MATLAB: 5G New Radio Polar Coding: Why is the rate defined as K/E

MATLAB

In the 5G New Radio Polar Coding example, why is the rate defined as K/E?
<https://www.mathworks.com/help/releases/R2019b/5g/gs/polar-coding.html>
That is, K includes both information bits and CRC bits (which are redundant). Shouldn't the rate be (K-crcLen)/E?

Best Answer

From an informational theory perspective - the information carrying bits are K-crcLen only.
However, for common comparisons across 3GPP member contributions, the rate is considered from the perspective of the component. In this case that would be the encoder + rate matcher. K bits are input and E bits are output from the combined Polar encoding and Rate matching components. This falls within the simulation assumptions agreed to within 3GPP, as the specifications were being developed prior to their release.
We have thus chosen to follow the simulation assumptions so as to make it easier for users to compare our results, with other published ones.
This is similarly handled at the receiver as well, where K bits are output from the decoder. We compare the (K-crcLen) bits at the end to exclude the CRC, and compare only the informational bits, in determining the bit/block errors.
Related Question