the frequency band defined for GSM is 890-915 Mhz & 935-960 Mhz (so 25MHz of bandwidth in UL and 25 in DL). It means that a GSM communication, by definition, should use a frequency within this range.
One GSM communication uses 200kHz in UL and 200kHz in downlink (and actually, this little slice of frequencies can be used by 8 communications at the same time !!). Therefore a GSM channel = 200kHz.
Who is in charge of these 2x25MHz ? It is the government.
Who wants to use them ? The operators.
So the government must share this band among different operators. One operator can receive for instance 2x8MHz of the GSM900 band.
These 8MHz = 8000/200 = 40 GSM channels.
I guess you know about frequency reuse ? A site can have 3 cells, using 4 TRX each. Each TRX must use one channel (it is a “non hopping” example). So this site uses 12 channels.
How many sites can you plan with 40 channels ? Not much… what happen is that the channels will be reused, far enough as to avoid interference with the other sites using the same channels.
But you understand the concept : with only 40 channels, it is very complicated to put site 3×4. You are limited in terms of capacity, in order to keep a “good” frequency reuse.
But the problem is gone if you receive more bandwidth 🙂 That’s what all the fuss is about.