Nyquist - the amazing 1928 BREAKTHROUGH which showed every communication channel has a capacity

In 1928, Harry Nyquist published a paper which would change the course of history [1]. But his original contribution was not the sampling theorem.

Inspired by the work of Fourier, Nyquist discovered that there is a maximum rate at which signals could be sent through a bandlimited channel. For a bandwidth of B, 2B signals per second is the limit (the capacity). This, of course, does not set the limit on how much information you can squeeze into a single symbol/signal, but it shows something remarkable - the bandwidth of the channel limits the signaling rate of a channel. 20 years later, and inspired by Nyquist, Claude Shannon would publish his Mathematical Theory of Communication [2], which combined Nyquist's signalling rate capacity in a bandlimited channel with the impact of noise.

Sources:
[1] H. Nyquist, "Certain Topics in Telegraph Transmission Theory," in Transactions of the American Institute of Electrical Engineers, vol. 47, no. 2, pp. 617-644, April 1928, doi: 10.1109/T-AIEE.1928.5055024.

[VisualElectric_]

Are there are any other historical figures in communication/electrical engineering I should do a video on?

[brownj]

Engineering does not collide with mathematics. Engineering is applied mathematics and applied physics.

[michaelogden]

It's a good think that a few really smart cookies come along from time to time! A really interesting video. Thanks!

[fishPointer]

Surreal to be glued to my screen the whole video and then see <1k views. Your quality is extremely good, I look forward to seeing this channel grow 🙌 please keep making videos

[mixedbytc]

That's not how "Fourier" is pronounced

[CoreyMinter]

Nice. Really like the style, animations, and way of connecting the history.

[MrDuracellHase]

Well made video! I really enjoyed it. Hope your channel gets more attention and that there will be more videos following. Keep up the good work. Thanks!

[dominicestebanrice]

Our cellular communications world is the best example of "to the user, technology, when sufficiently advanced, is indistinguishable from magic". Excellent video BTW!

[jimdigriz]

Nyquist didn’t include the influence of SNR on a channel … Anyone else remember that? Shannon barely got a mention

[Grateful.For.Everything]

Thank You, really enjoy the way You lay all this out! It’s really quite wonderful this ability You have to express knowledge.

[attica]

The Nyquist sampling theorem is essentially about trigonometric interpolation, and the relevant mathematical results were known for at least a hundred years before Nyquist. What was not appreciated was the connection of trigonometric interpolation with electronic communication.

[MirlitronOne]

I know you're a computer, but it's Foo-Ree-Ay.

[arizali_]

Thank you. Very good video. It is sad that it is underappreciated.

[agranero]

As most signals do are not periodic (even periodic ones must start and end some day and are not strictly periodic but we can ignore the high frequency transients at the start and end) we can not use Fourier series, we use a generalization of them: the Fourier transform where we have a continuum of frequencies and not only the harmonics, basically the um turns into an integral.
This is why Nyquist approach uses samples, but it is a natural way of thinking (as you say accidental) as in practice you can't really go to an infinite resolution.
Apart from that a very nice video.

This duality is brought by Fourier transforms (and more generally by all integral transforms) in many places not only in communication: in Quantum Mechanics for instance you can pass from position to momentum by a simple Fourier transform what leads directly to Heisenberg principle. This is the power of abstract algebraic structures: a simple unifying way to treat several at first apparently unrelated things. Summarizing: several ways to see the same process and a unified way to see several different processes.

[alexanders]

And not a word about Kotelnikov. Although he described in 1933 about the capacity of the channel. Kotelnikov V.A. On the transmission capacity of the ether and of cables in electrical communications

[michaelmoorrees]

Author of this video should look up how Fourier is pronounced. Fourier was French.

8:00 - Nyquist criteria. Claude Shannon shows that if you include signal to noise factor, you can really pack more data. In early modems the data signal could still be transmitted successfully, even if it was deeply buried in the noise. As modems got faster, constellation techniques got more complicated. The last old fashion telephone line modems were rated for 33K/56Kbaud. The 56K rate was only possible on a really noise free line, while 33K was a more realistic number. Old American analog telephone lines have a band from 300Hz to 3400Hz, and when digitally sampled, an 8000s/s rate was usually the minimum rate to meet the Nyquist criteria, when practical filtering is included.

[GaryL]

As a old Electronic Tech I respected the Nyquist limits but got lost when QAM came along. My mind just could not comprehend the capacity increase from 50 Baud to 9600 of a 4k bandwidth channel. Much respect to those who really understand this.

[jrjr]

Fantastic presentation about some truly special people.

[stevetaylor]

“It is difficult to understate the importance of electrical telegraphy…” I think you mean difficult to overstate…

[tedn]

Wow I learned about nyauist from his sampling theory. Interesting how he saw it from the opposite way.