Total: 1
In distributed computing systems, reducing the communication load during the data shuffling phase is a critical challenge, as excessive inter-node transmissions are a major performance bottleneck. One promising approach to alleviate this burden is Embedded Index Coding (EIC), which exploits cached data at user nodes to encode transmissions more efficiently. However, most prior work on EIC has focused on minimizing code length in wired, error-free environments-an objective often suboptimal for wireless multiple-input multiple-output (MIMO) systems, where channel conditions and spatial multiplexing gains must be considered. This paper investigates the joint design of EIC and transmit beamforming in MIMO systems to minimize total transmission time, an NP-hard problem. We first present a conventional optimization method that determines the optimal EIC via exhaustive search. To address its prohibitive complexity and adapt to dynamic wireless environments, we propose a novel, low-complexity multi-agent reinforcement learning (MARL) framework. The proposed framework enables decentralized agents to act on local observations while effectively managing the hybrid action space of discrete EIC selection and continuous beamforming design. Simulation results demonstrate that the proposed MARL-based approach achieves near-optimal performance with significantly reduced complexity, underscoring its effectiveness and practicality for real-world wireless systems.