Downlink Power Control for Cell-Free Massive MIMO With Deep Reinforcement Learning

Luo, Lirui; Zhang, Jiayi; Chen, Shuaifei; Zhang, Xiaodan; Ai, Bo; Ng, Derrick Wing Kwan

doi:10.1109/tvt.2022.3162585

Cited by 23 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reference [38] was the first scientific attempt to employ UL-based power control in cell-free m-MIMO orientations. In [94], a DRL framework has been considered for power control in cell-free m-MIMO systems. Results indicate that the proposed approach can improve all considered performance metrics (sum user rate, minimum user rate and SINR) with significantly reduced computational complexity compared to the complex optimization solver.…”

Section: Distributed and Cell-free Massive Mimo Configurationsmentioning

confidence: 99%

A Survey on Machine Learning Techniques for Massive MIMO Configurations: Application Areas, Performance Limitations and Future Challenges

Gkonis

2023

IEEE Access

View full text Add to dashboard Cite

The deployment of fifth-generation (5G) broadband wireless cellular networks has enabled the support of highly demanding applications, paving the way towards global broadband connectivity. In the same context, related advances in network infrastructure, such as network function virtualization via the decoupling of related functionalities from hardware equipment has leveraged end-to-end service support in highly heterogeneous environments. To this end, uninterrupted provision of service necessitates a holistic network redesign where access points (APs) should be placed and configured dynamically. One of the key supporting technologies of 5G communications are massive multiple input multiple output (m-MIMO) configurations, deployed over various APs, either in a centralized or in a distributed manner. However, such a configuration would pose an additional degree of complexity during network planning, since on one hand accurate channel estimation for all potential links is limited due to the pilot contamination effect, and on the other hand the number of actually deployed radio frequency chains is again limited when compared to the fully digital approach, in an effort to relax hardware and computational burden. Moreover, additional issues should be taken into consideration as well, such as the coexistence of m-MIMO configurations with other novel technologies in the 5G air interface as well as the need for provision of acceptable quality of service to a large number of mobile users. In an effort to deal with the aforementioned issues, machine learning (ML) algorithms have emerged over the last decade as a promising solution that can deal with complex optimization problems. In this context, the main goal of this survey paper is to present the state-of-the-art approaches in the deployment of ML methods in m-MIMO configurations. Unlike other related survey papers, this is the first scientific attempt to cover all aspects of m-MIMO practical implementations (i.e., channel estimation, MIMO precoding and signal detection, hybrid beamforming, distributed configurations, user pairing and power allocation in non-orthogonal multiple access), in the context of ML-aided calculations. To this end, all presented works are categorized according to the studied aspect, while in the same context potential limitations and future challenges are highlighted as well.INDEX TERMS Massive MIMO, machine learning, channel estimation, hybrid beamforming, cell-free orientations, NOMA.

show abstract

Section: Distributed and Cell-free Massive Mimo Configurationsmentioning

confidence: 99%

A Survey on Machine Learning Techniques for Massive MIMO Configurations: Application Areas, Performance Limitations and Future Challenges

Gkonis

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…For the power allocation based on infinite blocklength in [39], [40], the problem can be converted into a convex problem by introducing slack variables, which can be readily solved by the bisection search algorithm. However, maximizing the weighted sum rate is an NP-hard problem, which cannot be readily solved.…”

Section: A Problem Formulationmentioning

confidence: 99%

Resource Allocation for Cell-free Massive MIMO-enabled URLLC Downlink Systems

Peng¹,

Hong²,

Pan³

et al. 2023

Preprint

View full text Add to dashboard Cite

Ultra-reliable and low-latency communication (URLLC) is a pivotal technique for enabling the wireless control over industrial Internet-of-Things (IIoT) devices. By deploying distributed access points (APs), cell-free massive multiple-input and multiple-output (CF mMIMO) has great potential to provide URLLC services for IIoT devices. In this paper, we investigate CF mMIMO-enabled URLLC in a smart factory. Lower bounds (LBs) of downlink ergodic data rate under finite channel blocklength (FCBL) with imperfect channel state information (CSI) are derived for maximum-ratio transmission (MRT), full-pilot zero-forcing (FZF), and local zero-forcing (LZF) precoding schemes.Meanwhile, the weighted sum rate is maximized by jointly optimizing the pilot power and transmission power based on the derived LBs. Specifically, we first provide the globally optimal solution of the pilot power, and then introduce some approximations to transform the original problems into a series of subproblems, which can be expressed in a geometric programming (GP) form that can be readily solved. Finally, an iterative algorithm is proposed to optimize the power allocation based on various precoding schemes. Simulation results demonstrate that the proposed algorithm is superior to the existing algorithms, and that the quality of URLLC services will benefit by deploying more APs, except for the FZF precoding scheme.

show abstract

“…The ePCNet is demonstrated to outperform state-of-theart power allocation solutions such as WMMSE and greedy power allocation. Furthermore, in [24] and [25], deep reinforcement learning-based power allocation was leveraged to deal with the max-min and SE maximization power allocation problem in CF mMIMO systems, respectively. In [26], authors resorted to deep CNN to maximize SE in CF mMIMO systems, which outperforms the well-known use-and-then-forget-based power allocation.…”

Section: A Related Workmentioning

confidence: 99%

Heterogeneous graph neural network for power allocation in multicarrier-division duplex cell-free massive MIMO systems

Li¹,

Yang²,

Maunder³

et al. 2022

Preprint

View full text Add to dashboard Cite

In-band full duplex-based cell-free (IBFD-CF) systems suffer from severe interference problem including self-interference (SI) and cross-link interference (CLI), especially when cell-free (CF) systems are operated in a distributed way. To this end, we propose multicarrier-division duplex (MDD) as an enabler for full-duplex (FD)-style operation in distributed CF massive MIMO systems, where DL and UL transmissions take place simultaneously at the same frequency band but mutually orthogonal subcarrier sets. In order to maximize the spectral efficiency (SE) in the proposed systems, we present heterogeneous graph neural network specific for CF systems (CF-HGNN), which consists of an adaptive node embedding layer, meta-path based message passing, meta-path based attention and downstream power allocation learning. In particular, the adaptive node embedding layer can handle the varying number of access points (APs), mobile stations (MSs) and subcarriers, and the involved attention mechanism enables each AP/MS node in CF-HGNN to aggregate the information from interfering path and communication path with different priorities. Numerical results show that CF-HGNN is capable of using 10 4 times less operation time to achieve the 99% performance of the SE of quadratic transform and successive convex approximation (QT-SCA). Additionally, CF-HGNN also significantly outperforms unfair greedy method in terms of SE performance. Furthermore, CF-HGNN exhibits good adaptivity to varying number of nodes and subcarriers, and also generalization ability to different sizes of CF network.

show abstract

Downlink Power Control for Cell-Free Massive MIMO With Deep Reinforcement Learning

Cited by 23 publications

References 12 publications

A Survey on Machine Learning Techniques for Massive MIMO Configurations: Application Areas, Performance Limitations and Future Challenges

A Survey on Machine Learning Techniques for Massive MIMO Configurations: Application Areas, Performance Limitations and Future Challenges

Resource Allocation for Cell-free Massive MIMO-enabled URLLC Downlink Systems

Heterogeneous graph neural network for power allocation in multicarrier-division duplex cell-free massive MIMO systems

Contact Info

Product

Resources

About