Direct sequence spread spectrum (DSSS) with secret and long chipping sequences can be used to hide a signal "under the noise floor." It depends on what you mean by encryption, but this can ensure that no one notices the signal and its content remains confidential. One keyword for this is "low probability of intercept" (LPI). This can also be used to prevent jamming, because the signal is not only hidden but more robust to interference that is not spread.
However, this method faces similar challenges like encryption, all authorized receivers must know the chipping sequences in advance, so you have to distribute them and protect them, etc.
thanks Mr.Matthias .. else Can we use packet tracer and trace the needed info and then encrypt and send ?
To Rajarajan K: I think you misunderstood the answer provided by Matthias. It is an excellent answer and probably the only one practically possible. I got this feeling from your second question about the packet tracer and the encryption of the sensitive data. The transmission channel and the Matthias' answer do not quite well relate to the concept of packets. Here is my penny:
Your question was 'how to encrypt data in the transmission channel', meaning in the 'physical channel' (this also is known as the RF channel if you are using a wireless system or Layer-1 in ISO-OSI). It is usually assumed that the attacker (against whom you need to protect your data, possibly by encrypting it) has full access to the transmission channel. In the transmission channel we have signals (this is not quite the same as packets) which we assume that the attacker is able to receive. So here is my explanation of Matthias' excellent answer:
Usually all conventional methods for data encryption involve replacing one bit sequence (e.g. one byte, one packet, etc.) with another -- therefore this can only be done on one of the higher layers in the network, not in the transmission channel. E.g. this could be done in Layer-3, where the data exists as some bytes and bits kept in a certain type of memory. There you can read the bits from the memory, change them and write them to the same memory or to another area. This could be generalised to receiving the bits from one network layer, encoding them and sending them to another layer in the network, but in ALL cases the bits do not travel far, they remain at the location of the transmitter during the entire process of encryption. Of course, it is assumed that no one else has access to the part of the memory where the unencrypted data is kept.
Later, when you transmit the encrypted bits over the medium, such as an RF channel, you usually perform a type of modulation of a carrier. Thus, all protection of your data therough encryption MUST be done before the carrier is modulated and sent over the transmission channel. This also means that any attacks on the encryption of your data will occur not in the transmission channel, but after the signal has been received by the attacker from the transmission channel and demodulated and converted back to a series of bits which will need to be decrypted (the attacker will write the received bits into some type of memory and will try to pick-up a decryption key). To summarise: the methods for protecting the data in the transmission channel are not usually called "encryption" and they are briefly reviewed below.
To protect the data in the transmission channel, one could in principle try using an unknown (secret) frequency for his carrier in the hope that no one would be listening to that frequency in the transmission channel, which of course is naive -- it is safe to assume that the attacker will be capable of listening to all frequencies and will be doing this when we start our transmission, so there is no such thing as a secret frequency. The next step in the sophistication is to use a secret modulation type. This is also naive as technical means exist that allow very quickly to get some idea of the modulation type used. The next level in the sophistication is to use a parameter of your signal in the transmission channel which changes in time following a secret sequence, which sequence is difficult to guess or discover without having a prior knowledge of it. This parameter could be the frequency of the carrier in the transmission channel -- i.e. use a carrier which hops from one frequency to another (this is called frequency hopping) and the sequence of the hops is secret . Then it may be difficult to follow the hopping frequency of the carrier if one does not know the sequence in advance, and time will be needed to find the sequence from listening to the transmission channel. If your hopping sequence is non-repeating and very long then it will take a very long time to discover it. But still this would not be impossible.
Now, the next level in sophistication in the protection of your data in the transmission channel is to use SSS (spread-spectrum signal), which is what the frequency hopping becomes when the hops are done very rapidly (i.e. the frequency is changing continuously) with no dwelling time on any particular frequency. This is what Matthias has suggested as a good practical solution. If you implement an SSS system, then the data will be protected in the transmission channel because of the following points: (1) without knowing the spreading sequence it will take a very long time (it may be practically impossible) to discover it from the signal received in the transmission channel and therefore it will be impossible to de-spread the signal and convert it back to the original bit sequence (now, the original bit sequence could additionally have been encrypted in the higher layers, which would require finding also the decryption method -- this would doubly hinder to attacker); (2) because SSS signals have very low energy per hertz (meaning very low power per unit of bandwidth) they can even be below the level of the natural thermal noise and man-made noise, therefore the attacker will have difficulty detecting when he/she is getting your signal and when she/he is just getting noise in his/her antenna; this will make it even more difficult for the attacker to follow your signal in order to attempt to demodulate/de-spread it. This is precisely what Matthias has clearly explained and what perhaps is the only practical approach to protecting of your data in the transmission channel. Matthias went further to explain, that you should keep the spreading sequence secret and that this spreading sequence must be known (stored) to all your receivers so that they can correctly de-spread your transmitted signal. This storing of the spreading sequence in the receiver must be done in such a way that the attacker does not get access to the stored sequence in your receivers. Also, it is good to change the spreading sequence as often as practical, so that if the attacker has discovered your current sequence, by changing it you force the attacker to start from the beginning. However the distribution of the new spreading sequence to your receivers is vulnerable to the same attacks as those applicable to the distribution of secret encryption/decryption keys. This is what Mathias also explained in his answer. It means that while the data may be very strongly protected by spreading in the medium, it may be very vulnerable if the spreading sequence is not kept secret at all times and not changed as often as practical.
Old engineers, like myself, would also remember that before the advent of SSS (e.g. long before DS-CDMA became the basis for 3G in the cellular networks) a similar principle of protection was applied to the time domain (as opposed to the frequency domain as is the case with SSS). This was termed "scrambling" of the signal and relied on transmitting the parts of your signal in the wrong order. Assume that we work in milliseconds, then the first millisecond of your normal signal was not transmitted immediately, but delayed and transmitted later after some other parts of the signals were transmitted, other milliseconds were also delayed for a seemingly random time and transmitted later (there was, of course, no overlapping between the delayed parts of the signal -- the method made sure that each part was transmitted on its own and that there were no gaps between the transmissions of the different delayed parts). This was for a time used to protect analogue TV signals. An ordinary TV receiver would get the signal off the air but the picture on its screen would have been scrambled like a mosaic. This worked particularly well with TV signals since it would make it impossible to discover the line- and frame-sync impulses (they would come in the wrong place and possibly in the wrong order). Therefore, scrambling in time can be seen as applying the principle of SSS to the time dimension, however this has a number of disadvantages compared to SSS: the signal can easily be detected (although not decoded) because it is transmitted on the same frequency all the time; the strength of the protection is proportional to the delay required for transmitting the scrambled parts of the signal, therefore strong protection would require long delays in processing and reassembling the signal in the correct order, etc. But this is a useful example to illustrate the main principle of protecting your data in the transmission channel: you must vary at least one of the medium-related parameters in a seemingly unpredictable way, so that the attacker will not be able to quickly guess or discover how this is done. This parameter could be the hopping sequence of the carrier frequency (in SSS this is equivalent to the spreading sequence), or the scrambling sequence in time, or a combination of both (which indeed is widely used in order to protect the signal from the damages caused by fading: SSS protects against frequency-selective fading and against narrow-band jamming, scrambling in time protects against dips caused by flat fading and against impulse jamming) -- this was also explained by Matthias.
I apologise for the long-winded answer, I hope that this shed some light on the original and otherwise better answer provided by Matthias.
To recap for one last time: in order to "encrypt" (i.e. to protect) your data in the transmission channel you need to transmit it with at least one medium-related parameter (such as frequency, modulation type, timing, etc.) which is unknown to the attacker and which is difficult to guess or find out (it must take a long time to do so, by which time your data should not be of any interest and importance to the attacker).
The keyword you are looking for is "physical layer security"
There has been quite a bit of research in this area recently.
For instance in wideband channels people try to use the fact that the channel properties (the channel impulse response) is unique between the TX and the Rx so they can use it as a secret key for cryptography. (look for Scholtz as author)
Another approach I have seen is to look at the channel coefficients in MIMO (Jensen M.A. as author).
Personally, I would be a bit careful though. For instance with the channel properties the channel reciprocity is assumed, i.e. the channel is the same in both directions. This is true between the antenna ports but you have to consider that there are power amplifiers and low noise amplifiers at each end of the channel which are not the same at both ends.
This can be a bit of a problem. This is not to say that to some extent it will be possible to use the channel properties for secret key generation but I feel that a lot of "practical" issues have to addressed before we see real-world products.
@Plamen: Thanks for your insightful answer. This is exactly what I meant.
There is another possibility to hide a signal, the keywords are "cooperative jamming" or the "Wyner wiretap channel." The idea is to deploy several "friendly" jammers that make it harder for an eavesdropper to receive a protected packet, but you can choose the deployment so well that the legitimate sender may still receive the message despite jamming. However, this come from information theory, and I would say it will work "mainly in theory."
This should be a good read for anyone interested in PHY security:
http://dx.doi.org/10.1109/MSP.2013.2260875
@Rajarajan: When you do the protection on the physical layer you leave the digital world of bits and bytes, the methods you will use come from the field of electrical engineering there. This means especially that you will need to modify the sender and receiver hardware to get results, or start working on analytical results in information theory or signal processing.
@Vit: I have been working on PHY key generation, I agree with you that it has many practical issues, and the benefits over existing and proven alternatives like asymmetric crypto are small to non-existent for most scenarios. However, it could be a good choice for devices that don't have the computing power to support heavy-weight protocol but have a RF transceiver, like wireless sensor motes. The channel reciprocity deviations are there in practice, but you can design protocols that try to fix the errors. We did a measurement study + a light-weight protocol, and it worked quite well:
http://dx.doi.org/10.1109/JSAC.2013.130911
But I found key generation does not fully fit this topic here because you generate a key and then use it for conventional encryption, so I didn't bring it up in the beginning.
- Badges
- Science topic