HydrogenAudio

Lossy Audio Compression => Speech Codecs => Topic started by: QuantumKnot on 2006-01-18 02:13:08

Title: Speech coding for security and defense apps
Post by: QuantumKnot on 2006-01-18 02:13:08
Hi everyone,

I've heard that ultra low bitrate speech coding (< 800 bps) that use simple modifications to the LPC vocoder (such as segmented LPC vocoders), are of interest mainly for security and defense applications.  At such low bitrates, I assume that intelligibility is the only criteria here.  My question is why do we need to go to such low bitrates for defense and security?  Is it because we want to reserve bandwidth for error-correcting redundancy?  Or perhaps encryption?  The only thing that I've read which sort of remotely justifies the low bitrate is a lesser chance of being intercepted.

If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.

Thanks,

QK
Title: Speech coding for security and defense apps
Post by: clintb on 2006-01-18 03:12:07
Possibly because they (the men in black) can store more of your communications when encoded at a lower bitrate?  Oh crap, I've figured it out...they'll probably have some aliens abduct me now. 
Title: Speech coding for security and defense apps
Post by: boojum on 2006-01-18 03:32:21
I would ask the people who tell me this stuff.  If they know that much, they must know the whole story, eh??   
Title: Speech coding for security and defense apps
Post by: Shade[ST] on 2006-01-18 03:48:06
Why not steganography?  At such a low bitrate, you can practically hide information in powerline level fluctuations and no one would notice...
Title: Speech coding for security and defense apps
Post by: saratoga on 2006-01-18 04:42:47
Bandwidth is a good thing to save whenever you are doing radio communications.  It lets you use way narrower frequency bands, less power, etc.
Title: Speech coding for security and defense apps
Post by: HotshotGG on 2006-01-18 04:46:01
Quote
Why not steganography? At such a low bitrate, you can practically hide information in powerline level fluctuations and no one would notice...


When I started reading about Steganography I was amazed with what you could do with it. 

Quote
If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.


Why do we need GPS to begin with?  . GPS was mainly a government thing, until we turned it over for commercial use.
Title: Speech coding for security and defense apps
Post by: QuantumKnot on 2006-01-18 05:24:37
Quote
Bandwidth is a good thing to save whenever you are doing radio communications.  It lets you use way narrower frequency bands, less power, etc.
[a href="index.php?act=findpost&pid=357963"][{POST_SNAPBACK}][/a]


Oh, that makes sense.  My initial thoughts were that bandwidth was rarely a problem these days (if we can pump streaming video over the cellular network, I'm sure we can handle the CELP and MELP bitrates of 4.8 kbps or 2.4 kbps, respectively).  But the communication channels used by defense and security agencies may have much narrower bandwidth than the ones for civilian use, hence the need for lower bitrate. Yes, that makes sense.
Title: Speech coding for security and defense apps
Post by: Shade[ST] on 2006-01-18 05:30:46
Quote
Quote
Bandwidth is a good thing to save whenever you are doing radio communications.  It lets you use way narrower frequency bands, less power, etc.
[a href="index.php?act=findpost&pid=357963"][{POST_SNAPBACK}][/a]


Oh, that makes sense.  My initial thoughts were that bandwidth was rarely a problem these days (if we can pump streaming video over the cellular network, I'm sure we can handle the CELP and MELP bitrates of 4.8 kbps or 2.4 kbps, respectively).  But the communication channels used by defense and security agencies may have much narrower bandwidth than the ones for civilian use, hence the need for lower bitrate. Yes, that makes sense.
[a href="index.php?act=findpost&pid=357969"][{POST_SNAPBACK}][/a]

Having such narrow channels makes them much harder to detect, also, if you don't already know their frequencies. (try finding a pin in a haystack)
Title: Speech coding for security and defense apps
Post by: audio_geek on 2006-01-23 05:49:30
Do you need lossless eompression for speech communication between two high class delegations or for government purposes or security agencies ??
bandwidth is of course the BIG issue but do u see anything good in lossless compression of speech, though lossless compression is there in audio.
Title: Speech coding for security and defense apps
Post by: HbG on 2006-01-23 15:44:11
The less bandwidth the easier to transmit and thus the easier you can guarantee that the message arrives, even under tough circumstances, i guess that's underlaying thought.

Submarine and subterranean (mines) wireless communications for example rely on low bandwidth VLF/ULF and even ELF radio.
Title: Speech coding for security and defense apps
Post by: pepoluan on 2006-01-23 19:10:24
Quote
The less bandwidth the easier to transmit and thus the easier you can guarantee that the message arrives, even under tough circumstances, i guess that's underlaying thought.

Submarine and subterranean (mines) wireless communications for example rely on low bandwidth VLF/ULF and even ELF radio.
[a href="index.php?act=findpost&pid=359212"][{POST_SNAPBACK}][/a]
Ah you beat me to it 

Ever see the movie "Crimson Tide"? The VLF, ULF, and ELF waves are the only ones capable of penetrating the ocean. It requires antennae as long as a submarine, because the wavelength is very long.

Unfortunately at such low frequency, the bandwidth is very very limited, which is why in the film they got the message only partially. Think of a 300 bps modem. Even slower.

And don't forget that you need robust error correction for military orders. So everything that can save bandwidth is very helpful.
Title: Speech coding for security and defense apps
Post by: HbG on 2006-01-24 00:11:48
Actually the global submarine communications transmitters at 82 and 7-something HZ only do a few symbols per minute, so all you can really do is tell a sub to rise and seek contact via other means.

But there's plenty of other military and secret situations where data rates of 800bps are possible and voice preferred.

I don't think i've seen Crimson Tide (or read Tom Clancy for that matter), but wikipedia is a wonderful resource.
Title: Speech coding for security and defense apps
Post by: QuantumKnot on 2006-02-09 22:55:42
Thanks for the replies everyone.

Another reason I read in a paper is that with low bitrate, only low power transmitters are required, which I assume are harder to intercept.
Title: Speech coding for security and defense apps
Post by: Garf on 2006-02-09 23:20:31
I would say another advantage is that low bitrate = short messages = short transmission time = less time for tracking sender?
Title: Speech coding for security and defense apps
Post by: jmvalin on 2006-04-10 16:24:06
If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.


My guess is actually that they want to use huge error-correcting codes to make it robust to jamming. With a very small bit-rate, you can afford a lot of redundancy and eventually (with lots of jamming/interference), it may actually sound better than a higher-bitrate codec.
Title: Speech coding for security and defense apps
Post by: HbG on 2006-06-28 01:49:38
802.11 based wireless networks extend range by switching to lower bitrates, when reception gets poor. Currently this scales from 55 down to 1mbps, yielding a 23db increase in receive sensitivity in typical  implementations.

Military applications may do the same, and a 800bps voice channel may be at the extreme end of this.

Another way to look at it is that this flexibility effectively allows you to trade bandwidth for transmission power, hence it may also have a purpose for transmitting when low on power.
Besides the military this makes it useful for intelligence purposes where bugs have to be small and thus have small batteries, it may be more energy efficient to compress a voice signal and transmit it at low power. Question is whether such a bug's microphone can provide a clear enough signal for a 800bps stream to remain intelligable.
Title: Speech coding for security and defense apps
Post by: pepoluan on 2006-06-28 19:00:22
I would say another advantage is that low bitrate = short messages = short transmission time = less time for tracking sender?
Uh, Garf, shouldn't that be low bitrate = long transmission time?

I'm more to the error-correction reason.
Title: Speech coding for security and defense apps
Post by: Garf on 2006-06-28 19:07:32
I would say another advantage is that low bitrate = short messages = short transmission time = less time for tracking sender?
Uh, Garf, shouldn't that be low bitrate = long transmission time?


No.
Title: Speech coding for security and defense apps
Post by: HbG on 2006-06-28 19:40:14
I would say another advantage is that low bitrate = short messages = short transmission time = less time for tracking sender?
Uh, Garf, shouldn't that be low bitrate = long transmission time?

I'm more to the error-correction reason.


He means the low bitrate of the voice codec, not the data link. When, for example, you would transmit a 800bps voice message over a 1Mbps wireless link it would go at 1250x realtime, meaning a 10 second message takes a mere 8ms to transmit. This makes detection extremely difficult.

I know for a fact that such very short transmission time communications were used during the cold war to communicate with spies.
Title: Speech coding for security and defense apps
Post by: maxonen on 2006-08-09 14:58:21
If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.


When using low bitrates you can use spread-spectrum algorithms to distribute the signal power over a very  wide range - even to a degree where the signal power will be lower than the background noise - and thus "hide" the signal in the background noise and make it practically impossible for any other party to even detect that you are transmitting something.
Title: Speech coding for security and defense apps
Post by: Woodinville on 2006-08-09 19:01:44
Because the channels they use for communications are beyond what we would regard as "really, really awful".

It's a channel issue mostly.
Title: Speech coding for security and defense apps
Post by: kwwong on 2006-08-13 08:26:41

If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.


When using low bitrates you can use spread-spectrum algorithms to distribute the signal power over a very  wide range - even to a degree where the signal power will be lower than the background noise - and thus "hide" the signal in the background noise and make it practically impossible for any other party to even detect that you are transmitting something.


Can you enlightened me further in this area? How about some detail equations presentations?
Title: Speech coding for security and defense apps
Post by: Garf on 2006-08-13 10:49:24


If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.


When using low bitrates you can use spread-spectrum algorithms to distribute the signal power over a very  wide range - even to a degree where the signal power will be lower than the background noise - and thus "hide" the signal in the background noise and make it practically impossible for any other party to even detect that you are transmitting something.


Can you enlightened me further in this area? How about some detail equations presentations? 


http://en.wikipedia.org/wiki/Spread_Spectr...ecommunications (http://en.wikipedia.org/wiki/Spread_Spectrum#Spread-spectrum_telecommunications)
Title: Speech coding for security and defense apps
Post by: HbG on 2006-08-13 11:30:03
I recall bluetooth is a good example of spread spectrum communications, hopping the entire 2.4GHz ISM band.

How you can recieve something when the signal is below the noise level? Doesn't make sense to me.
Title: Speech coding for security and defense apps
Post by: beto on 2006-08-13 15:07:10
It's against common sense but it works.
See CDMA (http://en.wikipedia.org/wiki/Cdma) for a more technical explanation (and if you are familiar with mobile telephony.
Title: Speech coding for security and defense apps
Post by: kwwong on 2006-08-14 11:55:32


If anyone can enlighten me as to why we need such low bitrates (< 800 bps) for security and defense, that would be splendid.


When using low bitrates you can use spread-spectrum algorithms to distribute the signal power over a very  wide range - even to a degree where the signal power will be lower than the background noise - and thus "hide" the signal in the background noise and make it practically impossible for any other party to even detect that you are transmitting something.


Can you enlightened me further in this area? How about some detail equations presentations? 


http://en.wikipedia.org/wiki/Spread_Spectr...ecommunications (http://en.wikipedia.org/wiki/Spread_Spectrum#Spread-spectrum_telecommunications)


Opps, my data communication knowledge is ancient as the egyptian mummies... 
Title: Speech coding for security and defense apps
Post by: Woodinville on 2006-08-14 20:15:49
How you can recieve something when the signal is below the noise level? Doesn't make sense to me.


Perhaps you need to consider the meaning of "below the noise level"?

With a narrowband signal, as you narrow the bandwidth, the noise level drops (we're assuming wideband gaussian, for now), does it not?

With a wideband signal, one needs to consider what happens when one compacts (demodulates) the signal, without compacting the noise (which is noise, after all, yes?).
Title: Speech coding for security and defense apps
Post by: kwwong on 2006-09-03 09:30:01
How you can recieve something when the signal is below the noise level? Doesn't make sense to me.


Perhaps you need to consider the meaning of "below the noise level"?

With a narrowband signal, as you narrow the bandwidth, the noise level drops (we're assuming wideband gaussian, for now), does it not?

With a wideband signal, one needs to consider what happens when one compacts (demodulates) the signal, without compacting the noise (which is noise, after all, yes?).


Sounds like a Wiener filtering problem to me?
Title: Speech coding for security and defense apps
Post by: cabbagerat on 2006-09-03 13:09:32
I think Garf's theory that it is to avoid direction finding is closest to the target. Although PN spread spectrum technologies are making this less critical than it once was, transmitting less data makes it much easier to avoid Huff-Duff (http://en.wikipedia.org/wiki/Huff-Duff) type detection and tracking. Passive detection is still very useful and can detect transmitters of even very low level signals (even ones using spread spectrum like technologies).

Also, people often assume that military equipment is the best and the fastest. This isn't really true for two reasons. The first is most obvious - that equipment needs to be extremely rugged, which is hard to do with cutting edge technology. Another is that there is a lot of military equipment that is very old, stuff that was bought in the seventies (or earlier) and for a number of reasons is still in service. This isn't always a bad thing (it works, so why replace it with something possibly less reliable) but is mostly due to budgetary constraints. Militaries generally either have very nice new equipment or really old stuff.
Title: Speech coding for security and defense apps
Post by: DonP on 2010-11-09 13:20:54
Another consideration, and I really don't know the answer, is how many channels do you want to have available in a war zone if the terrain is such that it has to go through a satellite or high altitude aircraft based repeater?

On first read of the question, I was thinking about NRO sort of stuff where they may be archiving all phone calls in a region.