TECHNICAL NOTE Rev. 1.0 – 01.10.2015MH370: On the possible interpretation of the abnormal BFO values of 273 Hz and -2 Hz

Oleksandr N.

Terms and AbbreviationsADIRU Air Data Inertial Reference UnitAES Aircraft Earth StationAP AutopilotATSB Australian Transport Safety BureauBFO Burst frequency offsetBTO Burst time offsetCTS Constant Thrust SettingsGES Ground Earth StationGPS Global Positioning SystemIDKUT Navigational waypointIG Independent GroupKLIA Kuala Lumpur International AirportNILAM Navigational waypointROC Rate of ClimbSAARU Secondary Attitude and Air Data Reference UnitUTC Universal Time Coordinated

1. IntroductionThere are two obviously abnormal BFO values of 273 Hz (18:25:34.461 UTC) and -2 Hz

(00:19:37.443 UTC) recorded by Inmarsat’s [2].

The first of them is inconsistent with the other BFO records in the same cluster of BFOs 18:25 – 18:27, and it is also inconsistent with the known heading and speed of the aircraft by 18:22.

The second abnormal BFO value of -2 Hz considerably differs from the BFO value of 182 Hz just 8 seconds earlier. Should this value be correct, it would imply an extreme descent rate (~15,000 ft /min).

While attempts took place to explain the BFO of 273 Hz as a result of some maneuver, such as a lateral offset, the second anomalous value of -2 Hz is widely believed to be erroneous.

This technical note provides an alternative view, suggesting that both the anomalous BTO values are valid, but they are the results of the inability of AES to apply Doppler compensation due to missing position/velocity data.

2. Analysis of the Inmarsat records An extraction of the series of BTO and BFO records (Appendix A, Table 1) from Inmarsat logs

[2] corresponding to the two events in question reveals that:

1. Both the records of the log-on events 18:25 and 00:19 are comprised of the identical sequences of communication transactions.

2. Both the second recorded BFO values in these sequences are abnormal: 273 Hz and -2 Hz respectively.

3. Both the abnormal BFO values are paired with the abnormal BTO values, which are of the same magnitude: 51,700 μs and 49,660 μs respectively.

4. Both the first BTO values in the sequences include an additional delay of 4600 μs (12,520 μs and 18,400 μs) – the value, observed on the startup at KLIA. This delay is also consistent with other BTOs in the cluster 18:25 – 18:27.

5. Latencies (the time difference between the next and previous transactions) are similar in both the instances, except “Oxll- Log-on Confirm”.

6. Although not relevant to this study, it is worth of noting that the 18:25:27 BFO of 142 Hz is the same as it is estimated to be on the moment of the communication loss (17:22).

This analysis suggests the abnormality is caused by the same reason in both the instances. Notably, the sequence of the records during normal logon at KLIA was different, or it was trimmed/modified by Inmarsat.

3. Suggested interpretation of the abnormal BFO valuesAccording to [1] BFO can be expressed as the sum of the following terms:

BFO=ΔFup+ΔFdown+δ f comp+δ f sat+δ f AFC+δ f bias,11\* MERGEFORMAT ()

where:

ΔFup is the Doppler shift of the signal passing from the aircraft to the satellite;ΔFdown is the Doppler shift of the signal passing from the satellite to the GES;δ f comp is the frequency compensation applied by the AES;δ f sat is the variation in satellite translation frequency;δ f AFC is the frequency compensation applied by the GES receive chain;δ f bias is a fixed offset due to errors in the aircraft and satellite oscillators.

The terms ΔFup and ΔFdown represent physical phenomenon, and thus they cannot be erroneous. The sum of the terms δ f sat+δ f AFC is known, and it is provided by Inmarsat [1]. Given that this sum was correct in all other BFO samples, it is very unlikely that the terms of this sum could be a reason for the abnormality. The term δ f bias is constant, it is AES-specific, and it is estimated as a result of the calibration. Thus, the most likely, if not only possible, source of the abnormality is associated with the term δ f comp.

This term could be erroneous if AES compensation algorithm was not functioning properly, or, which is more likely, if AES received wrong velocity/position data or did not receive the required data at all. In case of the absent for whatever reason data when communication with the satellite is in progress, and a message has to be sent/received, the most logical behaviors of the compensation algorithm would be:

Do not apply Doppler correction, i.e. assume δ f comp=0; Apply Doppler compensation based on the last available position and velocity data.

However, the second option has two potential pitfalls:

It can do more harm than good, especially if an aircraft changes its heading by 180°. The algorithm is more complex, and it would require additional checks at least on the

appropriateness.

An assumption was made in the present study that the velocity/position data was not available to the AES for the estimation of the correction term δ f comp in case “OxlS - Log-on/Log-off Acknowledge” transaction in both the log-in instances, but it was available in the case of the first transaction “OxlO- Log-on Request (ISU)/Log-on Flight Information”. Under this assumption, abnormally long BTOs could be explained by the delays, during which the AES software was pending the required data.

In addition, it was assumed that changes in the aircraft velocity were immaterial during the time interval of 8-9 seconds between “OxlO- Log-on Request (ISU)/Log-on Flight Information”and “OxlS - Log-on/Log-off Acknowledge” transactions.

These assumptions yield the system of algebraic equations:

{ΔFup+ΔFdown+δ f comp+δ f sat+δ f AFC+δ f bias=BFO1 ,Δ Fup+ΔFdown+0+δ f sat+δ f AFC+δ f bias=BFO2 ,

22\*

MERGEFORMAT ()

where BFO1 is the first ‘normal’ BFO value in the sequence (142 and 182 Hz respectively for the first and second log-on event), and BFO2 is the second abnormal BFO value in the sequence (273 and -2 Hz respectively for the first and second log-on event).

This system allows for deriving unknown ground speed and heading from given rate of climb (vertical velocity component) at a given location. In this study, the system (2) was solved numerically; no analytical investigation was undertaken.

4. Results

4.1. Log-on event 18:25The 18:25 position of the aircraft was extrapolated based on the known position 18:22. The

position assumed for the calculations was close to the waypoint NILAM. It is worth of noting that the changes in the position within 3 minutes of flight do not notably affect the solution of the system (2). Ground speed and heading as functions of ROC are depicted in Figure 1. The plot of the ‘possible’ (or ‘feasible’) groundspeed and heading in the vector form is shown in Figure 2 along with the background ‘Lido’ image and radar network (courtesy Don T.).

Figure 1 Groundspeed and heading as functions of ROC for 18:25 log-on event.

Figure 2 Possible groundspeeds and headings for 18:25 log-on event.

The term ‘possible’ or ‘feasible’ is this context is defined as the solution, for which ground speed does not exceed approximately 0.9 Mach at the air temperature of 20°C.

It should be noted the turn at the waypoint NILAM towards the waypoint IDKUT combined with the descent of approximately 5 m/s does not require change in the groundspeed. The descent would also be consistent with the radar readings. In case of the ascent, the aircraft would likely reappear on the radars, particularly RTDADS-III Phuket or Lhokseumawe, one of which was presumably used as the source of radar data (see Figure 2), at least after the gap in the data.

4.2. Log-on event 00:19The same interpretation of the abnormal BFO of -2 Hz during the log-on event 00:19 appears to

be possible in terms of the existence of a meaningful solution of the system of Eqn. (2). However, in contrast to the log-on event 18:25, the position of the aircraft is not even approximately known. Therefore, three possible locations were considered in this study:

1. 88.57°E, 36.02°S, 11 km altitude. This is an approximate terminal location, which was predicted by IG basing on the AP-constrained flight model.

2. 98.995°E, 29.473°S, 7.1 km altitude. This location corresponds to the terminal point of the “Constant Thrust Settings” (CTS) model.

3. 105.5°E, 18.0°S, 5.0 km altitude. This is an approximate location of the northern intersection of the 7th arc and fuel endurance curve presented in ATSB report [1].

The solutions of the system of Eqn. (2) in terms of the ground speed (horizontal) and heading as the functions of ROC for each of the locations are depicted in Figure 3, Figure 4 and Figure 5 respectively.

Figure 3 Airspeed and heading as functions of ROC for 00:19 log-on event, location #1.

Figure 4 Airspeed and heading as functions of ROC for 00:19 log-on event, location #2.

Figure 5 Airspeed and heading as functions of ROC for 00:19 log-on event, location #3.

Remarkably, there are physically meaningful solutions for all the three locations. Moreover, the following observations are made:

All the solutions correspond to the descent rate in the range from approximately 15 to 26 m/s; there no solutions corresponding to positive ROC.

The northern locations along the 7th arc in the Indian Ocean are characterized by a wider range of possible descent rates, as well as by slower descents.

Each of the locations has two pairs of (ROC, heading) corresponding to the same ground speed, except the minimum groundspeed, for which only one solution exists. The latter visually appears to be corresponding to the solution with the heading perpendicular to the 7th arc, but the solution was not analytically analyzed to draw a generalized conclusion.

Possible groundspeeds and headings as functions of ROC for the 00:19 log-on event at the three selected locations are presented in Figure 6.

Figure 6 Groundspeed and heading as functions of ROC for 00:19 log-on event at the three selected locations. Note: the ranges of possible ROCs are individual for each of the locations.

5. Summary and discussion This technical note proposes an interpretation of the two anomalous BFO values recorded

during 18:25 and 00:19 log-on events as the results of the malfunction/intervention/unintended behavior of some electrical/electronic circuits, in result of which AES was not able to apply Doppler compensation for the “OxlS - Log-on/Log-off Acknowledge” transaction.

It was found that under such an assumption:

1. There are physically meaningful groundspeed and heading for both the log-on events.2. There are two solutions for a given groundspeed (horizontal), corresponding to the two

different headings and ROCs.3. The solutions for the log-on event 00:19 for various locations along the 7th arc correspond to

the descent rates in the approximate ranges:a. From 22.5 m/s to 26.4 m/s for the location #1 (terminus suggested by IG).b. From 20.0 m/s to 26.3 m/s for the location #2 (terminus of CTS model).c. From 14.7 m/s to 23.3 m/s for the location #3 (intersection of the 7 th arc and fuel

endurance curve from ATSB report [1]).4. The 18:25 BFOs may indicate a turn towards the waypoint IDKUT with the descent rate of

order 5 m/s.

It should be noted, however, it is not clear, what electrical/electronic malfunction/intervention/ unintended behavior could result in the correct AES compensation applied for first transaction in the log-on sequences (“OxlO- Log-on Request (ISU)/Log-on Flight Information”), and absent compensation for the following transaction (“OxlS - Log-on/Log-off Acknowledge”).

One potential explanation is the chain of following events:

Log-on is initiated upon restoration of the power supply to the AES; ADIRU / SAARU is ready to feed data into the AES instantly as they have their own

power backup, and continued normal functioning during a hypothetical power blackout; Simultaneous powering of some other units triggers ADIRU / SAARU to reset/re-

initialize using data from GPS; Due to the preceding power loss, GPS require some time to tune on (e.g. “locating

satellites”); alternatively, the process of ADIRU reset/re-initialization takes some time; In approximately 7-8 seconds AES requests for the updated position and groundspeed

information, but neither ADIRU nor GPS are immediately ready to provide such data due to the re-initialization in progress;

AES has to complete the log-on process and proceeds with no Doppler correction applied. Transaction is successful as the Doppler shift is not significant in both the cases.

The other possible explanation is that both BFOs in each of the sequences are erroneous. This case was not investigated yet.

6. References 1. ATSB, 26 June 2014. MH370 - Definition of Underwater Search Areas. ATSB Transport Safety

Report AE-2014-054, www.atsb.gov.au.2. Inmarsat, December 2014. Update to Signalling Unit Logs.

APPENDIX A

Table 1. Extracts from Inmarsat’s logs, which contain abnormal BFO and BTO values upon the two log-on events

Time Latency, s Channel Name Channel Type SU Type BTO, μs BFO, Hz18:25:27.421 IOR-R600-0-36E1 R-Channel RX OxlO- Log-on Request (ISU)/Log-on Flight Information =17120-4600 14218:25:28.852 1.431 IOR-P600-0-36FC P-Channel TX Oxll- Log-on Confirm18:25:29.572 0.72 IOR-P600-0-36FC P-Channel TX Ox40- P-/R-Channel Control {ISU)18:25:29.572 0 IOR-P600-0-36FC P-Channel TX Subsequent Signalling Unit18:25:30.213 0.641 IOR-P600-0-36FC P-Channel TX Ox41- T-Channel Control {ISU)18:25:30.213 0 IOR-P600-0-36FC P-Channel TX Subsequent Signalling Unit18:25:34.461 4.248 IOR-R1200-0-36ED R-Channel RX OxlS - Log-on/Log-off Acknowledge 51700 273

00:19:29.416 IOR-R600-0-36F8 R-Channel RX OxlO- Log-on Request (ISU)/Log-on Flight Information =23000-4600 18200:19:31.572 2.156 IOR-P600-0-36FC P-Channel TX Oxll- Log-on Confirm00:19:32.212 0.64 IOR-P600-0-36FC P-Channel TX Ox40- P-/R-Channel Control {ISU)00:19:32.212 0 IOR-P600-0-36FC P-Channel TX Subsequent Signalling Unit00:19:32.852 0.64 IOR-P600-0-36FC P-Channel TX Ox41- T-Channel Control {ISU)00:19:32.852 0 IOR-P600-0-36FC P-Channel TX Subsequent Signalling Unit00:19:37.443 4.591 IOR-R1200-0-36F6 R-Channel RX OxlS - Log-on/Log-off Acknowledge 49660 -2