Feedback experiments using entangled photons for polarization control in future quantum networks

Abstract

Control of the measurement frames that project on polarization entangled photons is an important experimental task for near term fiber-based quantum networks. Because of the changing birefringence in optical fiber arising from temperature fluctuations or external vibrations, the polarization projection direction at the end of a fiber channel is unpredictable and varies with time. This polarization drift can cause errors in quantum information protocols, like quantum key distribution, that rely on the alignment of measurement bases between users sharing a quantum state. Polarization control within fiber is typically accomplished using feedback measurements from classical power alignment signals, multiplexed in time or wavelength with the quantum signal that coexist in the same fiber. This thesis explores ways to use only measurements on the entangled photons for polarization control and perform entanglement measures without multiplexing alignment signals. This approach is experimentally less complex and can reduce the noise within the quantum channel arising from the alignment signals. In the first part of this dissertation, we study how to use distributed measurements on polarization entangled photons for polarization drift correction in a 7.1 km deployed fiber between the University of Maryland and the Laboratory of Telecommunication Sciences for two individuals sharing a near maximally entangled Bell state, $\hat \rho = |\Psi^-\rangle\langle\Psi^-|$. In the second part of the dissertation, we examine how to use feedback measurements to maximize the violation of a Bell's inequality used as an entanglement measure. Both polarization drift correction and the maximization of a Bell's inequality violation use iterative optimization algorithms to actuate upstream polarization controllers. In the Bell's inequality investigation, three numerical methods: Bayesian optimization, Nelder-Mead simplex optimization, and stochastic gradient descent are implemented and compared against each other. For complete polarization control and Bell's inequality violation experiments, we developed a polarization and time multiplexed detection system that reduced the number of photon detectors needed and mitigated the demand on the coincidence counting electronics for real-time feedback and control.