Practical Phase-Modulation Stabilization in Quantum Key Distribution via Machine Learning

Abstract

In practical implementation of quantum key distributions (QKD), it requires efficient, realtime feedback control to maintain system stability when facing disturbance from either external environment or imperfect internal components. Usually, a "scanning-and-transmitting" program is adopted to compensate physical parameter variations of devices, which can provide accurate compensation but may cost plenty of time in stopping and calibrating processes, resulting in reduced efficiency in key transmission. Here we for the first propose to employ a well known machine learning model, i.e., the Long Short-Term Memory Network (LSTM), to predict those physical parameter variations in advance and actively perform real-time control on corresponding QKD devices. Experimentally, we take the phase-coding scheme as an example and run the LSTM model based QKD system for more than 10 days. Experimental results show that we can keep the same level of quantum-bit error rate as the traditional "scanning-and-transmitting" program by employing our new machine learning method, but dramatically reducing the scanning time and resulting in significantly enhanced key transmission efficiency. Furthermore, our present machine learning model should also be applicable to any other QKD systems using any coding scheme or QKD protocols, and thus seems a very promising candidate in large-scale application of quantum communication network in the near future. * Quantum key distributions (QKD) [1][2][3][4] can provide information-theoretic secure keys between two communicated parties, usually called Alice and Bob, and has attracted extensive attention from the scientific world since the first BB84 protocol was proposed [2].To date, significant progresses have been achieved in this field both theoretically and experimentally, making it moving from laboratory research to practical implementation and from point-topoint communication to multiuser complex networks [5,6].In practical implementation of QKD, in order to maintain the stability and reliability of the system, especially for fast-speed QKD networks, real-time control are highly demanded due to very complex realistic environment and imperfect internal devices. In the history, the Faraday-Michelson Interferometer (FMI) was designed for phase encoding QKD systems to solve the problem of interference instability existing in Mach-Zehnder Interferometers (MZI) [7] caused by polarization changes in transmission fibers. Both theory and experiment have proven the polarization stability of FMI based QKD systems [8]. Nevertheless, neither MZI nor FMI can maintain phase stability for a long time due to ineluctable internal phase shift. Therefore, it is quite common to use a "scanning-and-transmitting" program to calibrate those physical parameters in present QKD systems [9][10][11][12]. Although present calibration programs can keep the stability of QKD systems, it is at the cost of transmission efficiency, since it takes time to run the calibration program and no signal data can be transmitted during the ca...

Machine Learning-Assisted Measurement Device-Independent Quantum Key Distribution on Reference Frame Calibration

“…The data structure is illustrated in Figure 1. A newly added data feature, i.e., partially disclosed QBER of XX basis, can provide the LSTM network valuable running status, which directly increase prediction accuracy compared to the former data features [23]. The training data conclude 25 batches, and each batch consists of 5400 data points.…”

“…The final MSEs of training set, testing set, and validation set are 0.0533, 0.1131, and 0.881 respectively. Additionally, an updating process is added periodically for the long-term reliability of network forecasting, in which the scanning program will be operated after predicting for a certain time to eliminate the cumulative error in the prediction period [23]. As a result, the duty ratio of the present MDI-QKD system has been increased from 85.7% (transmitting: 30 s, scanning: 5 s) to 96.9% (transmitting: 540 s, mismatch events: 2 s, updating: 15 s).…”

“…In order to overcome those shortcomings, we adopt long short-term memory network (LSTM) [21,22] onto the MDI-QKD system and predict out the phase drift between two users in advance. Immediately, real-time phase compensations can be realized, drastically increasing the key transmission efficiency [23].…”

280-km experimental demonstration of a quantum digital signature with one decoy state

“…The resulting interference visibility is about 99.7% and the duty cycle of transmission approximtes 86%. In fact, we can also use an active phase compensation scheme [33] to further improve the system efficiency.…”

Real-Time Phase Tracking Scheme With Mismatched-Basis Data for Phase-Coding Quantum Key Distribution

“…Therefore, the scanning process will occupy a proportion of the system time, and decrease the duty cycle [19] of the transmission process, resulting in a reduction of transmission efficiency and consequently the secure key generation rate of the QKD system. Most recently, an active phase compensation scheme based on the machine learning algorithm is proposed [21]. This scheme enables QKD systems to predict the parameter variations beforehand and actively perform real-time control on corresponding devices to perform the phase tracking, increasing the efficiency of the QKD system.…”