I noticed that countries with latin script use latin-1 encoding for sms, because they never really needed unicode. Then when software converts text to latin-1 or acsii, there's an option to find the best match character in ascii repertoire, I think in that case ı will be converted to i.
It's not latin-1, and it's not ASCII either. It's 7-bit GSM 03.38 charset, with optional shift tables. You can use either that or UCS-2 (though these days phones use UTF-16 instead), and UCS-2/UTF-16 significantly limits the number of characters that fit into the message (160 vs. 70).
