As we move towards beyond-fifth and sixth-generation wireless networks, new challenges emerge, including the need for higher data rates and lower latency. One promising solution to overcome some of these challenges is cell-free massive MIMO. However, a significant drawback of such systems is the high computational cost, particularly in downlink power allocation operations. This paper introduces a novel approach to tackle this issue by leveraging the proven capabilities of reinforcement learning (RL) algorithms. Specifically, we develop a multi-agent deep deterministic policy gradient (MADDPG) algorithm to maximize the sum spectral efficiency (SE) in power allocation schemes. In addition, a novel evolutionary hyperparameter optimization technique has been applied for the first time in this context. The proposed solution not only outperforms existing methods in the literature but also accurately replicates the behavior of the sum SE maximization approach solved as a convex problem. Notably, it achieves a significant computational efficiency improvement by reducing computation time by a factor of 1,000 compared to the convex solution. The results have been validated in properly configured cell-free networks, addressing the limitations of previous studies that did not consider a sufficiently large number of access points relative to user equipment. This work represents a significant advancement in the field, offering a promising and efficient solution for future wireless communication systems.

Cell-free massive multiple-input multiple-output (MIMO) is a technology that combines ultra-dense cellular networks and massive MIMO to enhance wireless transmission efficiency, being a promising solution for future communication scenarios. However, its implementation poses a significant computational load, which can hinder real-time implementation. The use of machine learning comes as a great solution to reduce this computational complexity. In this context, the present work explores the application of a supervised, low-complexity deep neural network (DNN) for power control, aiming to significantly reduce computational costs. The DNN was evaluated across various scenarios, generating a comprehensive dataset. The proposed DNN outperforms existing methods, demonstrating superior performance with three power optimization schemes: max-min spectral efficiency (SE) fairness, sum SE maximization, and fractional power control (FPC). Importantly, these results are consistent across diverse propagation conditions. A noteworthy outcome of this study is the achievement of a three-orders-of-magnitude reduction in computation time when compared to conventional techniques, paving the way for the practical implementation of cell-free massive MIMO systems.

In this paper, we investigate the tasks of binaural source distance estimation (SDE) and direction-of-arrival estimation (DOAE) using motion-based cues in a scenario with a walking listener. On top of performing both tasks as separate problems, we study two methods of solving the joint task of simultaneous source distance estimation and localization (SDEL), with a single model. Experiments are conducted for three different scenarios: a static receiver; a static receiver with a rotating head; and a freely moving listener inside a room. The study proposes rotation and translation features to include information about the receiver's motion during model training and studies the effects of these on the final performance. The work includes extended simulation of three datasets containing numerous testing scenarios for sound sources, covering a wide range of DOAs and a source-to-receiver distance up to 15 m. Results are further analyzed with respect to room reverberation, walking speed, as well as source-to-receiver distance. The presented outcomes show large improvements in both DOA and distance estimation for a model that uses motion-based cues as compared with a static scenario. These include a decrease of 9.50° in DOA and 1.56m in distance errors for a joint model, followed by 16.17° and 0.17m for separate models.