"All models are wrong, but some are useful" - George Box, statistician
Your goal shouldn't be to exactly follow someone else's model or work but I think you'll be better off trying to do the following.
You learn about the background theory of the topic, then you understand how modern research is being done in that area, and then when you lay out your research problem the model to usually falls into place. Or it'll become clear if no past works directly relate and then you need to lay out assumptions and why you are making those assumptions so that others can understand and trust the validity of your modeling.
As far as you working on OFDM specifically, there are many papers on OFDM as well as material on the theory of it. Once you learned about OFDM (why use it? how are synchronization, channel estimation, equalization typically done? what assumptions do people usually make?), then you may explore how do modern research works explore OFDM. I think you'll find either people exactly use a published signal (IEEE 802 for WiFi, LTE/5G for cellular, etc.) or they closely model one so that its not a far stretch. This is to say, don't have your work revolve around something too obscure without justification.