User-centric cell-free massive multiple-input multiple-output (UC CF mMIMO) is a key technology for 6G. In UC CF mMIMO each user equipment (UE) is jointly served by a small subset of a large set of cooperating access points (APs). The APs are connected through the fronthaul to a central unit (CU) for joint processing. The key to efficiency is the UE-APs association procedure together with the APs/CU cooperative signal decoding. The first is essential for reducing fronthaul signaling and the latter for removing interference. In this work, we design an algorithm to jointly perform UE-APs association and decoding. The aim is to maximize the mean squared error (MSE) between transmitted and CU decoded data, considering the fronthaul data rate limitations. In our approach, the two aspects are jointly considered, allowing improved performance while keeping the amount of fronthaul signaling low. Furthermore, the proposed method has low-complexity, requiring only the knowledge of a signal quality indicator of each UE at the CU to perform joint decoding and UE-APs association. The numerical results have shown that there is a tradeoff between performance and fronthaul requirements, where the proposed method with fewer APs may achieve the same bit error rate (BER) performance in comparison to conventional CF systems where each AP serves all UEs.

In this paper, we propose a distributed hybrid analog-digital multi-user equalizer combined with an analog precoder for a wideband millimeter-wave (mmWave) cell-free massive multiple-input multiple-output (CF mMIMO) systems. A low complexity angle of departure (AoD) analog-based precoder is applied at the transmitter, while a distributed architecture is considered at the receiver side, i.e., the hybrid analog/digital beamforming processing is executed at the access points (APs), while the final data fusion is performed at the central processing unit (CPU) combiner. Herein, the average mean square error (MSE) between the transmitted and received signals is used as a metric to optimize the APs hybrid equalizer. Then, the noise variance is included for optimal data combination. The results show that the performance of the proposed hybrid scheme with the AoD analog precoder achieves a close result to the fully digital counterpart and fully centralized approach.

The current cellular systems are conventional cell-centric based architectures, and thus the inter-cell interference greatly affects the quality of service (QoS) of users located at the border of those service areas. For that reason, future wireless systems will rely on a cell-less deployment by using an alternative promising network infrastructure known as cell-free (CF) technology. Here, in contrast to the current networks, there are no concepts of cells or cell boundaries since it is a disruptive paradigm where a large number of geographically distributed access points (APs) simultaneously and jointly serve a much smaller number of user terminals (UTs) to fulfill the exponentially increasing number of users. Therefore, this paper aims to contribute on this field and considers the design of hybrid analog/digital multi-user equalizer for broadband millimeter wave (mmWave) CF massive multiple-input multiple-output (mMIMO) systems, using a distributed beamforming approach (processing is distributed by APs and central unit (CU)) to reduce the fronthaul requirements. First, only a low complexity analog precoder is applied to the transmitter, while a joint design of APs hybrid equalizer and CU combiner are considered by minimizing the average mean square error (MSE) between the transmitted and CU estimated signals. To achieve this joint design, a cyclic minimization method is used to iterate between APs hybrid equalizer and CU combiner, where the information is exchanged between APs and CU based on a set of error variances to improve detection performance. The simulation results show that the proposed precoder and the hybrid linear equalizer can achieve performance results very close to those of the fully digital system. Moreover, just two iterations of the proposed algorithm are already enough to match the full centralized approach performance.