Analysis of suspected MH370 pings recorded by Ocean Shield | Hacker News Reader

Analysis of suspected MH370 pings recorded by Ocean Shield

iheartmatlab.blogspot.com.au

186 pointsntakasaki54 comments

More DIY youtube-video-telemetry-data-extraction fun from the illustrious Oona Räisänen:

http://www.windytan.com/2014/02/mystery-signal-from-helicopt...

She actually pulls out data from a binary stream. Her whole blog is amazing, it's been on here many times.

The potential Doppler shift is mentioned at the end. If there were three towed ping detectors and all of them were moving at a fixed speed, would you be able to immediately triangulate the position of the device? How many seconds of ping would you need in order for it to be reliable? I wonder if the software to do that (I guess it must exist) has a very clean UI showing approximate location getting better as more data flows in. (Think the indicator for GPS accuracy in map apps.)

You can do better than Doppler shift. If the clocks on three ping detectors are well-synchronized and their locations known, you can use the ping arrival times to triangulate. Sound propagation in water is nontrivial, but the people who are good at it are good at it.

Furthermore, as long as the clock on the pinger is fairly good, the ocean environment fairly stable, and you don't drop many pings, you can drive one ping detector around and triangulate from the ensemble of ping measurements.

You can do that with a single moving platform, since the target is not moving. Record the bearings for half an hour, make a 90° turn, and recore another half hour of bearing data. That should give a decent approximate location. Its also basic target motion analysis that a military sonar operator can do in their sleep.

> Its also basic target motion analysis that a military sonar operator can do in their sleep.

Which makes me wonder: Does the US not have military assets in the area? Or do they and they don't want to disclose their capabilities?

EDIT: I'm behind the times; I didn't know the US had a sub-hunter in the area.

I think it's safe to assume that someone(US, Russia, China, etc.) has something in the area, but nobody's going to be talking about it to the public.

It's not so much about disclosing capabilities as much as it's just simply operational secrecy. The entire point of a submarine is that it's hard to find.

>I think it's safe to assume that someone(US, Russia, China, etc.) has something in the area

accidentally, this search may find a lot of that "something"

It's in international waters, the Chinese have no authority here and in any event Australia is coordinating the search. Ocean Shield is an Australian Defence Force ship (although not grey like your typical warship).

http://www.theguardian.com/world/2014/apr/07/mh370-black-box...

The US has a specialist sub-hunter in the area, and they've made extensive use of P3 and P8 aircraft. I doubt they're terribly sensitive about their ability to find a deliberately pinging object 6000m below the surface.

"There are 11 military aircraft, four civilian aircraft and 14 ships taking part in the search on Wednesday."

http://www.theguardian.com/world/2014/apr/09/mh370-search-ar...

>since the target is not moving

assumption

True, but is very likely a correct one.

Not really - underwater currents can be quite strong. If you were creating a model to accurately position this thing, you'd be best to add a v⃗ component for both the transmitter and receiver sides of the equation.

I'm not that familiar with the bathymetry[1] of the area, but I have been seeing reports that the locations are between 3000m and 6000m deep. At that depth, even a strong persistent current is probably only a few cm/sec. For the precision of TMA, that's effectively stationary; the noise in the bearing rate measurement is going to be greater magnitude than the true bearing rate.

[1] Why is "bathymetry" not in Chrome's spelling dictionary?

It's in the 10s of cm per second and up to 1m/s in some places, though not SW Australia

http://hypertextbook.com/facts/2002/EugeneStatnikov.shtml

Sadly the towed pinger locator (TPL-25) is omni directional.

But by analysing the inter-ping interval you can determine a CPA point. Something below I wrote on reddit about this technique:

Due to the pulse length (10ms) there is an inherent 100Hz bandwidth in the signal, which is equivalent to the frequency shift you would expect if moving at 4m/s. In other words, Doppler measurements of a single pulse is NOT going to be useful in attempting to localise.

However, because the pinger has a relatively stable repetition rate (about 1.106s from what I could see) you can do a cool trick to localise it (assuming the source is stationary):

- Perform a pass over the area of interest (ie where the pinger is) - measure the ping times of arrival (either manually with waveform inspection or write an envelope detector) - plot the ping times of arrival versus the time of arrival modulo ping interval. ie in MATLAB: plot(toa, mod(toa, 1.106), '.'); - The plot indicates the relative distance from the source / receiver. Hopefully the plot should contain a local minimum representing a closest point of approach (ie like http://imgur.com/v1ZclaX ). This means that the source lies in a line perpendicular to the vessels course centred at the location of the receiver (ie TPL-25) at CPA time - Perform another pass over the area of interest on a course perpendicular to the first - measure the ping times of arrival and look for CPA as before - You should now have two intersecting lines. Guess what should be at the intersection? :)

This relies on having a stable ping repetition rate.

BTW this is the author of the linked post here :)

Rodney

Edit: @ISL has hit the nail on the head there. Hyperbolic trilateration using multiple simultaneous receivers is the way to go. RAAF will be trying this with sonobuoys deployed from P3-C Orions

You can do better than complete waveform time of arrival.

Sound propagation in water is dispersive (higher frequencies travel faster, but lower frequencies travel farther). With an accurate model if the chirp waveform, you can determine distance more accurately by looking at the amount of signal "spreadding" and the amount of attenuation of specific signal components due to this shifting.

You can do better than that, too. Theoretical accuracy of time of arrival distance approximation is directly related to the wavelength of the transmission signal. If you can model the transmitted signal and its transmission path accurately enough that you can get a strong correlation between the received signal and the modified signal (in this case taking into account Doppler effect and wave dispersion), then you can accurately measure both the time of arrival and the phase of the signal. Assuming a perfect model (HA!) the accuracy of the distance calculation is soon dominated by the various noise components (typically your timer noise, power supply noise, decimation noise from your ADC, and various thermal noise sources in both the transmission path and your measurement equipment).

Triangulation requires an angle and only two observers. What you're proposing is trilateration, http://en.wikipedia.org/wiki/Trilateration

I suspect it'd be challenging to get reliable and precise enough measurements from multiple detectors as close to each other as they'd have to be on a single ship. Three ships, perhaps a different story.

I was dubious at first that the movement of the ship would be fast enough for a 4ms delta in the click timing but it comes out to just 6.24 m/s speed if they're moving directly away from the object (this assumes a 1560 m/s speed of sound in water which isn't constant but it's the best I had).

A friend of mine works on ships that do towed array for seismic exploration. The software and systems they have behind this sort of thing is very comprehensive. In that case they are looking for things below the ocean floor. They're able to pick up seismic responses from interesting geological formations in the rock under the ocean while moving along towing a large string of lines behind the boat.

I would think finding a pinging box would probably be simple in comparison, if you knew where to look for the thing.

The change in interval of a few ms could also be because of mp3 binning. Extracting a file of Youtube means you have already been through at least three codecs.

Audio codecs are very good at retaining signal phase as the human ear is very sensitive to errors in this. Whilst I'm not saying it is impossible, I think it would be very unlikely that the interval change is due to transcoding.

Also, as it is a moving receiver this effect is almost expected.

Best would be to contact the video source and ask exactly what the audio track is.

I don't think the video shows the pings from the blackbox. More likely they are the chirp pings from a sidescan sonar. They are looking for an shadow (image) of the blackbox in the data. Here is a nice diagram showing what they are looking for: http://www.southwestcoastalgroup.org.uk/cc_mon_sidescan.html

Doubtful; why would they be worried about the black box pinger's battery running out then? Also, it's much easier to listen for the pings than to look for visual indication, since the pings can be analyzed by DSP software.

The video shows a moving spectrogram of the received signal from the TPL-25 towed pinger locator. It has nothing to do with sidescan sonar.

In fact they explicitly have not launched the sidescan capable Bluefin21 AUV because of fears of acoustic contamination of the towed pinger audio.

According to the Wikipedia, the cockpit voice recorder records only up to last two hours of the talks in the cockpit:

http://en.wikipedia.org/wiki/Flight_recorder#Cockpit_voice_r...

It appears that even once the recorder is found we can just discover that nobody talks anything substantial in the last two hours of the long flight, that is, that "interesting" (to find the motives) conversations are already overwritten.

The flight data recorder can save more flight data but maybe it can only be the confirmation of the flight path already relatively known. Maybe the mystery won't be solved even with these devices present?

FDR records huge array of instruments, this data alone has solved countless crashes.

I'm afraid there's still chance that we'll learn exactly when it crashed etc but not why all this happened if it was really the result of somebody doing something intentionally and everything important happened before the last two hours. The current estimates are already that the plane simply flew on autopilot during the last hours. It's possible that nobody spoke in the cockpit at that time.

It's unlikely that no noise at all was recorded in the cockpit. Other issues have been discovered from non-voice recorded noises - like the 737 Rudder failure problem that was deduced from the noise of a pilot grunting as he tried to move the rudder.

The FDR data will still tell them whether it flew (ie pilot commanded) into the water or crashed (ie ran out of fuel or suffered engine failure), whether it had been on autopilot, whether there was an earlier fire that knocked out communication and navigation etc.

There are plenty of events where the CVR provided no useful information and the cause of the crash was determined by the FDR.

See the accident that killed Payne Stewart[1] for an example of a not-terribly-useful cockpit voice recording.

[1] http://en.wikipedia.org/wiki/1999_South_Dakota_Learjet_crash

Examining the hull once found will dispel a great many theories.

Would there be any chance of recovering data from personal phones/laptops? Assuming the passengers were alive and aware, perhaps some of them tried to document what happened.

The only problem with that idea is that electronic devices may have been scattered on the ocean floor or not survived fire or other destructive forces that happened either during flight or during landing.

I find it annoying that the media call this thing a "pinger". To ping means to send out a signal in expectation of maybe receiving an echo or reply. It's a two-way process. The device's manufacturer calls† their product a beacon because that's what it is. It doesn't listen. It only sends. Until its battery runs out.

Arguably, a real pinger (underwater transponder?) would have much better battery life, needing only to listen for part of each second or minute for a page from a passing search vessel. It would squawk only when it hears such a page, and, encoded within the squawk would be the airplane's serial number, so there would be no doubts about marine mammals or stray locators. Also, the search vessel could measure the ping delay, revealing the range to the wreck. A number of pings from different positions would pinpoint the wreck's location.

The pinger could listen for search vessel pages on one frequency and respond on another. The page could contain a serial number, which the pinger would echo back. So, it would be simple to match page and response for the delay time measurement.

†http://www.radiantpowercorp.com/dukane-seacom-locator-beacon...

In college our underwater acoustics prof used an old copy of the Navy sonar manual to teach some fundamental concepts early on in the course. One of my favourite passages was when the Navy manual definitively answers the question of "if a tree falls in the forest and nobody's around to hear it, does it make a sound?" The answer, according to the Navy is that no - in order for a sound to exist there must be both a transmitter and a receiver component, and if both are not present then the energy emitted by the lone transmitter does not constitute "sound."

So, according to that - you're both right in that it's a two party process. However, I believe the term is appropriate in underwater acoustics for anything which dumbly sends out a relatively predictable signal on a regular interval.

The problem here is designing a microphone that is sensitive enough to listen for a page from a passing search vessel and also survive the crash.

I worked on a project involving this type of underwater audio beacon, and everyone called it a "pinger." So it's not just the media.

The GP's point is well-taken. The company that produces this particular widget calls it an "underwater locator beacon",[1] not a "pinger".

To "ping" implies you are the locator, not the locatee, whether you are sitting in front of a terminal window or are running the active sonar[2] on a boat. You send a signal out and await a response, be it a reply (as in the case of the ping command) or an echo (as in the case of sonar).

[1]http://www.radiantpowercorp.com/dk-series-underwater-locator...

[2]http://en.wikipedia.org/wiki/Sonar#Active_sonar

> To ping means to send out a signal in expectation of maybe receiving an echo or reply. It's a two-way process.

"Ping" is an Onomatopoeia. There is no distinction about whether or not you are supposed to get a response or acknowledgement in the definition of this word - this is your own view of the definition, perhaps biased by years of working with computers, but in the acoustic world a 'pinger' is just something that 'goes ping'.

>I find it annoying that the media call this thing a "pinger".

I find it annoying when people assume their definitions of words they use are the only 'right ones'. Use a dictionary - you'll see the definition right there:

    pinger  (ˈpɪŋə) 
 
    — n
    a device that makes a pinging sound, esp one that can 
    be preset to ring at a particular time

No. I hereby execute a proof by Monty Python: https://www.youtube.com/watch?v=arCITMfxvEc

Re: low battery, what's that circuit look like? I can imagine a real simple RC discharge/charge circuit (ping/delay) that would behave like that. Of course anything quartz-based would behave very differently.

Why no radio ping?

Deep ocean water is opaque to all but the lowest frequency RF.

http://en.wikipedia.org/wiki/Extremely_low_frequency

Which is why, during the Cold War, the US maintained very powerful long wave transmitters, both land-based and airborne. Using VLF†, they could send the nuclear launch codes to the nuclear missile submarines without the subs having to break the surface and create a possible radar return.

They kept an airplane aloft 24 hours a day carrying a SAC general. In the event of war, it could reel out a long-wire antenna a couple of miles long from its tail and send out the codes.

†http://en.wikipedia.org/wiki/Communication_with_submarines#V...

I wonder how this can have worked; according to my understanding VLF transmission for underwater reception needs very much power.

Did they bring really a powerful generator in the plane?

http://en.wikipedia.org/wiki/Naval_Communication_Station_Har...

This says 1 MW power.

I would say a mobile platform could get away with much lower power.

According to the Wikipedia†, the aircraft in question had a 200 kw transmitter, but the article is a bit vague about VLF in particular.

The idea, of course, was that while, hopefully, the ground stations would get the message out, there was that airplane you couldn't count on knocking out in a first strike.

†http://en.wikipedia.org/wiki/TACAMO

Since seawater is conductive, radio waves die out quickly. Thus, these use audio, which travels nicely in seawater.

Interesting, thanks.

I thought the audio frequency that was chosen for the black boxes was one that only came from a man made source?

its too bad that the data from the search process is not made available more openly. they used crowdsourdcing for the image analysis, but as this analysis shows -- its likely the people can really be useful at all stages of the search effort.

kudos for doing this! really great.

All he did was take an average of the pinging interval... I'm pretty sure the analysts actually working on this didn't need to crowdsource that.

In the OPs defense, he not only determined the average ping interval but also provided analysis on what the numbers could mean...without the second part, the numbers don't matter.

I would hope the people working this case could do the same but I also support the idea of crowd-sourcing.

It sounds like a good idea in theory but, in reality, I think you'd end up with a lot of people coming to different conclusions, all convinced that they are right.

The "crowdsourcers" would be demanding that their theories be looked into and, if they were, it would take away from the search operation vital resources.

It reminds me of when I'm troubleshooting network issues and my (very non-technical) boss calls and offers up his ideas of what might be wrong. It distracts me from what I'm doing and simply results in it taking longer to resolve the issue.

Reply on news.ycombinator.com