Wireless channel key generation is a crucial method for achieving physical layer security. An in-depth study of the channel reciprocity is the theoretical foundation for this technology. Existing channel modeling methods struggle to meet all the requirements of 5G applications and lack comprehensive analysis of 5G uplink-downlink reciprocity. By constructing a map-based hybrid channel model that combines the geometry-based stochastic channel model and ray-tracing techniques, accurate qualitative modeling is achieved. This paper further analyzes the impact of configuration parameters such as subcarrier spacing, channel bandwidth, user equipment mobility velocity, and DMRS configuration on channel reciprocity. Experimental results indicate that while increased subcarrier spacing and reduced symbol intervals boost channel reciprocity, higher device mobility velocity diminishes it due to doppler effects and other factors. Additionally, for cases where the Pearson correlation coefficient between uplink and downlink channels exceeds 0.7, we further confirm that the system’s security is not vulnerable to eavesdropping attacks.