This looks like an interesting paper. I only partly get the paper but it reads like it might be usefull for inspiration on building non-interactive transfers.
It’s more useful for accelerating the Initial Bytes Download by using Incrementally Verified Computation to verify existence of a valid kernel history, without having to download all those kernels. I.e. it would make IBD just UTXO sized.
A more recent paper that might offer further improvements is