coverage
interference
path unbalance (poor tx or poor rx in the bts, usually, it doesn’t generate alarms)
in some vendors, the sdcch drop also contains the sdcch congestion, so check that out also
and finally, you can check the SDCCH average duration. If it’s higher than 6 or 7 seconds on most cells of your network, there might be an issue during the authentication phase (dialog with HLR)
SD drop depends on formula used. considering phantom raches in this formula is a little bit weird. To Drop on SD you should have reached this SD. so , from my point of view (and my customer point of view 🙂 ) SD_drops are considered after SABM is received from MS. in that case Phantom raches do not impact sd_drop.
On the other and you have to defined an other KPI to check SD seizure