@nokiabot, to put it simply:

DL = downlink
UL = uplink
DL band = range of frequencies that the base station uses to transmit to the handsets
UL band = range of frequencies that the handsets use to transmit to the base station

(See also http://en.wikipedia.org/wiki/GSM_frequency_bands.)

In other words, your phone transmits at one frequency band and receives at another. To make a peer-to-peer link, one of the phones would have to swap bands which, as juiceme says, is impossible.

VoIP over an ad-hoc WLAN would work but is quite limited in range. Mind you, so would be GSM. The base stations ("towers") are at elevated positions and have much higher power output (and possibly better antennas allowing a more sensitive input) than the handsets. A handset can talk to a tower 20-30 km away in an open terrain, but two handsets at ground level would struggle to achieve more than 1-2 km [citation needed]. Granted, better than WLAN but perhaps less than what you'd expect.