Currently there is no option for selecting UTF-16 for Oracle connection.
Feature request is raised for the same: CLDDI-2935
However UTF-8 is expected to processes the data as expected.
Report if any discrepancies seen.
one follow-up question to your explanation:
In PowerCenter you cannot process characters with a code value beyond the so-called BMP (Basic Multilingual Plane, the first 64 K of Unicode), meaning no character with a code > U+0ffff can be processed.
Does the same apply to ICS / IICS? If so, then setting the connection to UTF-8 may not be sufficient; if the source data contain any characters with character codes outside the BMP, then IICS could not process them, meaning they would get lost in processing.
Yeah UTF-16 is there for future road map.
So as of now you might face issues.