Towards a self-tuning quantum key distribution transmitter using a genetic algorithm

For the adoption of QKD to grow, much effort has been devoted to making QKD systems more robust and efficient. Much of the complexity of a QKD system stems from its transmitter where quantum states encoded with bit values are prepared. Recently, optical injection locking (OIL) has emerged as a promising method to realize high-speed QKD transmitters with a compact design. This approach enables direct phase encoding without the need for external modulators, while simultaneously improving the laser characteristics. Due to these remarkable advantages, OIL has been widely applied to many QKD protocols, including BB84, MDI-QKD and TF-QKD. However, in practice, tuning the laser system to find optimal operating parameters is a very challenging task. This is because the underlying laser dynamics are rich and involve a complex interplay between multiple control parameters. It is therefore highly desirable to develop an efficient method to optimize the systems. Here, for the first time, we address this issue by demonstrating a self- tuning QKD transmitter by implementing a genetic algorithm to autonomously locate the optimum system parameters. Without any user intervention, our approach manages to optimize the quantum bit error rate down to similar to 2.5%, matching the state-of-the-art performance.