You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
One additional question:
Small updates to support normalized minsum algorithm, i.e., msg = sign_val * msg_e * 0.8 in _cn_update_minsum_mapfn function of class LDPCBPDecoder, which may make the ber performance slightly better.
I’m not exactly sure what you mean by “ARQ support”, but I assume it relates to the incremental redundancy (IR) of LDPC codes. Supporting IR is currently not our highest priority, but could be definitively integrated in a future release.
Regarding your question about minsum decoding: it is already possible to scale the VN2CN messages with trainable weights.If initialized with a constant value, these weights allow scaled min-sum decoding (see Weighted BP example).
Remark: Closing this thread as this is not really an issue. Please consider starting a new discussion in the forum for future feature requests.
Additional context
Excellent project!
Do you plan to support ARQ in future releases?
The text was updated successfully, but these errors were encountered: