I'm not from EE background and try to understand how digital is sent in a channel as a non-technical audience.

Let's say we use fiber optics to transmit digital data 010101....,

Since fiber optics use light to transmit information, I can picture something like this:

  • A torch can flash light
  • a flash means 1
  • no flash means 0
  • this torch is connected to one end of the fiber optics cable
  • as long as the torch's flash rate is fast enough like flash/no flash X 100 million times per second
  • theoretically we can transmit as many bits as we want if we can have a torch with unlimited flash rate
  • how many bits we can transmit is really determined by the torch, not the bandwidth of the channel (fiber optics)

I know my assumption is wrong, could someone give me an intuitive example to explain why it doesn't work like this with little EE terminology?

  • \$\begingroup\$ It's both you need to consider. \$\endgroup\$
    – Andy aka
    Commented Jun 4, 2020 at 14:51
  • \$\begingroup\$ Do you want an answer specific to your fiber optics example, or do you want a general answer for all channels? \$\endgroup\$
    – The Photon
    Commented Jun 4, 2020 at 15:04
  • \$\begingroup\$ Your manual example is fine, the issue is that you are turning the light on and off so slowly that the limitation of the bandwidth just isn't entering into your thinking. Imagine in instead trying to send information with a bell: you can do it, but not very quickly as the speed at which you can ring and damp the bell is rather limited, ie, it has a low bandwidth. Try to push data at extreme rates down fiber, and you run into similar issues. \$\endgroup\$ Commented Jun 4, 2020 at 16:18
  • \$\begingroup\$ @The Photon I want an general answer for all channels \$\endgroup\$
    – amjad
    Commented Jun 5, 2020 at 10:11

3 Answers 3


so how many bits we can transmit is really determined by the torch, not the bandwidth of the channel(fiber optics)

The "channel" includes everything: sender, receiver, and medium. Realistically light communications are limited by how fast you can turn the light on and off (known as "rise time" and "fall time"), and how fast the sensor can respond to that.

Diffraction may become important as well; the front of the wavefront can interfere with itself, so what starts out as a clean sharp rise at the transmitter may smear out along the way.

  • \$\begingroup\$ Thank you for your answer. So does it mean that fabric optics tunnel itself can have unlimited capacity so we can transmit as many bits as we want, as long as we can build the torch that can flash in any rate \$\endgroup\$
    – amjad
    Commented Jun 5, 2020 at 10:13
  • 1
    \$\begingroup\$ Unlimited? Nothing physical is ever truly unlimited: remember that light is quantized into photons.Timing-based signalling is going to be limited by how well you can construct a clock. As the signals get smaller the risk of mistaking noise for signal gets higher, and you can never eradicate the noise. \$\endgroup\$
    – pjc50
    Commented Jun 5, 2020 at 13:56
  • 2
    \$\begingroup\$ As a completely trivial upper limit, the frequency of the light itself is a problem: you can't send a signal shorter than one wavelength. \$\endgroup\$
    – pjc50
    Commented Jun 5, 2020 at 13:57
  • \$\begingroup\$ my question really is, is the capacity of a channel determined by the torch capacity only, not by the medium like fabric optics? \$\endgroup\$
    – amjad
    Commented Jun 6, 2020 at 8:28
  • \$\begingroup\$ The second paragraph of this answer contradicts the first. \$\endgroup\$
    – Simon B
    Commented Apr 26, 2021 at 21:11

You should research the very powerful Shannon-Hartley Theorem, which gives the maximum transmission rate of a channel -- any channel -- in terms of the channel's bandwidth and its signal-to-noise ratio. This provides a fundamental limit on information transfer rate.


OK I can give you the easiest way to logically understand what is going on. So lets do a thought experiment in the huge arena of time-invariant linear systems, and the basic idea of stimulus/response (also called input/output). Physicists and especially engineers analyze a huge category of problems assuming they can be analyzed with what are called linear differential equations. Engineers go to great lengths to design systems around the advantageous properties of linearity which among the advantages means that sine waves and cosine waves** when applied (as stimulus) to a linear system, generate in the response no additional sine waves of frequencies not present in the stimulus. So now that we establish the importance of linearity and sine waves, let's consider non-linear systems. (BTW the propagation of electromagnetism in free space, and in electrical and optical materials within certain power limits, is a linear system process thank Goodness.)

In communications, all modulation techniques (modulators) are non-linear, non time-invariant which (by the previously mentioned properties of linearity) means that with a constant amplitude sine wave*** applied to an example of such, a multitude of different sine wave frequencies result in its response (which we call its 'output'). The modulator output is what we call a modulated signal. This is the case for all means of electrical communications including optical, as the only differences between say coaxial cable, microwave links, and optical fiber is the matter of where in the electromagnetic spectrum the range of each lies. All of the information transferred is made up of the patterns of diverse sinusoids (a general term for sine and cosine functions) present in the signal. The more information per second (we say bits/sec) the more spread out this multiplicity of frequencies is.

Now suppose a hypothetical channel is ideally the narrowest bandwidth possible say zero BW, and you apply a signal that has been modulated to this channel and the channel straddles the original unmodulated sine wave signal. Now even though a channel like this is not possible even in theory, we can pretend it exists and say the channel is an "infinitesimally narrow passband filter" in engineering parlance and we apply the modulated signal to its input.

So this infinitesimally narrow channel would have, in its output, a strange effect on the modulated signal, as would any virtually zero BW or extremely narrow channel, by the properties of linear systems. The transferred signal would be stripped of its modulation and so the channel output would be the original constant amplitude sine wave applied to the modulator, in which the only information present would be a scalar value for its phase. No other data would be present. The longer this sine wave would be measured and analyzed, the effort would only increase the accuracy with which the scalar value would be established, based on information theory but that is it - the channel is useless for anything else.

So by this extreme example of showing why zero BW is not tolerable I hope to have shown why progressively larger BW would let through progressively more information per second or bits/sec. And this is actually quantified by the famous Shannon-Hartley theorem which is the governing principle for all types of coding/modulation, of which the questioner's on/off states (on/off keying) is the simplest and is not particularly bandwidth efficient. So it cannot be intuited by the maximum frequency possible of on/off states (or the related amplitude keying), whether optical or electrical. This last pitfall is something students (myself included, way back) are tempted to employ to grasp what is going on here but it is a pitfall.

** ~ what are called 'eigenfunctions' of linear systems ~

*** If it were varying amplitude, it would not mathematically be a sine wave.

  • 1
    \$\begingroup\$ Try breaking your wall of text into a few logical paragraphs. 2 x <Enter> in the markdown syntax. \$\endgroup\$
    – Transistor
    Commented Apr 23, 2021 at 21:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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