The ACARS messages are intended to be read by a maintenance crew sometime before the aircraft lands in order that they can be prepared for a repair/ box swap. They are relatively low urgency, low criticality messages; 'anomalies' could be seen for a number of reasons, such as:-
1. Source Message Generation Delay.
The originating sub-systems usually check inputs from sensors, possibly via an AtoD converter, against predefined criteria (range check, rate check, comparison) and may conclude that the input is 'bad'. The corresponding output(s) to other sub-systems over the ARINC 429 would be marked as 'invalid' with a default value; if further readings over subsequent cycles are still bad, then a 'failure' maintenance message may be generated by the sub-system. i.e. the fault goes from being possibly intermittent to 'hard'. The period of time between the first 'bad' reading and the failure message being generated could be 10's of seconds. The message may be given a time stamp when the fault count was started. If the Pitot readings were suspect some time in the minute(s) before 02:10; the AP could then be the first sub-system to complain, then later (message 14) the 'hard failure' was registered and (message 18) the NAV ADR Warning was generated. This possible maintenance message delay before a fault is declared 'hard' could explain why the Failure messages (19 & 20) time stamped at 02:11 are 'after' the message at 02:12 and the one at 02:13 (message 24) is after 02:14.
2. Potential for Missing Messages
It is possible that some messages were not transmitted due to a full buffer within the CMC. I would expect that the message queue handling software would have a wrap around buffer with protection to prevent the buffer 'over filling'. However, the assumption may have been made that only a few maintenance messages per flight would be the norm. The nominal transmission rate of one message every 6 seconds allows for a sustained rate of 10 messages per minute. A buffer size of 10 messages could (with lack of hindsight) seem sufficient. During 02:10 the flood of 15 (or more) messages within a minute may have caused some messages to be 'lost'.
3. Messages not exactly every 6 seconds
From the SITA website (
www.sita.aero/file/1569/Aircom_new_generation_services.pdf) the satellite communication uses a Carrier Sense Multiple Access (CSMA) protocol for Media Access Control (MAC). This mean that before an aircraft starts transmitting, it first checks to see if anyone else is transmitting. To avoid repeated retries if 2 aircraft happen to be in phase on a 6 second cycle, the time between transmissions for each aircraft would have a different, random 'jitter' applied to the basic 6 second cycle.
4. Messages at 15 second interval
There are 3 messages (10, 13 & 16) that all have a received interval gap of 15 seconds; my assumption is that these could be messages that were not acknowledged as having been received by the ground station. Comms protocol usually require that if the transmitting station does not receive an 'ACK' from the end receiver within an appropriate time interval (3 seconds would seem to me to be not unreasonable), it will hold off transmitting the 'next' message and resend the unsuccessful message again. This would give a 9 second delay on top of the normal 6 second cycle.
5. Message 18 gap of 31 seconds
Message 18 seems to be where the message buffered has just been flushed as the time stamp and received time are both in the same minute. That would suggest to me that the gap is due to no messages being generated for 5 cycles of the 6 second transmission slots
6. Loss of Signal
I seem to remember reading somewhere that the INMARSAT link uses a once per minute status message (unrelated to ACARS messages) to monitor which aircraft are still in range. In which case the sat comms would know if the link was 'lost' or not.
Conclusion: The flood of 15 (or more) messages, their interpretation, from diverse sub-systems and all at 02:10 could suggest that the aircraft went rapidly into a chaotic flight state, probably a bit before 02:10. In which case point of impact would be as likely S of LKP as N.