Tight scaling of key rate for differential-phase-shift quantum key distribution

Abstract

The performance of quantum key distribution (QKD) protocols is evaluated based on the ease of implementation and key generation rate. Among major protocols, the differential-phase-shift (DPS) protocol has the advantage of simple implementation using a train of coherent pulses and a passive detection unit. Unfortunately, however, its key rate is known to be at least proportional to $\eta^2$ with respect to channel transmission $\eta\to0$. If one can only prove the rate proportional to $\eta^2$ and cannot improve the analysis beyond that, then the DPS protocol will be deemed inferior to other major protocols, such as the decoy BB84 protocol. In this paper, we consider a type of DPS protocol in which the phase of each emitted block comprising $n$ pulses is randomized and significantly improve the analysis of its key rate. Specifically, we reveal that the key rate is proportional to $\eta^{1+\frac{1}{n-2}}$ and this rate is tight. This implies that the DPS protocol can achieve a key rate proportional to $\eta$ for a large number of $n$, which is the same scaling as the decoy BB84 protocol. Our result suggests that the DPS protocol can achieve a combination of both advantages of ease of implementation and a high key generation rate.