This paper investigates a two-timescale uplink transmission framework for a fluid antenna-enabled multiuser multi-input multi-output system (MIMO-FAS). Antenna positions are optimized based on statistical channel state information (CSI), while beamforming vectors at the base station (BS) adapt to instantaneous CSI. Under a Rician fading channel with imperfect CSI, we establish a linear minimum mean square error (LMMSE)-based channel estimation approach and derive a closed-form expression for the achievable uplink rate using a low-complexity maximal-ratio-combining (MRC) detector. The optimization problem is formulated as a minimum user rate maximization problem by optimizing the fluid antenna positions, subject to the feasible region and the minimum spacing distance constraints. To address this non-convex problem, a genetic algorithm (GA) method is proposed, encoding antenna configurations as population individuals. Additionally, an accelerated gradient ascent algorithm is proposed to enhance computational efficiency. Numerical results validate the mathematical derivations and demonstrate that the proposed two-timescale transmission strategy significantly outperforms traditional FPA systems, with both algorithms achieving enhanced gains.

This paper proposes an uplink two-timescale transmission scheme for a fluid antenna-enhanced multi-user multiinput multi-output system (MU-MIMO-FAS), where antenna positions are optimized based on statistical channel state information (CSI), and beamforming at the base station (BS) adapts to rapidly-varying instantaneous CSI. Using a Rician channel model with imperfect CSI, we employ the linear minimum mean square error (LMMSE) method for channel estimation and a maximal ratio combining (MRC) detector to derive a closed-form expression for the achievable rate. Subsequently, we formulate a minimum user rate maximization problem for antenna position design, subject to movement and spacing constraints, and utilize a genetic algorithm (GA) to solve this non-convex problem. Numerical results demonstrate that the proposed two-timescale MU-MIMO-FAS design significantly outperforms the traditional fixed-position antenna (FPA) system.