Paper 2020/326

An upper bound on the decryption failure rate of static-key NewHope

John M. Schanck

Abstract

We give a new proof that the decryption failure rate of NewHope512 is at most $2^{-398.8}$. As in previous work, this failure rate is with respect to random, honestly generated, secret key and ciphertext pairs. However, our technique can also be applied to a fixed secret key. We demonstrate our technique on some subsets of the NewHope1024 key space, and we identify a large subset of NewHope1024 keys with failure rates of no more than $2^{-439.5}$.