My question is very primitive :) I've read multiple articles today and answers here, on Stack Exchange, but still don't understand one thing. Why doesn't frequency itself affect data rate in mobile networks? 3G/4G networks use QAM modulation, that includes changes to amplitude and phase of signal. Let's say, we have 900 MHz signal with 10 MHz bandwidth and 2600 MHz signal with 10 MHz bandwidth. At the same "snippet" of incoming signal we'll have much more "cycles" to modulate on higher frequency (per same time), don't we? So why doesn't it matter?
It's not a bad question and shows quite a common misunderstanding of how Radio Frequency (RF) Systems work. The 900MHz and 2600MHz signals are called Carrier Frequencies. The actual information is contained in the 10MHz bandwidth. The original signal is a baseband signal that extends up to 10MHz. This is used to modulate the carrier signal. The reason we do this is so that we can have several channels sharing the same medium.
When received by a radio, the signal is downconverted back to the baseband (up to 10MHz in this case). The reason we do this, instead of sampling the signal directly, is the RF electronics are very complicated and relatively expensive, while baseband electronics is not.
So to answer your question, when both signals are downconverted, they are both 10MHz wide signals, so will transfer data at that rate (probably greater for QAM due the symbol rate but that's another story).
Because the bandwidth of the channel determines how rapidly the symbols on that channel can change, not the carrier frequency. It is the bandwidth of the channel which determines how fast the channel "rings down" from a sudden change in phase or amplitude.
So your two 10MHz bandwidth channels can each sustain a symbol rate of no more than \$10 \cdot10^6\$ symbols per second, regardless of whether the carrier frequency is 0Hz or 1THz or anywhere in between.
By quick analogy (the other answers are sufficient in their detail):
If you own a hotel where all the rooms are the same size (bandwidth), it doesn't matter how high a floor (frequency) they're on--they all hold the same amount of stuff. If you have some larger rooms (more bandwidth), they'll hold more stuff, whether they're on the 2nd floor or the 26th.
Consider Morse Code modulation applied to human audio as one of many possible examples.
- . . . - . . . - . - .
First, 880 Hz carrier frequency:
biyip bi bi bi biyip bi bi bi biyip bi biyip bi
Next, 55 Hz carrier frequency:
gruur gr gr gr gruur gr gr gr gruur gr gruur gr
10 MHz of bandwidth contains the same amount of information, regardless of whether it's centered at 900MHz or 2600MHz, or any other center frequency.
This is easy to show -- the center frequency of a signal can be shifted digitally or electronically without destroying it. You can move the 2600MHz signal down to 900MHz, and it will have the same bandwidth. Then you can move it back to 2600MHz and get exactly the original signal back, so obviously whatever information you can carry at one frequency, you can carry at the other.
By the the Nyquist-Shannon sampling theorem, we know that any < 10MHz signal at any center frequency can be reconstructed from samples taken at a 20MHz rate.
If one is using a single carrier to send information through an unshared, interference-free, communications medium, bandwidth will generally scale with carrier frequency. If, however, one modulates a signal with a 901,500,000hz center frequency and sends it through a communications medium which, while free of unwanted frequency content in the range 901,490,000Hz to 901,510,000Hz, may have unpredictable amounts of frequency content below 901,490,000Hz and above 901,510,000Hz, then even if one demodulates the signal perfectly, the demodulated signal may have unpredictable amounts of noise at any or all frequencies above 10,000Hz(*).
If one used a perfect filter to remove from the demodulated signal all content above 9,999Hz, one could faithfully recover all frequency content below that frequency, but would of course lose any content one had transmitted at frequencies higher than that. If one doesn't filter out everything above 10,000Hz, it might totally drown out everything else in one's signal. The usable frequency range will depend upon the distance between the carrier frequency and the nearest unwanted frequency content.
(*) Techniques called single sideband modulation or quadrature amplitude modulation may be used, respectively, to shield a channel from interference on one side (allowing one to use a carrier frequency which is off center toward that side, and thus further from the other side), or use a channel that is clear on both sides of the carrier frequency to send two signals, each of whose bandwidth would match the difference between the carrier and the nearest interference. The general problem of interference, however, remains.
First, all communication channels are noisy.
The maximum data rate depends on the signal/noise ratio and the bandwith. The particular frequencies of the band are of no importance.
Signal level depends on the transmitter power and the propagation conditions. Noise depends on the temperature and the presence of other transmitters.
Propagation conditions and thermal noise both DO depend on the frequency in more or less predictable manner, as well as the presence of other transmitters, but all of these factors have only a second-order relation to your question. But yes, they are related.
The bandwidth is allocated by the regulator authority (when using a common media), some standard, media properties or other considerations when you control the media.
A higher bandwidth dictates higher data rate which we all know. Bandwidth refers to a frequency range and not just a single frequency. If your wireless system is operating say at 5.85 GHz to 5.95 GHz, a bandwidth of 100 MHz is available. Also, when your wireless operating system is operating at say 2.4 GHz to 2.45 GHz, the bandwidth available is 50 MHz. Thus, data rate is higher in the former case.
I mentioned the former band as ranging from 5.85 GHz to 5.95 GHz. Bandwidth is the difference between the highest frequency of the band and the lowest frequency of the band. If regulatory authorities allow us to expand the existing bandwidth to say 150 MHz, then our highest band frequency is now 6 GHz. Thus, bandwidth has risen and hence the data rate has also risen.
Whenever the highest frequency of the band is increased, the bandwidth increases.