I’ve optimized such networks, with dualband cells. The “logical” principle is :
1800 cells are considered “inner zone”
900 cells are considered “outer zone”
Are those the same thing than what you’re referring too ?
Since you’ve explained about the physical part, I’ll tackle the logical part.
Those cells are basically concentric cells, and the handovers between the 2 zones are seamless (but the HO thresholds are a bit tricky to find). The GSM1800 zone has naturally a smaller coverage than the GSM900 zone, because of radio propagation properties of those two different frequency bands. This is the reason why the 1800 zone is the “inner” zone.
The BCCH and SDCCH are always located in the outer zone (gsm 900), on top of TCH. The inner zone contains only TCH. Therefore a call can be setup only in the GSM900 zone.
HO from outer to inner is based on RX_LEV of the outer zone, it is a “better condition” HO. The level of the outer zone is too high, so the MS is handovered to the inner zone, for capacity purpose.
HO from inner to outer is based on RX_LEV of the inner zone, and it is an emergency HO (too low RxLev in inner zone, so the MS has to jump back to the outer zone to receive a good rx lev).
A MS in the inner cell will preferably make a HO to the outer zone, rather than to a neighbor cell, even if there is such a possibility.