This may look good enough at first glance but it is not optimal.
For the given decode success rate and maximum number of symmetric differences, there are optimal k (number of hash functions) and h (hedge factor).
For the maximum number of symmetric differences up to 1000 and the decode success rate above 99.5%, optimal k and h can be found here.
Space-optimal k and h are obtained via a brute force search. Apparently, there is no analytical result that can meet the strict practical requirements.
This is how the parameters are used in IBLT construction.
This may look good enough at first glance but it is not optimal.
For the given decode success rate and maximum number of symmetric differences, there are optimal
k
(number of hash functions) andh
(hedge factor).For the maximum number of symmetric differences up to 1000 and the decode success rate above 99.5%, optimal
k
andh
can be found here. Space-optimalk
andh
are obtained via a brute force search. Apparently, there is no analytical result that can meet the strict practical requirements.This is how the parameters are used in IBLT construction.