Reliable communication between ground control stations and unmanned aerial vehicles (UAVs) is fundamental for safe, autonomous, and coordinated flight operations. This work presents a decentralized model predictive control (MPC) framework that jointly optimizes UAV trajectory planning and high-frequency MIMO communication performance, explicitly accounting for multipath channel effects. The proposed framework incorporates ergodic spectral efficiency, derived from realistic location-dependent channel modeling, into the control cost function, thereby enabling UAV trajectories to dynamically balance three key objectives: (i) minimize travel time to assigned targets, (ii) maintain probabilistic safety through minimum separation constraints under stochastic disturbances, and (iii) maximize the achievable uplink ergodic rate to the ground station. Simulation results confirm the effectiveness of the approach, showing that ergodic rate-aware path planning leads to improved trade-offs between communication quality, safety, and mission efficiency under practical channel conditions.