My response to the Inmarsat "garbage" question

 This statement, sent to me by Mike Exner, a satellite industry expert,  via Twitter, asks those who believe that the 00:19:27 BTO value is garbage to provide evidence that it is bad.  Here is my response to this question and the reasoning behind my belief.



First off lets start with the data in question (see table below). Notice that the first Log-in request/Log-on acknowledge at (15:59:55, 16:00:13 ) seemed to have been measured as part of a normal and complete handshake. The value (14820us) for the 16:00:13 Log-on acknowledge seems to have been measured correctly without any anomalous effects and seems to match the derived value (14780us)  for the the Log-in request at 15:59:55 within a reasonable margin of error. So I have no issues with this first handshake measured values.


Time (UTC)

Message Channel Offset (us) Measured BTO value (us) Corrected BTO value (us)
7/03/2014 15:59:55 Log-on Request R600 -4600 19380 14780*
7/03/2014 16:00:13 Log-on/Log-off Acknowledge R1200 0 14820 14820
7/03/2014 18:25:27 Log-on Request R600 -4600 17120 12520*
7/03/2014 18:25:34 Log-on/Log-off Acknowledge R1200 0 51700 unknown
8/03/2014 00:19:29 Log-on Request R600 -4600 23000 18400*
8/03/2014 00:19:37 Log-on/Log-off Acknowledge R1200 0 49660 unknown

* corrected values

However the partial handshakes at 18:25 and 00:19 seem to tell a different story. Clearly there were some anomalous effect here that affected the Log-on/Log-off Acknowledge measured values at 18:25:34 (51700us)  and at 00:19:37 (49660us). Obviously these values were erroneous and had to be ignored.

My question is simply this,  what was this anomalous effect that affected these signals and could this anomaly have affected their respective  Log-on Request BTO values as well since these messages were part of the same logon power up/power down cycle and these signals were happening only seconds apart from each other within the same terminal unit.

Inmarsat has stated (see NOTE 1) that they believe these anomalous results at 18:25:34 and 00:19:37 were the result of Channel/Frequency switching interference within the SATCOM terminal during logon and should be ignored.  If this is the case then it is also theoretically possible that these Log-in request messages at 18:25:27 and 00:19:29:may have have also been affected by this internal switching. Even though the derived results they give us might seem reasonable we cannot say with a 100% certainty that they weren't affected  by this anomaly because the problem is we have no valid measured data from their respective  Log-on/Log-off Acknowledge messages to compare to like we have in the first handshake at 15:59

What gives me doubt about the 18:25:27 BTO value is that this corrected  value (12520 us) is less than the next succesive BTO value at 18:27:03 which was correctly measured at 12560us. The problem with this is that, at this time,  the plane was supposedly moving towards the satellite so the corrected value should have been higher than 12560us because it occured a couple of minutes before this time event (18:27:03) and was supposedly further away from the satellite at this time.   The value I estimate should have been around 12660us (17260us measured with a margin of error +/-40us ) based on the pattern of the 5 successive descending BTO values. Notice that the BTO value at 18:27:04  was measured correctly at 12520us the same value as the corrected value (12520) at 18:25:27.   Impossible for this to occur unless the plane was circling back at this time. (Please refer to the table below)


Time (us) Message Channel Offset (us) Measured BTO value (us) Corrected BTO value (us)
7/03/201418:25:27.421 Log-on Request R600 -4600 17120 12520*
7/03/201418:27:03.905 Eleven Octet User Data R1200 0 12560 12560
7/03/201418:27:04.405 Four Octet User Data R1200 0 12520 12520
7/03/201418:27:08.404 Acknowledge User Data R1200 0 12520 12520
7/03/201418:28:05.904 Access Request R1200 0 12500 12500
7/03/2014 18:28:14.904 Acknowledge User Data R1200 0 12480


Which BTO value does not seem to belong with the others?

Obviously this corrected value (12520us) may have been measured too low by a 140us or more perhaps. This is outside the acceptable margin of error of +/- 40us which is considered as normal jitter.

We can either choose to believe that the measured value( 17120us) is correct and the plane circled back, or magically zig-zagged, between 18:25:27 and 18:27:04, which makes no sense at all OR we can believe that the plane was flying in controlled flight,on autopilot, to waypoints, and we may have a another case of BAD MEASURED DATA (17120us).This is why I believe this signal (17120us) was also affected by channel/frequecy switching interference as well during logon.

If this derived value (12520us) is in error then this derived value (18400us) on the 00:19:27  time event is mostly likely erroneous as well. Both these measured values (17120us, 23000us) may just be erroneous measured data which may have been affected by internal switching interference inside the terminal during logon just like their respective Log-on Acknowledge signals were believed to have been affected by this anomaly (51700us, 49660us) on these specific partial handshake events only.

There is reasonable doubt here to question the validity of these 2 BTO values (18:25:27: 17120us, 00:19:29: 23000us) and we should make every effort we can to validate them since this information supposedly tells us the last known position of the plane.  If we cannot validate them with 100% certainty then perhaps it is better that we just assume that their correct values are unknown like the values for these time events (18:25:34, 00:19:37). This will inevitably make our math problem infinitely more complex and the search effort infinitely more impossible but if this is what we must do to find MH370 then this is what we must do.

Perhaps it would be a good idea to do testing in a lab on similar SATCOM units to try and replicate these anomalous results and try to determine if this switching effect during logon would have had any effects on the Log-in requests measured values. In the interest of leaving no stone unturned we should definitely test this hypothesis. These tests may provide evidence that this hypothesis is correct.

Since we don't know for sure what caused these power up/power down cycles at 18:25 and 00:19 UTC, was it human intervention or was something else like fuel exhaustion, at this point we simply don't know this answer so all possibilities must be considered here.

Based on the raw data only we cannot say with a 100% certainty that these values (18:25:27: 17120us, 00:19:29: 23000us) were or were not affected by channel and frequency switching during logon but there seems to be evidence in the pattern of BTO data that suggests this may be the case and we need to be absolutely certain of this if we are to have any hope of finding this plane.

To those that persistantly claim that the BTO value at 00:19:29 is correct, need to prove why it is correct, and why we should believe it was NOT affected by channel/frequency switching interference like these other 3 erroneous BTO values seem to have been.

Time (UTC) Message Channel Offset (us) Measured BTO value (us) Corrected BTO value (us) Measured BFO value (Hz)
7/03/2014 18:25:27 Log-on Request R600 -4600 17120 12520* 142
7/03/2014 18:25:34 Log-on/Log-off Acknowledge R1200 0 51700 unknown 273
8/03/2014 00:19:29 Log-on Request R600 -4600 23000 18400* 182
8/03/2014 00:19:37 Log-on/Log-off Acknowledge R1200 0 49660 unknown -2

I see no reason to have any faith in any of these measured BTO/BFO values for these time events. They should all be considered as "unreliable" delay mesasurements or as I like to say "garbage".



Ken S

twitter: @kstaubin


NOTE 1:  Inmarsat statement concerning the BTO values. See page 7 section 3.3

NOTE 2: Factual report statement concerning these questionable BFO values

NOTE 3: After submitting my information to the ATSB in January 2015, about a month later, the ATSB announced that they were widening the search area saying it was due to uncertaintity in the BTO data. I have no idea if this was a result of the information that I sent them but they did seem to cocncur with this analysis.



ATSB Publication:  Updated MH370 signalling unit log

ATSB Publication: MH370 Burst Timing Offset (BTO) Characteristics



View in block: 
Do You think some device may have caused interference?

Do You think some external device may have caused interference with the SATCOM terminal during these logon events at 18:25 and 00:19 UTC? This is why their Logon Acknowledge measured values were so erroreous, perhaps? 

Food for thought!

Sy Gunson
Misplaced BTO rings

The compelling evidence why BTO rings are in the wrong place is that two objects the length of Boeing 777 wings and a debris field of 300 floating objects sighted by satellites from five nations from 16-25 March. China reverse calculated their drift back to 45.30 South, 85.30 East on 8th March 2014. If translated to a radii from INMARSAT this impact point would be on a BTO ring 300nm further away from the satellite.

(Correction I first estimated the impact point at 44.12 South, 85.00 East and that was 220nm further from the 7th Arc - the Chinese estimate of impact at 45.30 S, 85.30 E is 300nm further from INMARSAT than 7th Arc -apologies)


More reasons to believe the sat data is wrong

Thanks for that info, who knows if this is debris from Mh370 then signal interference could explain the descrepency here and why the impact point was so far off from the 7th arc