# Confusion in Channel Encoding and Convolutional Encoding

If I'm making NRZ signal by first encoding 1 input bit into 10 bits multiplied by -V/+V before sending it into a channel, is it called channel encoding? If I then modulated it using GMSK technique, can it be considered Convolutional Encoding?

• Yes, adding 9 bits of redundancy to each bit of info is a channel code. So, what's your actual question? Is it "What is the definition of a channel code?" That would be a sensible question, since you seem to have a slight misunderstanding of what convolutional coding is. (and why the physical modulation is not part of channel coding) – Marcus Müller Apr 22 '17 at 17:41
• I don't know why the physical modulation is not a part of channel coding. Can you please help me in understanding the concepts of channel coding and convolutional coding? – anonymous Apr 22 '17 at 17:48
• I can point you to the wikipedia pages, which are longer and better than what I can write here? – Marcus Müller Apr 22 '17 at 17:49
• I have been trying to get these concepts from different places but I am getting more and more confused. Your help will mean a lot! – anonymous Apr 22 '17 at 17:53
• It's actually not that easy to recommend a book without knowing what your background is; you could start with services.bibliothek.kit.edu/primo/… (which I think has a high chance of your uni library having it, if you're studying). But: this might not be what you specifically need. As said, it's hard to know what to recommend. If you can, ask someone who's doing digital communications with the same background as you (or your digital comms professor, if you're at a university) – Marcus Müller Apr 22 '17 at 18:31

Like other comments, I don't really understand your question. What I am trying to do is to list basic things so that other ones can suggest edit because I find it is much easier to write in answer part :). Please let me know if it makes you less confused.

The general communication system is

At transmitter, before modulator is digital, modulator is the block to convert from digital world to analog world because physical signal is transmitted in analog (at least at our normal scale). Modulator converts bit sequences to voltage (NRZ), amplitude and phase of sinusoid (QAM), frequency (FSK), phase of of sinusoid (MSK, GMSK), ...

Signal is transmitted in physical channel always suffers degradation (noise, fading, ...) which is random. To combat this negative effect, channel coder adds redundant information to help us to recover the correct signal. For example adding 9 supplementary bits instead of sending only one, there is the chance that more than 5 bits are correctly decoded and with majority rule, the transmitted data is recovered. The operation is called channel coding. There are a lot of channel codes : turbo code, LDPC, convolutional code, ...

Thus, channel coding and modulation are two different steps that have different roles. For more details, you need to read the recommended resources in the comments.

It seems that you understand "convolutional coding" as modulation. Maybe it comes from the fact that signal transmitted over channel is modeled as convolution operation ? People usually use "convolutional code" to talk about a specific family of code in channel coding step.

About the source coder block in the figure, on the contrary of channel coder that adds redundant information to combat random degradation, source coder is here to remove redundant information from source. Huffman code, Shannon code, ... are famous. Applications such as winrar, winzip, ... are well-known. Source coder and channel coder have different roles, but they are related in the sense of "optimality" because they both deals with "information". Thus it is not always correct to say that source coder and channel coder are independent. You can search for "joint source channel coding" if you are interested. The reason why the myth exists and an counter-proof can find in this article Source Channel seperation Theorem. Note that it could be an advanced topic, depending on your background.

• It is interesting how Source Coding typically removes redundancy and Channel Coding typically adds it back in! (Although a systematic redundancy of course but still interesting) – Dan Boschen Apr 23 '17 at 0:17

Channel coding adds redundancy (making 1-bit to 10 bits what ever u mentioned; 9 redundant bits are added)

Whereas GMSK is a form of modulation technique. Since it is a form of continuous phase modulation technique and the partial response signalling concept used in GMSK, the instantaneous phase not only depends on the current bit but also previous bits. The span of the Gaussian filter decides the total number of bits on which instantaneous phase depends.

The above statements can be correlated to a convolution encoder. Where output bits depend on current and previous bits.

Thus GMSK has inherent convolution encoding (rate-1). since rate=1, do not add any redundancy.

Hope the explanation is clear