This paper investigates the integration of fluid an- tennas (FAs) into uplink non-orthogonal multiple access networks suffering from imperfect successive interference cancellation (SIC). The dynamic reconfigurability of FAs offers significant potential for mitigating interference and enhancing network per- formance by adapting antenna positions in response to changing channel conditions. In this study, we propose a joint optimization framework to maximize the system’s sum rate by optimizing key parameters, including FA positions, beamforming vector at the base station, and transmit power allocation for each user. The problem is formulated as a non-convex optimization task and solved using a new deep reinforcement learning (DRL)- based framework. The proposed DRL model incorporates a structured exploration strategy and reward shaping to efficiently learn optimal policies for resource allocation and antenna posi- tioning in dynamic environments. Extensive simulations validate the effectiveness of the proposed approach, demonstrating that integrating FAs significantly improves the sum rate, particularly in scenarios with imperfect SIC

Keywords