At 13:15 29 May 2015, Mike Schumann wrote:
On Thursday, May 28, 2015 at 8:59:12 PM UTC-5, Andy Blackburn
wrote:
I'll buy the beer.=20
=20
Prediction, projection, position, protection. I'm not very clear on the
f=
iner points of my mother language apparently. ;-)
=20
Here's how I look at it. FLARM calculates a projected flight path with
a
=
probabilistic "error radius" determined based on flight parameters to
estim=
ate a likely future position (assuming no change in control inputs) and
the=
n maps these paths to estimate likely conflicts and warns if it finds
one.
=
I consider that warning a prediction - if nothing changes the two
aircraft
=
will likely collide. Sure, it's an estimate but I still consider that
more
=
of a prediction than just putting airplane-shaped dots on a display and
tel=
ling the pilot "you figure it out". I'd rather have a microprocessor and
an=
algorithm than burying my head in a traffic display. I'd call that
warning=
a prediction but maybe I'm being sloppy with the definitions.
=20
The point is - FLARM will give you a warning only for pretty real
threats=
.. A traffic advisory system can only annoy you with constant
warnings of
ai=
rcraft in the vicinity whether they are a threat or not or leave it to
you
=
to find threats.by staring at a display. You need to project a flight
path
=
with some precision to strike a balance between too may false
positives
and=
leaving too many possible threats suppressed until too late. Given
how we
=
fly I think FLARM does quite well. It's also why ADS-B will struggle to
ope=
rate as a collision warning system for gliders - even if someone tries
to
p=
laster a collision projection algorithm on top of some Garmin ADS-B
unit.
P=
owerFLARM throws out the ADS-B information if it detects a target
with
both=
FLARM and ADS-B Out.
=20
9B

The discussion in this thread raises a significant question as to how
much
=
"predictive" information is actually transmitted as part of the FLARM
proto=
col. I suspect that the differences in the information transmitted by
FLAR=
M and ADS-B are relatively minor.
I don't question the sophistication of FLARM's algorithms in filtering
out
=
collision threats that are real based on the trajectories of the aircraft
i=
n a soaring environment. Much of this processing must take place in
the
FL=
ARM's receiver, as the trajectories of both the receiving aircraft and
the
=
threat aircraft must be known to determine if there is a collision
threat.

My problem is with the assertion that a similar system can not be
implement=
ed using ADS-B. This is patently false. Numerous low cost iPhone
and/or
A=
ndroid apps already provide collision threat warnings today using low
cost
=
ADS-B receivers. These apps not only report collision warnings for
ADS-B
e=
quipped aircraft, but also for all transponder equipped aircraft that are
i=
dentified by TIS-B received from an ADS-B ground stations (Note:
You need
=
to be ADS-B OUT equipped to reliably receive TIS-B traffic).

It is unlikely that these existing apps are optimized like FLARM is to
reco=
gnize the unique flight profiles that are common with glider traffic.
We
w=
ill have to wait and see whether or not these companies will find it
worth
=
their while to invest in more sophisticated algorithms to specifically
targ=
et the glider market. However, there are glider specific apps (iGlide,
etc=
..) that are specifically aimed at the soaring market, which would have
a
st=
rong marketing advantage in including this kind of advanced ADS-B
based
cap=
ability.

In Europe, given the large installed FLARM base, this may be an
academic
di=
scussion. In the US, FLARM will never be widely deployed outside of
the
co=
mpetitive soaring community. Here, an economical ADS-B based
solution
that=
provided reliable warnings, not only for other gliders, but also all
other=
GA aircraft, would be a big hit.

The differences in the information transmitted by Flarm and ADS-B are
not minor. There is plenty of information included, or linked to, earlier

in this thread to explain what Flarm transmits and why.

John Galloway