In unmanned aerial vehicle (UAV)-assisted device-to-device (D2D) caching networks, the uncertainty from unpredictable content demands and variable user positions poses a significant challenge for traditional optimization methods, often making them impractical. Multi-agent deep reinforcement learning (MADRL) offers significant advantages in optimizing multi-agent system decisions and serves as an effective and practical alternative. However, its application in large-scale dynamic environments is severely limited by the curse of dimensionality and communication overhead. To resolve this problem, we develop a scalable heterogeneous multi-agent mean-field actor-critic (SH-MAMFAC) framework. The framework treats ground users (GUs) and UAVs as distinct agents and designs cooperative rewards to convert the resource allocation problem into a fully cooperative game, enhancing global network performance. We also implement a mixed-action mapping strategy to handle discrete and continuous action spaces. A mean-field MADRL framework is introduced to minimize individual agent training loads while enhancing total cache hit probability (CHP). The simulation results show that our algorithm improves CHP and reduces transmission delay. A comparative analysis with existing mainstream deep reinforcement learning (DRL) algorithms shows that SH-MAMFAC significantly reduces training time and maintains high CHP as GU count grows. Additionally, by comparing with SH-MAMFAC variants that do not include trajectory optimization or power control, the proposed joint design scheme significantly reduces transmission delay.

In massive multi-user multi-input–multi-output (MU-MIMO) downlink transmission systems, the total power consumption of the base station is mainly attributed to high-resolution digital-to-analog converters (DACs). Therefore, utilizing one-bit DACs can effectively reduce the system power consumption. This work investigates the one-bit DACs precoding problem in massive MU-MIMO systems. Typically, the one-bit DACs precoding problem based on the minimum mean square error criterion is an NP-hard problem, making it challenging to solve. By exploiting the structural properties of nonconvex constraints, the nonconvex optimization problem is transformed into an equivalent continuous-domain mathematical programming problem with equilibrium constraints. The projected gradient method is used for obtaining the quantized precoding signals, and the method is optimized by introducing a step-size adjustment coefficient. Simulation results demonstrate that the improved projected gradient method can effectively reduce the uncoded bit error rate and achieve a performance gain of approximately $\text{2}\,\text{dB}$ with 16 quadrature amplitude modulation signaling. The simulations also demonstrate the better robustness of the suggested method with imperfect channel state information. Specifically, we evaluate the power consumption of the proposed algorithm to that of the zero-forcing precoder with different resolution DACs, illustrating its superior power efficiency. Furthermore, we demonstrate the effectiveness of the proposed algorithm through a theoretical analysis of the convergence, complexity, and exact property of the optimal solution.