Hi,
probably I will not be able to answer your questions completely, but I hope I will give you some hints. This is because you are describing your system in a general way, but you questions are quite particular - but please don't take it as an offence.
1. It depends how you implement your system, in particular how you allocate resources to different users. There are two ways how to do that - allocate 50 Mbps OFDM channel for a single user at a time, or allocate same channel to several users at a time (some of the sub-carriers would be allocated to one user and some other sub-carriers - to another, etc)
In case your system is the first one and your choice is pure 50 Mbps then data rate is independent on input data bits. It should be rather dependent on channel conditions.
2. In order to assess BER of the system, you have to take A LOT of samples. The more samples you take - the higher accuracy. Resolution is one bit, so if you transfer only 2048 bits, then the maximum achievable accuracy is:
1/2048 = 0.5E-3
I assume that your result is very consistent. Furthermore transfer of 2048 bits over 50Mbps channel takes tiny amount of time - the rf channel could be very different next second, but the simulations ignore this fact.
3. It depend on what you mean by modulation technique. If it is QPSK or 16QAM - then it is not possible to say what is data rate. It is possible to state that QPSK carries 2 bits per symbol and 16QAM - 4 bits per symbol. In order to know data rate, it essential to know symbol rate.
Say you have 50 000 symbols per second and your modulation is QPSK - then data rate is 100Kbps. However if FER (forward error correction) is employed in the system, then data rate is much lower, as some of the bits are used for error correction purposes.
Systems are designed to have certain Data rates and simulation only tries to imitate that system. If you are building simulation of a system, then you have to know the data rate of that particular system. It is not the other way round.
I hope I haven't confused you even further...