Public key cryptography is a cornerstone of today’s private communication. As quantum computing is a
major threat for the privacy of public key cryptography, the development of appropriate alternatives is
of critical importance. One promising alternative to achieve security against quantum computer attacks
is symmetric encryption, whereby the secret key is established using quantum key distribution (QKD).
Especially interesting for a wide application are continuous-variable QKD (CV-QKD) systems, as these can
potentially be implemented using standard components of coherent optical communications. However,
a major challenge for implementing CV-QKD is to achieve phase noise mitigation of the optical quantum
signal with a received power of less than 1 photon per symbol. In this thesis, two different approaches
for implementing CV-QKD systems and to overcome this challenge are proposed and investigated. To
achieve phase noise mitigation, a key element of both systems is the use of Bayesian inference methods
in the digital signal processing. Both systems rely on standard components and are therefore promising
for a practical implementation. Based on the experimental results, the suitability for CV-QKD is
confirmed, limiting factors are identified and the achievable performance is discussed. Additionally, the
compatibility of CV-QKD with existing optical networks is investigated based on system experiments with
a CV-QKD channel embedded in a commercial C band WDM system.