I wonder if such a bunch of trajectories would draw the attention of the Indonesian airspace surveillance chain (the location of the unidentified A/C matters versus a RoI, but also its radial speed ?), the best fit trajectory is slightly (a few NM) off Banda Aceh:
https://drive.google.com/file/d/0B3s...it?usp=sharing
The 18:40 C-channel BFO 88 Hz value is not an outlier for the predicted BFO enveloppe of my MC simulation and it does not degrade the BFO error optimization (as well as the 3rd handshake around 18:28), the following result (30 000 runs) is for a south turn occuring 6 min after 18:28, early trajectory not constrained by the ADS-B and radar tracks:
https://drive.google.com/file/d/0B3s...it?usp=sharing
The inclusion of this new BFO value does not alter the most probable area (between 28 and 32°S) along the last ping ring, the best fit trajectories last south leg remains at a relatively low true ground speed (around 410 kts) ending round 30°S ...and curvy (not a a great circle or a rhumb line):
https://drive.google.com/file/d/0B3s...it?usp=sharing
It does not modify the mean BFO error for each handshakes (as a function of time and true ground speed):
https://drive.google.com/file/d/0B3s...it?usp=sharing
The inclusion of a new BFO data point or of a delay (from 18:28) for the south turn does not seem to impact the simulation end result, what really does seem to have an impact is whether I allow the 1st part (till 18:28) of the trajectories to fluctuate & drift (resulting in a wider variety of trajectories and of injection points for the last south leg) or if I constrain them to the ADS-B & radar tracks (which seems to shift the most probable area more than 1° toward the south).