Coding, Modulation & the relationship between frequency and data rate

[Turksen]

Hi,

This is a general question rather than an explicit question, so apologies for not using the template.

I'm a bit confused by the relationship between frequency and data rate in a communications system and wondered if someone could help clarify please.

Say I want to achieve a data rate of 10Mbit/s - 10 millions bits transmitted across a link every second - and am using on-off modulation (say +5V = '1' bit and 0V = '0' bit), which by default is NRZ (non return to zero). What is the required frequency: 10 MHz, or could we say that each cycle can transmit 2 bits - so a frequency of 5 MHz would do?

Also, to then improve this could I use a different modulation scheme or form of coding? I've read that QPSK or QAM are more 'spectrally efficient' and can transmit more bits per symbol, but isn't the frequency determined by the BIT rate not SYMBOL rate?

Thanks for any tips!

 

[phinds]

something to think about: if you use a simple 5mhz or 10mhz "square-type" wave, that is, one where the signal pops up to 5v when there is a 1 and stays at 0v when there is a 0, what does your detector do when you transmit 10,000 successive 0's or 10,000 successive 1's ?

 

[Turksen]

I appreciate, with no transitions, that the clock signal can't be regenerated & things aren't pleasant.

So say we have scrambled data, then what frequency would be sensible? Could you say that since a square wave is composed of many frequencies, we'll just transmit the fundamental (10 * 0.5 I think)? Or fundamental + first harmonic?

Thanks

 

[phinds]

I appreciate, with no transitions, that the clock signal can't be regenerated & things aren't pleasant.

Well, yes, but that actually is NOT necessarily a problem. You could, for example, have a space-based transmission system that was slaved to a very precise 10Mhz (to use the example number from your post) frequency. Then the ground based system has only to lock onto where one edge is and phase-shift the signal accordingly and then it can easily detect 10,000 consecutive 1's or 0's because it knows exactly where they are and can then just look in the middle of where they are to see are they 1 or 0.

My point is that signal transmission is a highly complex topic and different coding techniques give different degrees of reliability, often traded off against bandwidth, so I'm not trying to NOT answer your question, I'm trying to get you to THINK about different ways a signal processing waveform might be used to transmit information. That is lots more interesting than just getting an answer of XXXhz.