The constant-redundancy FER conjecture for convolutional code–CRC pairs

Consider a minimal convolutional code with $m$ memory elements and a degree-$v$ distance-spectrum-optimal CRC code, used together under serial list Viterbi decoding at a fixed SNR. The constant-redundancy FER conjecture. Any such convolutional code–CRC pair has the same frame-error-rate performance as any other pair satisfying the same value of $m+v$.

The claim concerns pairs with the same total number of redundant bits, $m+v$. The paper presents it as an empirical conjecture based on pairs having nearly the same SNR gap to the random-coding union bound, while their decoding complexities can differ.