in GSM, during a voice call, the MS is allocated on one circuit only. There is no risk of confusion between several MS, because one circuit = one call.
A circuit in GSM is a radio timeslot, which is identified by its timeslot number and by its ARFCN.
In EGPRS however, one radio timeslot can carry several users (up to 6, usually) and as you noticed, such a system requires a need for each “block” sent on a RTS to be specifically identified.
Back to voice call, in half rate, two users are sharing one timeslot. But each user is actually using exactly one timeslot every other two. The scheduling is static, which prevents possible confusion.
In EGPRS the scheduling is dynamic.