Jennifer Bartlett

Continuous-Variable Quantum Key Distribution (CV-QKD) uses continuous variables of the electromagnetic field, such as amplitude and phase, to transmit quantum information between two users, Alice and Bob. In this regime, there are two standard practical implementations: the transmitted local oscillator (TLO) and the local-local oscillator (LLO), where the LLO has recently reached 100km in laboratory-based fibre. Despite this success, the LLO also suffers major disadvantages like reduced quantum signal bandwidth capacity. Typically, only 50% of the transmission is for exchanging common phase references between Alice and Bob using reference signals. Subsequently, the key rate is reduced by half over all the transmission distances. In this work, we propose using a machine learning algorithm such as Long Short-Term Memory(LSTM) to predict the reference phase. The algorithm is trained with a sufficient number of phase reference data. Then, it predicts the phase drift with minimal usage of the reference signals, thereby reducing the reference signal bandwidth and increasing the key rate.