FEC calculation logic - More details on how it works

[shrish MV]

Hi All,

I have been trying to understand the code around the file webrtc/modules/video_coding/media_opt_utils.cpp, to understand how the FEC rate is calculated give the different parameters like RTT, loss, bitrate , resolution , framerate etc. I am able to understand a few things , 

- the FEC rate tables have been calculated by some empirical experiment for 576x704 resolution.

- for other resolutions, we try to normalize to the above resolution and calculate based on loss

- boost protection for key frames

- convert the fecrate from 0-128 to 0-255 range exponentially

There are a few more details and reasoning i would like to understand.

- i dont see any unittest for this class ?

- in the code , we calculate the correction factor for lower rates , but it is never used anywhere ?

- the code for calculating "resolnFac" and "effRateFecTable" and "rateIndexTable" , i would like to understand the reasoning behind the code ?

- the kFecRateTable in fec_rate_table.h , how are these values calculated ?

If someone could explain me the above or point to a document regarding the above questions, that would be really helpful.

Thanks in advance.

Regards,
Shrish

[yishen wang]

I would also like to understand  resolnFac