tion
Heritage ter bone
Infrastructure Monitoring of Pipelines, Bridges, and Equip.
Building Inspection with their Support, Altitude Rotor Stick In- Electric Aileron GMT
Roads, Railway Tracks, Canal Inspec- Speed put Deflec-
tion, Power Plants, Support in Con- tion
struction, and Landfills Airspeed Gas Flap/Slat Hydraulic Rudder Flight
Agriculture Physical Interaction with Farming and Exhaust Control Index
Cultivation, Harvesting, and Planting Temp.
Law Enforcement and Public Rescue Missions and Search, Counter- Angle of Fuel Speed Navigation Elevator Weight-
Safety Privacy, Private security, Firefighting, Attack Con- Brake Takeoff
Fires in Forest, Border Patrol, Environ- sump-
mental Protection, Medical Support tion
Emergency Response Quick Response in Natural Disasters, Rate
Artificial (Man-made) disasters, Usage Angle of – – Environment Landing Synchro-
in Epidemics Slideslip Control Gear nization
Transportation and Logistics Cargo and Food Delivery and Private Position
Transportation ... ... ... ... ... ...
Mining and Energy Eng. Exploration of Natural Fossil Fuels,
Support in Mining
Communication Employment in Communication Re-
lays, Conventional Wireless Commu- 2) Size Restriction in Small UAVs: FDR in small UAVs,
nication Networks, TVs Broadcasting,
Network Coverage, Internet, and Other micro and mini UAVs, poses a challenge. An alternative
Support in Emergency solution is to implement a real-time data transfer mechanism
Journalism and Cinematogra- News Coverage and Motion Photogra- to receive flight data, requiring seamless network connectivity.
phy phy
Wildlife Management Protection of Animals with Counter-
Another possibility is to develop a nano FDR that records
Poaching, Pest Control, Monitoring of minimal operational standard parameters of UAVs.
Bird Lives and for Scientific Research 3) Weight and Payload: In the preliminary design phase of
on Animals, and Monitoring of Sea
Creatures.
a UAV FDR, it is important to consider the weight require-
Hobbies, Arts, Research Activ- Recreational Activities, Flight Re- ments, including the take-off weight and installed payload,
ities search, UAV as a Hobby, photography for both fixed-wing and rotary UAVs. For smaller UAVs,
Miscellaneous Missions Other Missions in Safe Operations of the number of parameters can be limited, and a lightweight
Civil Airspace
FDR may suffice. On the other hand, installing an FDR with
a parameter set may be mandatory for larger UAVs with a
high take-off weight and payload capacities to ensure safe
check more parameters, please refer to ICAO Annexes and
and efficient operation, that complies with regulations and
EUROCAE parameter characteristics for flight data recorders
provides critical information. Thus, a lightweight FDR can
in [44] and [45]. The classification in this study is based on
be considered in the design phase.
our findings and previous literature.
4) Airspace Regulatory Requirement: The requirements of
airspace access for UAVs must also specify the FDR require-
D. FDR Design and Operation Requirements ments of the air vehicles operating within those airspaces.
These related regulations must be proposed within the given
A growing need for UAVs in various industries requires course to meet the growing application needs of UAVs. The
establishing FDRs design and operation requirements for these ICAO, FAA, and EUROCAE continuously collaborate to set
aircraft. This subsection aims to explore certain considerations the standards regarding the UAV regulatory aspect consider-
that must be taken into account while designing and operating ing the UAV design type, altitude, frequency spectrum, and
the FDRs of UAVs, which include technical requirements, airspace operating permissions. In this regard, two surveys
regulatory guidelines, and potential challenges described as provide the current state of regulation concerning UAV op-
follows. erations [51], [52].
1) Power Constraints: Long-range UAV missions often 5) Encoding and Decoding Strategies: The parameter en-
require significant stored energy to ensure successful comple- coding and decoding are performed in the DAU unit, where
tion. There is always a trade-off between various characteris- all parameter sampling processes of the actual physical and
tics, such as range, payload, altitude, communication power, recorded values occur. The DFL document outlines the record-
and mission endurance. Power requirements must be assessed ing methods, parameter locations, encoding methods, number
to ensure safe UAV operation and real-time data transfer. To of bits for encoding parameters, and types of encoded data.
meet these requirements, either a lightweight FDR with data The process is then reversed for decoding purposes. During
recording capabilities and a set number of parameters can be this review, new encoding methods may be implemented to re-
implemented, or a real-time data transfer strategy must be duce the computational complexity of the algorithm employed
developed to transmit the data. by the FDR.
Fig. 4. Conventionally UAVs Protocol by Evans AR[47]
6) Quality of Data Recording: The quality of data record- The decisions of autonomous systems, often implemented with
ing plays a crucial role in flight data analysis by the operators. technologies such as neural networks, are black boxes, making
Regarding the data quality, ICAO Annex 6, Part-I refers to decision interpretation substantially difficult, which means that
“the remedial action must be taken if the recording of the the set of recorded attributes must be sufficient to interpret
mandatory parameter is not recorded correctly”. Also, the and comprehend decisions made by the system’s automated
ICAO Annex 6 [45], EUROCAE ED-55/ED-112 and ED-112A elements.
specify that the “significance period of poor quality data” can Figure 5 depicts the need for installing an FDR in different
lead to severe consequences while analyzing the data 3 . This is types of UAV classes. Depending on the size of a UAV, for
the same case that can be applied to UAVs while recording the instance, micro, mini, very small, and small UAV types, the
data; the data quality standards should be similar for UAVs. data recording requirement can be fulfilled by a minimum set
The poor data quality can be due to loss of information, of parameters due to its size and power limitations. Because
overlapping of the time period, or anachronism. For the UAVs, there is commercial employment of UAVs where it is not
as the number of vehicles will be in millions, the standards applicable to install the FDR onboard hence real-time data
for the quality of data must be taken into reconsideration. transfer scheme can be introduced, which can provide us
7) Requirements Based on Autonomy: Many different sub- with the data at the distributed sensor network node. We can
systems establish the flight capability of a UAS. These systems integrate the features of FDR in the flight control system of
perform tasks such as detection and avoidance, flight planning, UAVs both can work as a common goal for the command
fault identification and recovery, and many others. NASA center and data transfer purposes. On the other hand, it is
presents [53] multiple domains where autonomous processes fully suggested to integrate the FDR in large UAVs such as
can be applied and would benefit the overall system. Depend- Medium, Large, and other UAVs sizes. The integrated Internet
ing on which and how many of these tasks are performed of Things (IoT) devices in today’s world are proven to be the
autonomously, a UAS can be categorized into one of sev- best solution for this purpose having better energy efficiency
eral levels of automation, according to NASA [54]. These and robust communication links among the overall network
autonomous systems present new challenges for FDRs as [56], [57].
additional data needs to be recorded compared to conventional
aircraft. In general, it can be stated that all data used to E. Current FDR Technologies for a UAV
make an autonomous decision should be recorded. This usually
includes sensors for sensing and surveying the environment, At the moment, the lack stands for parameters set for UAVs
such as LiDAR, ultrasound, and cameras. This results in new despite efforts by aviation governing bodies to identify the
challenges. Current estimates suggest that autonomous cars parameter requirements of FDRs for UAVs, the significance
will generate up to 3 Gbit/s at low autonomous levels and in the aerospace industry, and research on FDR standards that
up to 40 Gbit/s at high autonomous levels [55]. Even if the are lagging behind. The lack of a standard parameter set for
use case is different, these values can also be applied as a UAVs can make it difficult to compare data from different
rough estimate of the data volumes of autonomous UAVs. UAVs or to conduct meaningful research on FDR standards.
Additionally, without standard parameters, it can be difficult to
3 NPA 2019-12 - Installation and maintenance of recorders – certification
ensure that all necessary flight data is being recorded, which
aspect, EASA, November 13, 2019 https://www.easa.europa.eu/en/document can compromise the safety of the UAVs. The reason might
-library/notices-of-proposed-amendment/npa-2019-12 be UAVs’ unmanned nature hence the need for associated
Fig. 5. Need for FDR based on Classification of UAVs
research get less compelling and there is no crew on board, the lightweight recording device that can be installed on manned
difference among type, is primitive usage (if compared with and unmanned vehicles. Additionally, this can be integrated
today’s UAVs) except for military missions. Hence, forming a into a maritime platform that records all the data from the
standard be challenging and subsequently requires less atten- sensors and control commands, such as servos, ECU, etc.
tion than its counterpart. Preferably, the rapid development and Similarly, it can record the instruction or control commands
mentioned technological challenges constitute less attention acquired from the GCS.
to the regulators to define the FDR parameter identification Other nano, macro, and small recorder devices are provided
set. Therefore, a comprehensive set of recording parameters is by DTS-Ultra Small Slice [60] recording devices designed
required in order to enhance safety and serve various purposes, especially for small drones. These devices are modular in
such as airworthiness development, accident investigation, and design and include a microprocessor, excitation, and full
training. The current lightweight FDR technology is provided signal conditioning. They can survive accidents, handle vibra-
as follows with their specifications. tion, operate in various weather conditions, and provide data
Lilium Jet chose to use the L3Harris Extra Light Data recording in low-pressure environments like high altitudes.
Recorder (xLDR) for electric vertical takeoff and landing These devices meet military standards such as MIL-STD-
(eVTOL) aircraft [58]. The eVTOL is an aircraft configuration 810E, a globally accepted manufacturing, and environment
that uses electrical power for vertical hover, takeoff, and testing standard. The environmental specification for this type
landing lies under the UAM potentially designed for urban air of device is operating temperature ranges between -40°C to
transportation. The standard version of FDRs manufactured by 60°C (-40°F to 140°F), an altitude of 15240m (50000 ft) at
L3Harris are used in fifteen aircraft models because they are -40°C and the vibration range is random, which exceeds 810-
lightweight and meet the requirements and recommendations E standard vibration. They can be employed in various UAV
of the global governing bodies. Larger aircraft typically use applications.
two FDRs, such as the CVR and the FDR. However, on The Hensoldt Group has developed a lightweight crash
an eVTOL, only the FDR is used to collect data from the recorder (LCR) [61] that has received European Technical
sensor due to the vehicle’s smaller size. Crashworthiness is Standard Order (ETSO) certification for aircraft equipment
an essential factor in assessing survivability in the event of from the European Aviation Safety Agency (EASA). These
a crash. The xLDR can withstand pressure depths of 6000 lightweight recorders, such as SferiRec LCR-100, are suitable
meters, operating temperature -55°C to +70°C, static crush for lightweight aircraft such as UAVs, drones, business jets,
1000 lbs., and fire protection up to 1100°C for fifteen minutes. and helicopters due to their low weight, volume, and power
In designing the eVTOL, careful consideration must be given consumption [61]. Previously, data acquisition and recording
to the aircraft’s weight along with all other components. The functions were distributed among different devices. Still, these
xLDR is designed to survive in the event of an accident. functions are now performed in a single device containing the
FDR-01 [59], designed by UAV Navigation, is another complete autonomous operation package. Because of its size
and lightweight, it has become possible to fit it into small

aircraft. A high-resolution camera (Flight Eye-FE 320) is in-
tegrated into the lightweight recording devices in cooperation
with KAPPA Optronics GmbH [60].
F. Survey of Reported Accidents

This section examines reported aviation accidents from three
well-known organizations in the U.S., which typically record
all fatal accidents and survival rates for drone accidents.
One can report accident and incident data to the National
Transportation and Safety Board4 (NTSB) or the FAA5 if one’s
drone crashes, and there is also a voluntary portal, the Aviation
Safety Reporting System (ASRS)6 , from NASA. The informa-
tion obtained from the following three sources is evaluated to
assess the importance of incorporating a lightweight FDR into
UAVs.
Fig. 6. UAV Accidents and Incidents in the World and U.S. per Year according
While experts can identify many potential hazards, there to NTSB
could be a variety of causes (primary, secondary, and tertiary)
that can contribute to a UAS incident or accident7 . These
UAVs sometimes crash due to programming or human error,
but also due to weather events, high winds, lithium-ion poly-
mer battery (LiPo) fires, and other technical problems8 .
1) National Transportation and Safety Board (NTSB):
The NTSB is an independent public safety department for
investigating traffic accidents in the U.S. that determines the
causes of accidents and makes safety recommendations. The
NTSB also investigates marine, rail, pipeline, and hazardous
material transportation accidents. The responsibility of the
NTSB includes preventing accidents and improving transporta-
tion safety. It also issues guidelines and protocols for safety
recommendations to prevent the recurrence of an incident or
accident. In this article, we have analyzed aviation accident
data to summarize the number of reported accidents and
evaluate the significance of this study [62].
Figure 6 shows all accidents involving UAVs worldwide
and in the U.S. The data was taken from the NTSB’s online Fig. 7. UAV Accidents and Incidents per Year according to the FAA
accident portal. In 2006, only one accident was reported, which
has been common for many years. It can be seen that the
number of reported accidents has been increasing since 2018. 3) Aviation Safety Reporting System by the NASA: The
2) Accident and Incident Data by FAA: The FAA continu- NASA-administered ASRS also referred to as the “NASA Re-
ously conducts and sponsors research to enhance commercial port” or ”NASA Form,” has introduced a voluntary participa-
and recreational aviation safety. To analyze accident history, tion portal for the reporting of aviation accidents. This program
data is extracted from the Accident and Incident Data portal by is designed to ensure and improve aviation safety. The ASRS
the FAA. Figure 7 shows the number of accidents from 2010 to plays an important role in contributing to improving safety
2014. We found limited data on this portal to depict the number and preventing accidents. Data is collected from operators,
of accidents. In 2011, there was a noticeable spike in the pilots, and others involved in aviation accidents. In addition,
number of reported accidents. Fortunately, in the subsequent the data is available to the FAA, organizations, and the aviation
years from 2011 to 2014, the number of accidents decreased industry for scientific purposes and is used to promote aviation
[63]. safety worldwide. Therefore, we have collected their data to
determine the number of reported accidents for our studies.
4 https://www.ntsb.gov/Pages/AviationQuery.aspx The FAA occasionally gathers accident data from the Avi-
5 https://www.asias.faa.gov/apex/f?p=100:446:::NO:446 ation Safety Reporting System (ASRS) portal. This collab-
6 https://asrs.arc.nasa.gov/,
(accessed: 02/10/2023) oration helps ensure the reporting of any aviation safety
7 An Overview of Human Factors in Aviation Maintenance http://www.atsb incident or situation in which aviation safety may have been
.gov.au/media/27818/hf ar-2008-055.pdf
8 Operation and Certification of Small Unmanned Aircraft Systems https: compromised. This program is similar to the Voluntary Safety
//www.regulations.gov/document/FAA-2015-0150-4721 Reporting Programs (VSRP) but differs from its counterparts
Fig. 9. FDRs’ Improvement over Time
introduced to reduce the number of maintenance tasks where

the data is gathered from various data recorders. This will
simplify gathering and analyzing data and make it easier to
analyze technical conditions using recorded data. After an
aircraft has landed, the recorded parameters are transferred
by radio to the servers chosen to gather this information
and stored at a specific airport. This procedure would be
Fig. 8. 50 Latest UAS Accidents and Incidents according to the NASA automatic and remove the need for service workers (in the
(reported May 18, 2022)
case of autonomous systems) to act as an intermediary and
the time required to complete the data analysis. To fulfill this
in structure [64]. Figure 8 shows the fifty most recent accidents requirement, the aircraft must establish reliable and robust
in the U.S., as shown in the NASA ASRS system portal, communication continuously with the ground base nodes of
which includes a monthly report from June 2021 to January the networks. These kinds of networks are required across the
2022. The third quarter of 2021 shows no accidents of a globe. This kind of network already exists and is regarded as
UAV. However, from October 2021, the number has increased the IoT network operational in various applications sectors.
significantly. Especially in October, November, and December, The above-mentioned suggestions do not cover all potential
an exceptionally high number of accidents can be observed. solutions for the general aircraft category. However, they
This could be due to more challenging weather conditions in support the premise that there is room for improvement in
the fall and winter. the design of recorders, particularly for the software used
to track the values or behavior of parameters in ”online”
mode. For instance, an airborne mission data recorder can
VI. P ROPOSED FDR F RAMEWORK , O PPORTUNITIES , AND
directly provide the parameter data from an aircraft directly
C HALLENGES
to the ground server. Moreover, the recorders’ development
This section illustrates the challenges to FDR technology will progress in response to newly developed requirements that
that require airlines to upgrade existing FDR technologies will emerge in connection with the already existing technical
for a number of reasons, including meeting new regulatory possibilities after examining the historical impact of demands
requirements, addressing obsolescence, weight, size, and per- on modifications in recorders (including their software).
formance issues, or collecting additional data. As mentioned
earlier, this is not a problem for general aircraft technology, but
more data is needed for UAVs, so changes need to be made to B. Data-Driven FDR Technology for UAVs
the FDR considering the modularity and scalability of UAVs.
FDR’s massive amount of data has challenged aviation
Consequently, a framework for designing a lightweight FDR
experts regarding accessibility and analysis. To tackle this is-
for UAVs is proposed to address parameter requirements and
sue, modern data-driven frameworks are employed to perform
other challenges.
data mining, an analytical process aiming to uncover patterns
To make modifications to FDR technology, a proposed
and relationships within large datasets. Data mining aims
framework must follow a strategy depending on the needs and
to support the knowledge discovery process, where funda-
nature of the aircraft. Also, while designing the newer version
mental insights are extracted and transformed into actionable
of FDR it must follow the characteristics illustrated in figure
information using AI/ML technologies [65]. These techniques
9, which demonstrates the characteristics from older FDR
enable aviation experts to effectively analyze the vast amounts
technology to the newer version hence collecting adequate data
of data generated by FDRs and make informed decisions
for investigation and meeting the standard requirements.
based on their findings. This section presents a unique design
for constructing a UAV FDR that meets the requirements
A. General Aircraft FDR Technology outlined in the previous section, including parameter specifi-
Opportunities exist for improvements to aircraft onboard cations, power limitations, data integrity, and communication
recorders. A single data analysis program or database can be protocol design. Furthermore, we demonstrate how machine
Fig. 10. Distributed Autonomous IoT network for UAVs
learning approaches can be implemented to enhance its per- overall different from classical servers. There are enormous
formance. The authors in [75] employed Gaussian Mixture benefits of using agent-operated programming methods, such
Model (GMM) based clustering to digital flight data which as reduced network load (i.e., it does not need a connection
detects some flights with unusual data patterns. These different all the time with the user), which means agents can reconnect
flights mean there’s more risk because regular flights tend to to their creator when the network is available, a decentralized
follow similar patterns, while unusual ones stand out. Aviation structure, and asynchronous processing.
and safety experts can review these patterns in the flight in These agents need to work together to form a multi-agent
depth which provides the risk assessment and they named this system because a single agent can’t handle complex problem-
approach ”Cluster AD-DataSample”. solving. Thanks to the distributed nature of networks, they
1) Intelligent FDRs for UAV: Based on [66], [67] and [68], work to solve the problem of data cooperatively. The knowl-
we propose an FDR framework for UAVs and highlight the edge is available and distributed among participant agents
significance of machine learning algorithms in contributing in a multi-agent system via node-to-node communications
to knowledge discovery (KD) and its potential as a promis- networks.
ing AI approach. Additionally, we discuss the importance Participant agents typically have their own goals, and con-
of designing software agents, also known as intelligent or tact is made only when the information exchange is required or
autonomous agents, in various computing environments. These it may contribute to solving the problem of the overall system.
agents function independently or collaboratively with other In the context of FDR’s comprehensive mechanism, three
entities in the system and possess the unique ability to kinds of agent layers collaborate, the first one is the general
employ learning and intelligence to achieve their assigned agent layer, the second is the task handling layer, and the third
tasks or goals autonomously. The behavior of these agents is the database layer. In the general layer, individual agents
can vary based on changes in the environment and can adapt such as electrical, mechanical, and decision agent, interacts
dynamically. Today, software agents are widely used in various with the flight control system of UAV, and DAU agents. The
applications, including autonomous driving cars, traffic control task-handling layer is a communication medium responsible
systems, autonomous missions, email filters, and complex for all communication between agents. The database layer
environments. saves the information every agent utilizes and the task handling
Programming frameworks that use software agents differ layer.
from other models, such as remote procedure calls and client The single agent-based system can be easily connected to
models used in distributed processing. Autonomous agents can a multi-agent subsystem for developing multi-agent systems.
provide various services simultaneously, adopting the server Most complex systems are distributed in nature and work
role while acquiring services at other times, hence acting together via node-to-node communication, and the knowledge
as a client. These agents also possess autonomy, which is needed to solve a problem is distributed among the agents.
Each agent can contribute to a coordinated solution for the 2) Communication Protocol Design: The need for robust
overall distributed problem. It is possible only when the par- and reliable communication emerges in the design of in-
ticipant agents are fully cooperative and communicative with telligent FDRs. UAVs have already integrated transmitters
each other to solve the commonly distributed problem similar and receivers, which are used by various ground or airborne
to the approach used in [66] but for UAVs. We believe that control stations to control UAVs. Numerous techniques have
a multi-agent-based approach will be suitable for intelligent been suggested for designing the communication protocol for
FDR design because it has several benefits over a single agent- UAVs [70], which are not fundamentally different from current
based system, as UAVs can cooperate with each other in terrestrial communication networks, provided they are imple-
airspace for data and knowledge sharing. Table V provides the mented correctly. Various types of networks can collaborate
KD methods for constructing an intelligent FDR model. In this to construct the IoT network, from satellite to underwater
article, we have created an intelligent FDR model based on a acoustic communication networks. The question arises, “how
learning algorithm and employing KD techniques, which is a do these communication networks cooperate in the design
database of previous flights. We have provided a framework of UAVs, whether for military or commercial purposes, in
for this model. This will help a UAV make decisions by using aiding the real-time data transfer and current status of UAVs
the data given to the model. responsively”. The aeronautical telecommunication network
Initially, data can be gathered from both flights with and (ATN) is intended to deliver full duplex, high data rate,
without incidents. This data can then be analyzed to identify and seamless connectivity among the aerial vehicles, aircraft,
specific sequences of events or patterns that may indicate ground control station (GCS), Aeronautical Administrative
the causes of unsafe flight conditions. As described in the Communication (AAC) and various aviation-related services
literature [66], this technique can also be applied to unmanned suppliers Aeronautical Operational Control (AOC), Aeronau-
vehicles, eliminating the need for an onboard pilot. Implement- tical Passenger Communication (APC) and local governing
ing a real-time data transfer system is crucial in this scenario, authorities which provide ATC and flight information services
as it can detect previously unknown occurrences or events (FIS). These types of networks cooperate to form a common
during flight, thereby providing early warnings to the UAV. ATN, which is operated as a single, cohesive, and virtual data
In the proposed method, the DAU must be installed on network.
the UAV along with the participating agents, which inter- 3) Radio and Electromagnetic Spectrum: The growing need
act with the UAV’s flight control system. Meanwhile, the for internet connectivity for various purposes and needs in
other participating agents in the network are ground-based everyday life has made it possible to propose new spectra for
computing systems. The task-handling layer is responsible conventional and future-oriented communication networks. It
for maintaining communication among the agents. The KD is, therefore, necessary to elaborate on the communications
box stores the information utilized by the agents and the spectrum used by UAVs for specific missions, whether for
task-handling layer. Figure 10 presents the structure of an military or civilian purposes. Radiofrequency (RF) is one of
autonomous distributed framework where communication can the parts of the electromagnetic spectrum often used in a
be made through various types of networks in the UAV. This specific frequency range, which is then called frequency bands.
demonstrates the interconnectivity of IoT communication net- Their range lies between 3KHz to 300GHz all current wireless
work devices with various contributing networks, ranging from connectivity networks benefit from the RF spectrum, such as
satellites to underwater/underground networks, facilitating data analog radio, navigation of aircraft, radio employed in marine,
transfer across networks. amateur radio, television broadcasting, mobile networks, and
satellite communications. The international organization that
TABLE V regulates the generation, transmission, and receiving of RF
K NOWLEDGE D ISCOVERY T ECHNIQUES FOR INTELLIGENT FDR S D ESIGN
[66], [68], [76] bands is referred to as the International Telecommunication
Union (ITU) [71]. The international body is responsible for
Knowledge Discovery Approaches Reinforcement Ensemble preventing interference among users and collaborates with
Learning (RL) Method and
Rule-Based each country’s national interest with respect to its laws. More-
Probabilistic and Statistical Meth- Model Based Association over, the organization headquartered in Geneva, Switzerland,
ods, Q-Learning Rule monitors and provides recommendations for the shared global
Classification based on Bayesian Deep Q- Random Forest
Techniques, Inductive Logic, Pat- Learning
usage of the radio spectrum and then promotes international
tern Discovery, Data Cleaning, cooperation while assigning the satellite orbits. The selection
Transfer Learning, Support Vector of RF frequency is crucial in designing the communication
Machine (SVM), Support Vector
Regression (SVR)
system (continuously sending or receiving the data). RF or
Genetic Algorithms Q-Network AdaBoost electromagnetic is a traditional source of communication
Deep, Artificial, Recurrent, Con- Deep Q- Decision Tree among UAVs, ground control stations (GCS), satellites, and
volutional, and Long Short-Term Network other contributors throughout the network. As shown in Figure
Neural Networks
Combination of two techniques, Flat Expert System 11, the overall methodology and functioning of the intelligent
for instance, Hybrid Approaches Hierarchical- FDR with real-time data transfer are depicted.
by integrating two or more above- RL, Meta RL, Laser beam communication as a data link can also be
mentioned methods HRL
used between aircraft, UAVs, and satellites as well. A white
paper from the University of Oxford [72] initiated the Hy-
Fig. 11. Proposed FDR Model for UAVs
perion Project, which researches laser communication among ping, and Airborne intercept
aerospace network participants. Because most utilization of • HF – Frequency range- 3-30 Mhz, Current Deployment
RF, the exponential growth of UAVs, and the number of in Radar Surveillance
satellites cause bandwidth allocation problems in communi- • V HF – Frequency range- 30-3002 Mhz, Application:
cating with any type of aircraft. On the other hand, laser Long Range Surveillance
communication is limited to a few kilometers (Km) hence • U HF – Frequency range- 300-1000 Mhz, Application:
distance constraints are present during the propagation of Long Range Surveillance
signals. Therefore, it is suitable for small UAVs to employ Based on the application area, we can use these frequency
laser communication, reducing the load of bandwidth ca- ranges for the data link and familiarize ourselves with the
pacity of RF frequency. Moreover, communication from and current status of the UAV. In the autonomous distributed
to UAVs, satellites, and other skies is done mainly through network, the data is transferred and saved through different
the employment of RF applications. The following frequency clouds, servers, and routers. For instance, the FDAU of the
bands are commonly employed for the goal of communication UAV can be associated with any link of a communication
[48], [69] network to acquire the data collected by FDRs and spread
• Ku - Frequency range- 12-18 Ghz, Area of Applications: it throughout the network. In this way, we have enough data
Satellite Altimetry and High-Resolution Mapping (HRM) for many purposes, i.e., training, designing, investigating ac-
• Ka -27-40Ghz, Area of Applications: Airport Surveillance cidents, and performing preventive maintenance. Additionally,
and Very HRM we shall identify the recording resolution of each parameter,
• K– Frequency range- 18-27 Ghz, Use: Water Vapor measurement range, recording interval, and accuracy of pa-
Absorption (Rare) rameters with its maximum sampling.
• S– Frequency range- 2-4 Ghz, Applications: Medium Use
in Surveillance, Terminal Traffic Control, and Weather
Purposes C. Data Sharing Phase in IoT Networks
• L– Frequency range- 1-2Ghz, Area of Applications: En- This section provides a data-sharing mechanism among the
route Traffic Control and Surveillance small UAVs throughout the IoT-based autonomous network.
• C– Frequency range- 4-8 Ghz, Area of Applications: The possible assumptions we have made through our under-
Tracking and Weather detection of Airborne standing are in two phases. The concept of a timestamp is
• X– Frequency range- 8-12 Ghz, Area of Applications: essential to be introduced for the given data to a data store
General Tracking, Marine radar, Missile Guidance, Map- or the collection of data in the KD and keeping records
of information online. This concept signifies the particular intelligent FDR with the help of the autonomous distributed
data storage at different times. Firstly, the data generated communication network. The roadmap and recommendations
from UAVs can be classified into two categories, i.e., UAVs’ proposed in this article will undoubtedly help any aircraft deal
mission data and UAV’s flight data. However, storing both with mishaps and provide sufficient data that can be used for
data is essential for mission and flight current characteristics. further safety improvements and other purposes.
In this study, our research is based on the FDRs of UAVs
regarding flight data. Therefore, we use the data link layer of ACKNOWLEDGMENTS
a communication system for opting flight data and then store The Federal Aviation Administration partly supported this
it at any node of the wireless communication network. research under Grant No. ASSURE A55.
1) Phase-I: Sending information from a UAV itself is a
challenge in observing the device’s computational resources.
The sensor’s data streams are processed to the FDAU; then it’s Legal Disclaimer
transmitted through a communication channel. The relevant The information provided herein may include content sup-
data can be saved to a particular owner of the UAVs or the plied by third parties. Although the data and information con-
GCS of that drone or to any network. tained herein have been produced or processed from sources
2) Phase-II: For the classification based on Figure 5 in believed to be reliable, the Federal Aviation Administration
micro, mini, very small, and small UAVs, we assume the makes no warranty, expressed or implied, regarding the accu-
flight data is generated and saved either on-board or the real- racy, adequacy, completeness, legality, reliability, or usefulness
time data policy is integrated into it. The sensors’ data is of any information, conclusions or recommendations provided
processed per FDR’s methods used in general aviation and herein. Distribution of the information contained herein does
then transferred to a network node. Each data frame enters not constitute an endorsement or warranty of the data or
a server with a timestamp that records every data. It also information provided herein by the Federal Aviation Admin-
can check the data in the server, and repeatedly data can be istration or the U.S. Department of Transportation. Neither
omitted from this process. Different UAVs can fly at different the Federal Aviation Administration nor the U.S. Department
times; the timestamp also remembers the data coming from a of Transportation shall be held liable for any improper or
particular drone at different intervals. incorrect use of the information contained herein and as-
The association between UAV’s flight control system and sumes no responsibility for anyone’s use of the information.
FDAU with IoT communication network are the remaining The Federal Aviation Administration and U.S. Department of
portions of the proposed intelligent FDR design for sending the Transportation shall not be liable for any claim for any loss,
real-time data of UAV parameters. A simulation platform can harm, or other damages arising from access to or use of data or
be generated with this design and then enlarged to a hardware information, including without limitation any direct, indirect,
design for testing and to check the feasibility of the methods, incidental, exemplary, special or consequential damages, even
which we shall consider in the following research phase. if advised of the possibility of such damages. The Federal
Aviation Administration shall not be liable to anyone for any
decision made or action taken, or not taken, in reliance on the
VII. C ONCLUSION information contained herein.
This paper studied FDRs for conventional aircraft systems
and UAV systems. We provided a brief introduction about R EFERENCES
FDRs illustrating the necessity to install them in any air vehi- [1] Public Broadcasting Service (PBS)
cle. Starting from previously used technologies and the initial https://www.pbs.org/newshour/nation/watch-live-boeing-ceo-tells-
phase of FDRs, we captured the whole picture of FDRs and congress-mistakes-were-made-on-737-max-jets (accessed: Dec.11,2022)
[2] France Ministry of Transport and Tourism, Official ”Report on the
briefly discussed data acquisition system theory, operational Airbus A320, registered F-GGED, crashed on 20 January 1992” Mont
verification checklist, and the calibration process. An overview Saint Odile in Alsace, France, 26 November 1993.
of the technical development of FDRs is given and how these [3] Safety Regulation Group ”Approval, Operational Serviceability and
Readout of Flight Data Recorder Systems and Cockpit Voice Recorders”,
technologies evolved. Our research focused on the regulation UK Civil Aviation Authority, 2011.
of the FDRs by aviation organizations, professional societies, [4] Dub, M., and Pařı́zek, J. (2018). ”Evolution of Flight Data Recorders.”
and governing bodies. These entities develop standards and Advances in Military Technology, vol. 13, no. 1, pp. 95–106, May 2018.
https://doi.org/10.3849/aimt.01226
guidance materials, which are subsequently translated into [5] Annex 6 to the Convention on International Civil Aviation, ”Operation of
laws that all airlines and aviation industries are obligated to Aircraft”, Part III — International Operations Helicopters, Ninth Edition,
follow. The next section starts with a detailed introduction to July 2018.
[6] ICAO, Annex 6 to the Convention on International Civil Aviation,
UAVs which covers the background and paramilitaries of a ”Operation of Aircraft”, Part I, International Commercial Air Transport
UAV, identifies the parameter and operational requirements, Aeroplanes, eighth edition, July 2001.
and then provides the state-of-the-art of current lightweight [7] Joint Aviation Requirements, ”JAR-OPS 1”, Commercial Air Trans-
portation (Aeroplanes), Global Engineering Documents, 15 Inverness
FDRs. To establish the significance of this study for future Way East, Englewood, Colorado 80112, USA on behalf of the Joint
generations of FDRs, we conducted a comprehensive review Aviation Authorities Committee, Sep.2005.
of UAV-related accidents reported by prominent U.S. gov- [8] Crotty, D. and Churchward, M. (2010) ARL Flight
Memory Recorder in Museums Victoria Collections
ernment agencies, including the NTSB, the FAA, and the https://collections.museumsvictoria.com.au/articles/3720 (accessed
NASA accident portal. We elaborated a road map to design Jan.23,2023)
[9] Flight Black Box, ”US 2959459”, [33] Kaleta, Ryszard, Janusz Niczyj, and Ryszard Winiarski. ”Trends of
https://en.patentoffice.ir/patent/107/Flight-Black-Box (accessed changes of flight data recorders in the field of their utilization.” AUTO-
Jan.23,2023) BUSY Technika, Eksploatacja, Systemy Transportowe 24.6, pp 303-310,
[10] EUROCAE Technical Work Programme, ”Edition 2022”, EUROCAE 2019.
Technical Advisory Committee, 10 November,2021 [34] European Commission H2020, Mobility for Growth MG-3.1-2016,
[11] Erik Gregersen, ”flight recorder”, Encyclopedia Britannica, 4 Aug. ”Addressing aviation safety challenges”, D3.2 Data-driven Research
2022, https://www.britannica.com/technology/flight-recorder (accessed Addressing Aviation Safety Intelligence, SafeClouds.eu, Project Report,
Dec. 20,2022) March, 2018
[12] Flight Data Recorders, The Background on the ‘Black Box’, The Air- [35] M Sudolsky. ARINC 573/717, 767 and 647a: The logical choice for
craft Electronics Association, Avionics news, Feb 2010. https://aea.net, maintenance recording and IVHM interface control or frame updates.
ANArchives TechSpeakFeb10.pdf (accessed Jan.23,2023) In Proc. of Prognostics and Health Management Society, 2009.
[13] Kowaleczko, Grzegorz, and Wieslaw Sobieraj. ”Reconstruction of an [36] Gavrilovski, Alek, Hernando Jimenez, Dimitri N. Mavris, Arjun H. Rao,
aircraft accident using an inverse technique”, 41st Aerospace Sciences Sanghyun Shin, Inseok Hwang, and Karen Marais. ”Challenges and
Meeting and Exhibit. 2003. opportunities in flight data mining: A review of the state of the art.”
[14] Morcom, A.R. (1970), “Flight Data Recording Systems: A brief survey AIAA Infotech@ Aerospace (2016): 0923.
of the past developments, current status and future trends in flight [37] Taha, Zahari, Y. R. Tang, and K. C. Yap. “Development of an onboard
recording for accident investigation and operational purposes”, Aircraft system for flight data collection of a small-scale UAV helicopter.”
Engineering and Aerospace Technology, Vol. 42 No. 7, pp. 12-16. Mechatronics 21.1 (2011): 132-144.
https://doi.org/10.1108/eb034655 [38] Qin, Guojie, Guoman Liu, and Hui Feng. “Design and implementation
[15] Helder Mendes. Study of mathematical algorithms to identify abnormal of a solid-state flight data recorder using multichannel technique.” 2013
patterns in aircraft flight data. Master’s thesis, Instituto Superior Técnico, Fourth International Conference on Intelligent Control and Information
Universidade Técnica de Lisboa, 2012. Processing (ICICIP). IEEE, 2013.
[16] International Civil Aviation Organization (ICAO). Annex 6 Operation [39] Kaygusuz, Y., T. İnal, S. Uyar, S. S. Ersoy, and Z. S. Mor. ”Aconceptual
of Aircraft, Part I, International Commercial Air Transport-Aeroplanes. design study of a flight data/voice recording system architecture with ac-
[17] Civil Aviation Authority (CAA). CAP 739: Flight Data Monitoring. The cident/incident statistics.” In Ankara international aerospace conference,
Stationary Office (TSO), CAA Aviation House, Gatwick Airport South, Ankara, Turkey. 2011.
West Sussex, RH6 0YR, 8 2013. ISBN: 0-86039-930-0. [40] Office of Research and Engineering Office of Aviation Safety, ”Flight
[18] European Aviation Safety Agency (EASA). Acceptable Means of Com- Data Recorder Handbook for Aviation Accident Investigations”, U.S.
pliance and Guidance Material to Part-ORO. European Aviation Safety National Transportation Safety Board, December 2002
Agency, Cologne, Germany, 2 2016. [41] European Operators Flight Data Monitoring Forum (EOFDM), ”Flight
[19] Kowaleczko, Grzegorz, and Wieslaw Sobieraj. ”Reconstruction of an data monitoring, analysis techniques and principles”, December 2021
aircraft accident using an inverse technique.” In 41st Aerospace Sciences [42] ATR Training Center, ”Flight Data Monitoring On ATR Aircraft”, 2016
Meeting and Exhibit, p. 581. 2003. https://www.easa.europa.eu ( (accessed Dec. 30, 2022))
[43] Jesse Garrison, ”Aeronautical and Aerospace Engineering
[20] Siegel, Greg. Technologies of accident: Forensic media, crash analysis,
Research Guide”, Iowa State University University Library,
and the redefinition of progress. The University of North Carolina at
https://instr.iastate.libguides.com/c.php?g=49290p=317896
Chapel Hill, 2006.
[44] International Standards and Recommended Practices, Annex 6 to the
[21] Norman, James. 2022. ”Open-Source Flight Data: Free,
Convention on International Civil Aviation, Operation of Aircraft, Part
but at What Cost?” Engineering Proceedings 28, no. 1: 9.
I International Commercial Air Transport — Airplanes, Ninth Edition
https://doi.org/10.3390/engproc2022028009
July 2010, International Civil Aviation Organization.
[22] Neil A. H. Campbell, The evolution of flight data analysis. Australian [45] International Standards and Recommended Practices, Annex 6 to the
New Zealand Society of Air Safety Investigators Conference, ANZSASI, Convention on International Civil Aviation, Operation of Aircraft, Part
Wellington, New Zealand, 2007. II — International General Aviation — Aeroplanes Tenth Edition,
[23] J.R. Sturgeon (1969), Technical Prospects of the Use of Digital Flight November 2018 International Civil Aviation Organization.
Recorders for Operational Research and Accident Prevention, RAE [46] S. M. Malaek and E. Alipour, “Intelligent Flight-Data-Recorders;
Technical Report 69201, 1969. a Step Toward a New Generation of Learning Aircraft,”
[24] Aeronautical Radio Inc. Enhanced Airborne Flight Recorder: ARINC 2022 8th International Conference on Control, Decision and
Characteristics 767-1. Aeronautical Radio Inc., Annapolis, MA, U.S., Information Technologies (CoDIT), pp. 1545-1549, 2022, doi:
2009. 10.1109/CoDIT55151.2022.9804136.
[25] Michael H. Thompson, A vision of future crash survivable recording [47] Evans, Andrew, ”The Hazards of Unmanned Air Vehicle Integration into
systems, Transportation Recording: 2000 and Beyond. International Unsegregated Airspace”, MS Thesis, The University of York, Sep. 2006
Symposium on Transportation Recorders, Arlington VA, United States [48] Sadraey, Mohammad H. Design of Unmanned Aerial Systems. John
1999-5-3 to 1999-5-5, pp 337-349, Jan 30 2008. Wiley and Sons, 2020.
[26] “Black Box” by Vanda Sendzimir; Invention and Technology, Vol. 12. [49] Gundlach, Jay. Civil and Commercial Unmanned Aircraft Systems.
no. 2, Fall 1996, https://www.inventionandtech.com/content/black-box-1 American Institute of Aeronautics and Astronautics, Inc., 2016.
(accessed: Dec.11,2022) [50] Proceedings of the First Symposium on Aviation Maintenance and
[27] Bureau d’Enquêtes et d’Analyses (BEA). Flight data recorder read-out: Management-Volume II, Springer Berlin, Heidelberg, Published: 10
Technical and regulatory aspects. Technical report, Bureau d’Enquêtes April 2014, https://doi.org/10.1007/978-3-642-54233-6
et d’Analyses pour la Sécurité de l’Aviation Civile, 5 2005. [51] Fotouhi, Azade, Haoran Qiang, Ming Ding, Mahbub Hassan, Lorenzo
[28] Aeronatical Radio Inc. Flight Data Acquisition and Recording System: Galati Giordano, Adrian Garcia-Rodriguez, and Jinhong Yuan. ”Survey
ARINC Characteristics 717-15. Aeronautical Radio Inc., Annapolis, on UAV cellular communications: Practical aspects, standardization ad-
MA, U.S., 2011. vancements, regulation, and security challenges.” IEEE Communications
[29] Lyko, Klaus, Marcus Nitzschke, and Axel-Cyrille Ngonga Ngomo. surveys tutorials 21, no. pp 3417-3442, 2019
”Big data acquisition.” New Horizons for a Data-Driven Economy: A [52] C. Stöcker, R. Bennett, F. Nex, M. Gerke, and J. Zevenbergen, “Review
Roadmap for Usage and Exploitation of Big Data in Europe, 39-61, of the current state of UAV regulations,” Remote Sens., vol. 9, no. 5, p.
2016 459, 2017
[30] Fuchs, Christian M., and Advisors Stefan Schneele. ”The evolution of [53] Fong, Terry. Autonomous systems, Nasa capability overview. No. ARC-
avionics networks from ARINC 429 to AFDX.” Innovative Internet E-DAA-TN62415., Oct. 2018.
Technologies and Mobile Communications (IITM), and Aerospace Net- [54] Young, Larry, Jeffrey Yetter, and Mark Guynn. ”System analysis applied
works (AN) 65, 2012. to autonomy: Application to high-altitude long-endurance remotely
[31] Mukherjee, Debasis, et al. “Implementation of ARINC 429 16 channel operated aircraft.” In Infotech@ aerospace, p. 7103. 2005.
transmitter controller on FPGA.” Information Technology and Mo- [55] Heinrich, Stephan, and Lucid Motors. ”Flash memory in the emerging
bile Communication: International Conference, AIM 2011, Nagpur, age of autonomy.” Flash Memory Summit (2017): pp. 1-10, 2017
Maharashtra, India, April 21-22, 2011. Proceedings. Springer Berlin [56] Moussa Aboubakar, Mounir Kellil, Pierre Roux, A review of IoT
Heidelberg, 2011. network management: Current status and perspectives, Journal of King
[32] Pascal Andrei, Uwe Bartels, and Volkmar Neeb. Flight data recovery: Saud University - Computer and Information Sciences, Volume 34, Issue
Time for evolutions. Airbus FAST Magazine, 48, 2011. 7, 2022, Pages 4163-4176, doi.org/10.1016/j.jksuci.2021.03.006.
[57] Jianpeng Zhang, Real-time detection of energy consumption of Waleed Raza received a B.E. degree in Electronic
IoT network nodes based on artificial intelligence, Computer Com- Engineering from the Department of Electronic En-
munications, Volume 153,2020, Pages 188-195, ISSN 0140-3664, gineering, Dawood University of Engineering and
doi.org/10.1016/j.comcom.2020.02.015. Technology Karachi, Pakistan in 2017. He received
[58] L3-Harris- Extra Light Data Recorder, an M.S. degree in underwater acoustic communi-
https://www.aviationtoday.com/2022/07/11/l3harris-lilium-extra-light- cation engineering from the College of underwater
data-recorder/ (accessed: 01/23/2023) acoustic engineering, Harbin Engineering University,
[59] FDR-01 Flight Data Recorder, Harbin China where he researched OFDM commu-
https://www.uavnavigation.com/products/peripherals/flight-data- nication for underwater technology. He is the edito-
recorder (accessed: 01/23/2023) rial board member for Engineering, Technology, and
[60] Data Recorder Small Enough for Drones, Applied Science Research (ETASR) (2021-present),
https://www.electronicspecifier.com/industries/robotics/data-recorder- He is an active reviewer of a few journals including IEEE Sensor Journal,
small-enough-for-onboard-drones (accessed: 02/10/2023) IEEE Access, and International Journal of Electronics and Communications
[61] Hensoldt Flight Recorder, https://verticalmag.com/press- (2018-2022), he has recently joined the IEEE as a student member. Currently,
releases/hensoldt-launches-new-flight-recorder/ (accessed: 01/23/2023) he is pursuing a Ph.D. at Embry Riddle Aeronautical University in Electrical
[62] NTSB Data, Aviation Investigation Fields, https://data.ntsb.gov/carol- and Computer Engineering. His research area of interest includes underwater
main-public/basic-search acoustic OFDM communication, underwater acoustic target detection, ma-
[63] FAA Safety Information Analysis and Sharing chine learning for communication engineering, and autonomous unmanned
https://www.asias.faa.gov/apex/f?p=100:446:::NO:446 systems such as UAVs.
[64] NASA’s Aviation Safety Reporting System
https://asrs.arc.nasa.gov/uassafety.html.
[65] Brachman, R.J., and Anand, T. The Process Of Knowledge Discovery In
Databases: A Human-Centered Approach. In Advances In Knowledge
Discovery and Data Mining, eds. U.M. Fayyad, G. Piatesky-Shapiro, P.
Justus Renkhoff earned bachelor’s and master’s
Smyth, and R. Uthurusamy, AAAI Press/MIT Press, Menlo Park, CA.
degrees in media and computer science from Trier
PP 37-57, 1996
University of Applied Sciences, Germany in 2019
[66] Aborizka, Mohamed and Krishna M. Kavi. “An Intelligent Flight Data
and 2021, respectively. He is currently pursuing a
Recorder And Real-Time Monitoring System.”, 2000.
Ph.D. degree in Electrical Engineering and Com-
[67] Davies, Winton HE, and Pete Edwards. “Agent-based knowledge dis-
puter Science at Embry-Riddle Aeronautical Uni-
covery.” Working Notes of the AAAI Spring Symposium on Informa-
versity. Previously, he worked in research at the
tion Gathering from Heterogeneous, Distributed Environments. Stanford
Institute of Software Systems at Trier University,
University, Stanford, CA., 1995.
held a teaching position there at the university, and
[68] Wang, Lizhe Yan, Jining Mu, Lin Huang Liang, Knowledge discovery
worked as an adjunct instructor at St. Bonaventure
from remote sensing images, Interdisciplinary Reviews Data Mining,
University, New York. He focuses on deep learning
and Knowledge, 2020
cyber-physical systems and software engineering.
[69] J. Li, Y. Zhou and L. Lamont, “Communication architectures and
protocols for networking unmanned aerial vehicles,” 2013 IEEE Globe-
com Workshops (GC Wkshps), Atlanta, GA, USA, pp. 1415-1420, doi
10.1109/GLOCOMW.2013.6825193., 2013
[70] A. Sharma, P. Vanjani, N., C. M. Wijerathna B., Dushantha Nalin
K. Jayakody, H. Wang, P. M., Communication and networking Olayemi Ogirimah earned his bachelor’s degree
technologies for UAVs: A survey, Journal of Network and Com- in Computer Engineering from the Department of
puter Applications, Volume 168, 2020, 102739, ISSN 1084-8045, Computer Science and Engineering, Ladoke Akin-
doi.org/10.1016/j.jnca.2020.102739. tola University of Technology, Nigeria. He is cur-
[71] ITU, Press release, https://www.itu.int/en/mediacentre/Pages/pr31-2020- rently pursuing his master’s degree in Unmanned
open-research-group-autonomous-networks.aspx (accessed: 02/10/2023) and Autonomous Systems Engineering at Embry-
[72] UK, Hyperion Projectm laser-based aircraft tracking system, Riddle Aeronautical University, Daytona Beach
https://optics.org/news/7/3/15 (accessed: 01/26/2023) Campus, Florida. He is currently a student member
[73] R. Chen and J. Jin, ”Detection of Influential Features of Hard Landing of the Association for Uncrewed Vehicle Systems
Based on QAR Data,” 2021 IEEE 3rd International Conference on Civil International (AUVSI). He has lots of experience
Aviation Safety and Information Technology (ICCASIT), Changsha, researching, building, and flying UAVs. He is also
China, 2021, pp. 526-529, doi: 10.1109/ICCASIT53235.2021.9633658. an FAA-certified Part 107 remote pilot. His current and future research focus
[74] X. Li, J. Shang, L. Zheng, Q. Wang, H. Sun and L. Qi, ”CurveClus- remains in the field of Unmanned Aircraft Systems.
ter+: Curve Clustering for Hard Landing Pattern Recognition and Risk
Evaluation Based on Flight Data,” in IEEE Transactions on Intelligent
Transportation Systems, vol. 23, no. 8, pp. 12811-12821, Aug. 2022,
doi: 10.1109/TITS.2021.3117846.
[75] Li, Lishuai, et al. ”Anomaly detection via a Gaussian Mixture Model for
Richard Stansbury is an associate professor
flight operation and safety monitoring.” Transportation Research Part C:
of Computer Science and Computer Engineering
Emerging Technologies 64 (2016): 45-57.
at Embry-Riddle Aeronautical University, Daytona
[76] Janakiraman, Vijay Manikandan. ”Explaining aviation safety incidents
Beach Campus. He received his BS and MS degrees
using deep temporal multiple instance learning.” In Proceedings of the
in Computer Engineering (2002 and 2004 respec-
24th ACM SIGKDD International Conference on Knowledge Discovery
tively) and Ph.D. in Computer Science (2007) from
Data Mining, pp. 406-415. 2018.
the University of Kansas. As a graduate research
assistant, he developed autonomous ground vehicles
for operation in Greenland and Antarctica. His Ph.D.
dissertation was titled “Constraint-based Task Se-
lection and Configuration for Autonomous Mobile
AUTHOR ’ S B IOGRAPHY Robots.” He serves as a research leader at ERAU as the University leads for the
ASSURE FAA Center of Excellence for Unmanned Aircraft Systems (UAS).
In this role, he serves as an administrator over ERAU’s business within the
Center and as a frequent research investigator for FAA-sponsored grants. His
unmanned systems research includes topics of airspace integration including
system requirements for urban air mobility, aviation data analysis, etc. His
other areas of research involve applied artificial intelligence/machine learning
and mobile robotics.
Houbing Song (M’12–SM’14-F’23) received the

Ph.D. degree in electrical engineering from the Uni-
versity of Virginia, Charlottesville, VA, in August
2012. He is currently a Tenured Associate Professor,
the Director of the NSF Center for Aviation Big
Data Analytics (Planning), the Associate Director for
Leadership of the DOT Transportation Cybersecurity
Center for Advanced Research and Education (Tier
1 Center), and the Director of the Security and Op-
timization for Networked Globe Laboratory (SONG
Lab, www.SONGLab.us), University of Maryland,
Baltimore County (UMBC), Baltimore, MD. Prior to joining UMBC, he
was a Tenured Associate Professor of Electrical Engineering and Computer
Science at Embry-Riddle Aeronautical University, Daytona Beach, FL. He
serves as an Associate Editor for IEEE Transactions on Artificial Intelligence
(TAI) (2023-present), IEEE Internet of Things Journal (2020-present), IEEE
Transactions on Intelligent Transportation Systems (2021-present), and IEEE
Journal on Miniaturization for Air and Space Systems (J-MASS) (2020-
present). He was an Associate Technical Editor for IEEE Communications
Magazine (2017-2020). He is the editor of eight books, the author of more
than 100 articles, and the inventor of 2 patents. His research interests include
cyber-physical systems/internet of things, cybersecurity and privacy, and
AI/machine learning/big data analytics. His research has been sponsored by
federal agencies (including the National Science Foundation, US Department
of Transportation, and Federal Aviation Administration, among others) and
industry. His research has been featured by popular news media outlets, includ-
ing IEEE GlobalSpec’s Engineering360, Association for Uncrewed Vehicle
Systems International (AUVSI), Security Magazine, CXOTech Magazine, Fox
News, U.S. News & World Report, The Washington Times, and New Atlas.
Dr. Song is an IEEE Fellow, an ACM Distinguished Member, and an ACM
Distinguished Speaker. Dr. Song is a Highly Cited Researcher identified by
Clarivate™ (2021, 2022) and a Top 1000 Computer Scientist identified by
Research.com. He received the Research.com Rising Star of Science Award
in 2022 (World Ranking: 82; US Ranking: 16). Dr. Song was a recipient of
10+ Best Paper Awards from major international conferences, including IEEE
CPSCom-2019, IEEE ICII 2019, IEEE/AIAA ICNS 2019, IEEE CBDCom
2020, WASA 2020, AIAA/ IEEE DASC 2021, IEEE GLOBECOM 2021 and
IEEE INFOCOM 2022.

Flight Data Recorders Past Present and Future

Uploaded by

Copyright:

Available Formats

Flight Data Recorders Past Present and Future

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flight Data Recorders Past Present and Future

Uploaded by

Copyright:

Available Formats

This article has been accepted for publication in IEEE Aerospace and Electronic Systems Magazine.

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 1

Flight Data Recorders: Past, Present, and Future

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 2

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 3

Fig. 1. FDR’s Technology Devices Evolution [32] [33]

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 4

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 5

divided into four different one-second frames. In a one-second TABLE I

B. Operational Verification Checklist for Recording

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 6

Fig. 2. Principle of Recording [42]

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 7

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 8

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 9

Fig. 3. Classification of UAVs[48]

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 10

TABLE III TABLE IV

Category Applications Aero- Motor/ Control Status and Configur- Peripheral

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 11

Fig. 4. Conventionally UAVs Protocol by Evans AR[47]

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 12

Fig. 5. Need for FDR based on Classification of UAVs

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 13

and lightweight, it has become possible to fit it into small

F. Survey of Reported Accidents

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 14

Fig. 9. FDRs’ Improvement over Time

introduced to reduce the number of maintenance tasks where

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 15

Fig. 10. Distributed Autonomous IoT network for UAVs

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 16

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 17

Fig. 11. Proposed FDR Model for UAVs

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 18

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 19

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 20

IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE, VOL. X, NO. X, 2023 21

Houbing Song (M’12–SM’14-F’23) received the

You might also like