goodput

In conclusion, the tendency in WRNs would be to use large buffer sizes independently of the traffic load experienced.
In [7], TCP/NC was tested on a TCP-Reno flow running over a single-hop wireless link. The experiment is performed over 802.11a with a bit rate of 6Mb/s and a maximum of 5 link layer retransmission attempts. The quantity measured during the experiment is the goodput, we can define a goodput as the application level throughput, i.e. the number of useful information bits, delivered by the network to a certain destination, per unit of time. The amount of data considered excludes protocol overhead bits as well as retransmitted data packets. This is related to the amount of  time from the first bit of the first packet is sent (or delivered) until the last bit of the last packet is delivered.
The principal observation in the results was that in the lossless case, TCP performs better than TCP/NC. This could be because of the computational overhead that is introduced by coding and decoding operations, and also the coding header overhead. However, as the loss rate increases, the benefits of coding begin to outweigh the overhead. The goodput of TCP/NC is therefore higher than TCP, and hence the fall in goodput is more gradual with coding than without, [7].