Take the 2-minute tour ×
Cryptography Stack Exchange is a question and answer site for software developers, mathematicians and others interested in cryptography. It's 100% free, no registration required.

I have a basic understanding of how strong encryption works on digital data. Bits can be changed so that they appear completely random and map one-to-one back to the original set, given the correct decryption key (and possible other parameters).

But we always talk about digital encryption, never anything that would work on an analogue system as far as I know. Good encryption didn't exist until relatively recently (or cryptanalysis would not have been as useful as it was during World War II) while phones and televisions, both using mostly analog signals until the internet came along, have existed for much longer. To my knowledge, nobody seems to have ever come up with secure and practical ways of encrypting them without converting it to a digital signal first.

Upon searching for analog encryption systems I came across DigiCipher (the successor to Videocipher), but it seems to use binary data since Wikipedia mentions that it uses an mpeg-2 compressed stream. Other sites mention that scramblers were used to prevent television channels from being watched by non-paying consumers, but scramblers have little to do with encryption since they (as the name suggests) are not intended to make the message unintelligible. Wikipedia's secure voice article only briefly mentions analogue signals and then goes on to describe different digital systems.

The only real analogue encryption scheme I found was on that Wikipedia page about secure voice communications. It works by adding noise before transmission and subtracting it on the receiving end. This sounds similar to a one-time pad, but it probably wasn't used only once, which means that all encryption became useless if the phonographic record(s) with the noise were stolen.

Thinking this through, digital encryption algorithms may be used to generate new analogue records so that they're no longer a one time pad. A secret key and variable number (e.g. yymmdd) could be used as input to a digital encryption algorithm, which then produces the noise to be added to the analogue signal. Modern (secure) encryption algorithms are not doable by hand at a reasonable speed far as I know though, so I'm not sure how practical this idea is when all you have is mechanical or analogue technology. They would have needed a computer.

Have we not (yet) figured out how to securely encrypt signals in a truly analogue way (without the use of impractical one time pads), is it impossible (proven, somehow), or are existing algorithms just obscure because there are few applications for them (now that almost everything is digital)?

share|improve this question

migrated from security.stackexchange.com Feb 21 '14 at 10:57

This question came from our site for Information security professionals.

Analog data should be much more resistant to corruption in transit. If you flip a bit inadvertently in a digital encrypted stream, you can rely on error correction codes to get back to the pristine state. A chunk of random interference noise will turn an analog stream stream with digital encryption into undecipherable gibberish. –  Deer Hunter Feb 20 '14 at 4:41
Cryptophon 1100 is mostly analog time-scrambler with digital PRNG. It converts signal to digital form to simplify time-scrambling, but the scheme could be made purely analog, although it is seems to be impractical. –  aland Feb 20 '14 at 11:56
@aland It's not what I'm looking for since it has already been cracked, but thanks for your reply! –  Luc Feb 20 '14 at 16:15
Downvoter: I don't mind you downvoting, but please tell me what was wrong so I can improve it! –  Luc Mar 15 '14 at 18:45

2 Answers 2

It depends on your definition of "Encryption". Encryption; as opposed to encoding is some method of scrambling the text or signal. Enigma was arguably one of the best text encryption methods ever devised. There have been several people working to adapt a rotor-based encryption system for voice. Not sure if any have worked out. Digital is faster and more secure against computer-based cryptanalysis but analog is still very effective against all but the most sophisticated eavesdroppers.

share|improve this answer

Some people define "encryption" in such a way as to exclude analog techniques -- there is no such thing as "analog encryption", they would say, because any analog method of obscuring the contents of a transmission is, by their definition, "scrambling", not encryption.

There have been several such scrambling ("analog encryption") techniques, including:

voice scrambling:

  • some early analog secure voice techniques, as you already mentioned, which added noise (or classic symphony music) from a phonograph at one end, and subtracted the "noise" from an identical, synchronized phonograph at the receiver.
  • Some analog voice scramblers do frequency inversion
  • Some analog "rolling code" voice scramblers frequently jump to a new frequency, hopefully faster than any eavesdropper can track, similar to frequency-hopping spread spectrum
  • Various combinations of the above -- for example, the analog A-3 and the 5B machine built to "decrypt" it which "the speech spectrum was being split into 5 bands, inverted and translated ... the cipher controlling the switching repeated cyclically after 36 sections."
  • I've heard rumors that some early speech scramblers used an analog vocoder to dissect voice (using a bunch of analog bandpass filters and an analog envelope detector for each one) into a few "control signals", each of which was transmitted on its own (very narrow) analog band (each one independently frequency-hopped, and random gibberish transmitted on unused bands?), and then the received control signals were used to reconstruct an approximation to the original voice.

Some analog "television scrambling" techniques:

  • Some scrambling systems transmit the visible part of each line exactly the same as normal analog video, but mangle the Horizontal Sync so that standard analog TV receivers can't lock on.
  • Some scrambling systems invert the black-white luminance of some lines, each line inverted/normal according to a bit from some pseudorandom number generator synchronized with the video signal.
  • Some scrambling systems invert the red-blue chroma of some lines, each line inverted/normal according to a bit from some pseudorandom number generator synchronized with the video signal.
  • Several scrambling systems use various combinations of the above.
  • Videocipher I, Leitch Viewguard, VideoCrypt-S, and Nagravision Syster all scramble video by sending the video scan lines in a scrambled order.
  • Telefirst ("Tele1st") apparently used some sort of analog scrambling technique (?)
  • Zenith Z-Tac cable scrambling system apparently used some sort of analog scrambling technique (?)
  • VideoCrypt I, VideoCrypt II, and Eurocrypt use line cut-and-rotate -- each line of the image is cut into 2 parts and swapped; the exact cutpoint is determined by a pseudo-random sequence generated by a smartcard from a seed value.

All of these systems were developed under severe cost constraints, so they sacrificed some security for lower cost to manufacture legitimate receivers. So for a long time there was something like an arms race between the "defenders" (pay-per-view broadcasters, military radio manufacturers, etc.) and the "attackers" ( pirate decryption, military codebreakers and cryptanalysts, etc.).

share|improve this answer
Thanks for the elaborate answer, but still, none of these do actual "encryption" (where I define encryption as a signal that cannot be distinguished from white noise) without digital components. I wouldn't trust any of these to be illegible to an advanced attacker, like I would with AES, and thus not use them for anything beyond obfuscation. –  Luc Mar 15 '14 at 20:09
I dunno; I suspect that with a bit of tweaking the vocoder + frequency hopping using a "good" pseudo-random number generator + random chaff would be pretty close to indistinguishable from white noise. –  David Cary Mar 15 '14 at 22:14
I agree that most of these systems are mostly just obfuscation. Most people define things like the Playfair cipher as "encryption", although it is relatively easily distinguished from white noise. Even using your unnecessarily strict definition of "encryption", I suspect that it may be possible to build an analog system that would be practically indistinguishable from white noise; with a vocoder + frequency hopping using a "good" pseudo-random number generator + random chaff. –  David Cary Mar 15 '14 at 22:23
But how would the random number generator work? How could you deterministically generate an analog signal without a digital system? So something that could have worked in WO I/II, where microprocessors to generate/process digital data were not available. –  Luc Mar 16 '14 at 0:10
I mean WW I/II, not WO. WO is "wereldoorlog," the Dutch term for world war. My bad. –  Luc Mar 16 '14 at 0:16

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.