To size the ring the simplest way is to calculate an expected traffic load from the GSM and CDMA networks. Both GSM and CDMA networks provide data and voice services so you will need to calculate these demands.
Based on the peak demands for both voice and data you will then need to provide at a minimum an equivelant capacity in the optical backbone.
For a two node SDH ring (based on you description this is what you seem to say) it is pretty straight forward. If you are handling at peak level 1900 calls and 4 Mb/s data, then you only really need an STM-1 (144 Mb/s) ring or point to point connection.
1900 calls => 1900/32 ~= 60 E1s
4 Mb/s ~= 2 E1s
63 E1s = 1 STM-1
This would not allow for growth.
The two main network protection schemes are SNCP/UPSR (ITU/ANSI) and MS-SPRING/BLSR. In over simplified situations is best for rings where traffic is primarily between two node in the ring. MS-SPRING is better for dispersed traffic across all nodes in the ring. There are operational and service implications of using one protection mechanism over the other. Ask the operations team.