It has been shown that in the asymptotic case of infinite-key length the 2-decoy state QKD protocol outperforms the 1-decoy state protocol. Here, we present a finite-key analysis of the 1-decoy method. Interestingly, we find that for practical block sizes of up to $10^8$ bits, the 1-decoy protocol achieves for almost all experimental settings higher secret key rates than the 2-decoy protocol. Since using only one decoy is also easier to implement, we conclude that it is the best choice for practical QKD.
Download