IEEE - Aerospace and Electronic Systems - July 2023 - 53

Laaksosaari et al.
separate configurable comparison variables and other
P4 samples with different configurable comparison
variables. However, our testing achieved relatively
low error rates ranging from 0.02%-0.17%. Therefore,
we leave such further variables fine-tuning as future work. In
addition, another interpretation problem is present when
Mode S preamble is identical to part ofthe Mode A/C all-call
message. In turn, if the Mode A/C check fails on Mode A/C
all-call message, the program will loop until it reaches the
start of P3 and marks it as a Mode S message. However,
Mode S check uses stricter variables by default than the P4
check, which can mitigate the amount of false detection in
this scenario. Furthermore, Mode C all-call has significantly
more false detection (7.05 %) than Mode A all-call (1.87 %)
because the failure of detecting Mode C message is much
more likely due to longer length, which in turn leads to more
comparisons. Comparing values from the start ofthe pulse P6
data block could fix this misinterpretation since there is a continuous
1.25 ms pulse [16], [17], and we leave this marginal
improvement for future work.
SAMPLING-RATELIMITATIONSOFRTL-SDR
RTL-SDR dongles have a theoretical maximum sample
rate of 3.2 MS/s; however, in practice, most dongles start
dropping samples after passing 2.4 MS/s [18], [19]. Testing
was conducted successfully with RTL2832 U, while
the FC0012 equipped dongle repeatedly failed with
dump1030 at the required sample rate of 2.5 MS/s. For
more detailed introduction to RTL-SDR dongles, their
types as well as their practical applications and constraints,
we reference the interested readers to relevant
works [20], [21], [22].
The Mode S interrogation requires a minimum sample
rate of 2 MS/s due to preamble pulse width of 0.5 ms and
P6 data block of 56 bits or 112 bits being encoded in pulse
position modulation throughout 56 or 112 ms, respectively.
Pulse position modulation uses one pulse and one
nonpulse to represent every bit; therefore, a 112-bit message
over 112 ms leads to a pulse/nonpulse width of
0.5 ms. The P6 data block of Mode S uplink message differs
in two ways from the P6 data block of Mode S downlink
message: it uses differential phase shift keying
(DPSK) and its length in time is 30.25 or 16.125 ms.
DPSK requires one pulse per bit; thus, 112 bits over
30.25 ms leads to a pulse width of 0.25 ms [16], [17]. The
pulse width of 0.25 ms would require a minimum sample
rate of 4 MS/s, which is beyond the theoretical maximum
sample rate of RTL-SDR devices. Hence, decoding the P6
data block of Mode S uplink messages (similar to what
offline decoders from OpenSky and pyModeS offer-see
notes in the " Introduction " section) requires more capable
SDR receivers (e.g., HackRF or BladeRF) and a nontrivial
improvement of the dump1030 software; therefore, we
leave this as future work.
JULY 2023
TOWARD OPEN-SOURCE TCAS IMPLEMENTATIONS
The 1030-MHz interrogations are an essential part of the
TCAS [26], [27]. Therefore, to move forward toward
open-source TCAS implementations, whether passive
monitoring-only receivers or active transceiver systems, it
is essential to have open-source implementations able to
receive the 1030-MHz interrogations at least. In Figure 7,
USE-CASE SCENARIOS
IMPROVED BIG-DATA ANALYTICS OF ADS-B
COMMUNICATIONS
Projects, such as OpenSky [8], FlightRadar24 [23], collect
ADS-B data globally, and then store and process that data
for multiple purposes, such as research reports and analytics
[24], [25]. To the best ofour knowledge, unfortunately,
at present, all such (crowd-source) systems collect only
downlink messages (e.g., 1090-MHz responses) [25]. This
means the collected data may lack complete or accurate
context, mainly when it aims for TCAS and SSR interrogations
analysis [24], [25]. Therefore, tools, such as
dump1030, can bring essential improvement and granularity
to the ADS-B data collection ofpaired and contextualized
1030-MHz (uplink) and 1090-MHz (downlink)
transmissions. Moreover, dump1030 can also help build
more advanced and mature ML models considering more
communication, such as 1030-MHz uplink messages.
Figure 7.
Conceptual design of different vehicles having onboard TCAS
implementation based on open-source projects, such as
dump1090 and dump1030. The design assumes the depicted
FOSS/TCAS box is present in all airborne vehicles.
IEEE A&E SYSTEMS MAGAZINE
53
IEEE - Aerospace and Electronic Systems - July 2023

Table of Contents for the Digital Edition of IEEE - Aerospace and Electronic Systems - July 2023
