For example, Is the TCP throughput observed for a SNR value in free space largely different from that observed for the same SNR in small offices inside a building?
If the throughput at physical layer is same under different environments then the TCP throughput must not change. Throughput of IEEE802.11 in indoor is different from that in out-door environment, which is because of change is RSS. IEEE802.11 WLAN was designed for indoor environment and its throughput is higher in indoor because signal bounces from the walls and remains contained inside; such an effect can't be obtained in open environment.
When you are considering throughput in open space versus inside building, make sure the transmitter is same in both cases, i.e compare apple to apple comparison. You can't compare SNR of two situation from two different transmitters.
In case of IEEE802.16 wireless networks, the network was designed for open environment through BTS. You get maximum throughput in open environment but it decreases inside the building due to wall attenuations and other losses.
Summarizing, if the throughout at RF layer is same then it shouldn't change at TCP layer.
Aftab is right. However, i want to add a point which is important for the answer of the question.In free space you have a strong line of site signal and weak or null multiple path reflection signals. The channel model is then a Rician one. In case of propagation inside building there will be mulipath channel because of strong multiple reflections from the walls of the buildings, the channel model will be rayleigh fading channel. The rician channel has higher throughput than that of rrayleigh channel for the sane SNR becaue the presence of strong line of sight signal.
Thanks Aftab and Abdelhalim for your answers. It has given me some explanation on our field measurement observations. We observed that the throughput observed for Free space and Indpoor environments were comparable when signal was strong (SNR>30dB). However when signal becomes grey and week (SNR