Deep Learning-Based Limited Feedback Designs for MIMO Systems

Jang, Jeonghyeon; Lee, Hoon; Hwang, Sangwon; Ren, Haibao; Lee, Inkyu

doi:10.1109/lwc.2019.2962114

Cited by 33 publications

(25 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, we can also regard the LSTM layers as a refinement module that refines the initial output of the baseline decoder by using the antenna correlation. Specifically, once the baseline decoder outputs the initial reconstructed CSI Furthermore, the binary representation can be used to generate the bitstreams [33]. To an extent, this binary presentation-based CSI feedback can be regarded as a codebook-based datadriven feedback strategy [47].…”

Section: B Nn Modules In the Cocsinetmentioning

confidence: 99%

“…To an extent, the binarization operation can be regarded as a special case of quantization with B = 1. Unlike the general quantization, as in (17), the binarization operation introduces quantization noise with a zero-mean property, thereby making b(x) an unbiased estimator for 33]. We also compare the binarization operation and the common one-bit quantization in Fig.…”

Section: B Performance Of Nn-based Csi Magnitude Feedbackmentioning

confidence: 99%

“…Moreover, to further reduce the feedback overhead, an entropy coding module following a quantization module is used to further reduce the feedback bits in [32]. Unlike other quantization strategies, binary representation is exploited in [33] to generate bitstreams.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DL-based CSI Feedback and Cooperative Recovery in Massive MIMO

Guo¹,

Yang²,

Wen³

et al. 2020

Preprint

View full text Add to dashboard Cite

In this paper, we exploit the correlation between nearby user equipment (UE) and develop a deep learning-based channel state information (CSI) feedback and cooperative recovery framework, CoCsiNet, to reduce the feedback overhead. The CSI can be divided into two parts: shared by nearby UE and owned by individual UE. The key idea of exploiting the correlation is to reduce the overhead used to repeatedly feedback shared information. Unlike in the general autoencoder framework, an extra decoder is added at the base station to recover shared information from the feedback CSI of two nearby UE, but no modification is performed at the UE. For a UE with multiple antennas, we also introduce a baseline neural network architecture with long short-term memory modules to extract the correlation of nearby antennas. We propose two representative methods for converting CSI into bitstreams at the UE through quantization and binarization. Given that the CSI phase is not sparse, we propose a magnitudedependent phase feedback strategy that introduces CSI magnitude information to the loss function of the phase feedback. Simulation results show the effectiveness of the proposed CoCsiNet and explain the mechanism of the CoCsiNet via parameter visualization.

show abstract

Section: B Nn Modules In the Cocsinetmentioning

confidence: 99%

Section: B Performance Of Nn-based Csi Magnitude Feedbackmentioning

confidence: 99%

See 1 more Smart Citation

DL-based CSI Feedback and Cooperative Recovery in Massive MIMO

Guo¹,

Yang²,

Wen³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…This method is extended in [17], [18] for unsupervised-learning to maximize the system weighted sum-rate. In [19], a deep learning-aided transmission strategy is proposed for single-user MIMO system with limited feed back, which is capable of addressing both pilot-aided training and channel code selection. The authors of [20] develop a deep learning-based beamforming design to maximize the spectral efficiency of a single-user millimeter wave (mmWave) MISO system, which achieves higher spectral efficiency than conventional hybrid beamforming designs.…”

Section: Introductionmentioning

confidence: 99%

Machine Learning-Enabled Joint Antenna Selection and Precoding Design: From Offline Complexity to Online Performance

Chatzinotas

Nguyen

et al. 2021

IEEE Trans. Wireless Commun.

View full text Add to dashboard Cite

We investigate the performance of multi-user multiple-antenna downlink systems in which a base station (BS) serves multiple users via a shared wireless medium. In order to fully exploit the spatial diversity while minimizing the passive energy consumed by radio frequency (RF) components, the BS is equipped with M RF chains and N antennas, where M < N . Upon receiving pilot sequences to obtain the channel state information (CSI), the BS determines the best subset of M antennas for serving the users. We propose a joint antenna selection and precoding design (JASPD) algorithm to maximize the system sum rate subject to a transmit power constraint and quality of service (QoS) requirements. The JASPD overcomes the non-convexity of the formulated problem via a doubly iterative algorithm, in which an inner loop successively optimizes the precoding vectors, followed by an outer loop that tries all valid antenna subsets. Although approaching the (near) global optimality, the JASPD suffers from a combinatorial complexity, which may limit its application in real-time network operations. To overcome this limitation, we propose a learning-based antenna selection and precoding design algorithm (L-ASPA), which employs a deep neural network (DNN) to establish underlaying relations between the key system parameters and the selected antennas. The proposed L-ASPD is robust against the number of users and their locations, BS's transmit power, as well as the small-scale channel fading. With a well-trained learning model, it is shown that the L-ASPD significantly outperforms baseline schemes based on the block diagonalization [5] and a learning-assisted solution for broadcasting systems [29] and achieves higher effective sum rate than that of the JASPA under limited processing time. In addition, we observed that the proposed L-ASPD can reduce the computation complexity by 95% while retaining more than 95% of the optimal performance.

show abstract

“…Rayleigh fading situations. Also, recently, a number of approaches that utilize deep neural networks (DNNs) have been proposed to enable efficient CSI quantization [15]- [17]; yet, these publications mostly consider relatively low resolution quantization, as neural networks are hard to train for large quantization codebooks. When the channel exhibits temporal S. Schwarz is with the Institute of Telecommunications, TU Wien, Austria; email: sschwarz@nt.tuwien.ac.at; tel.…”

Section: Introductionmentioning

confidence: 99%

Recursive CSI Quantization of Time-Correlated MIMO Channels by Deep Learning Classification

Schwarz

2020

IEEE Signal Process. Lett.

View full text Add to dashboard Cite

In frequency division duplex (FDD) multiple-input multiple-output (MIMO) wireless communications, limited channel state information (CSI) feedback is a central tool to support advanced single-and multiuser MIMO beamforming/precoding. To achieve a given CSI quality, the CSI quantization codebook size has to grow exponentially with the number of antennas, leading to quantization complexity, as well as, feedback overhead issues for larger MIMO systems. We have recently proposed a multi-stage recursive Grassmannian quantizer that enables a significant complexity reduction of CSI quantization. In this paper, we show that this recursive quantizer can effectively be combined with deep learning classification to further reduce the complexity, and that it can exploit temporal channel correlations to reduce the CSI feedback overhead.

show abstract

Deep Learning-Based Limited Feedback Designs for MIMO Systems

Cited by 33 publications

References 16 publications

DL-based CSI Feedback and Cooperative Recovery in Massive MIMO

DL-based CSI Feedback and Cooperative Recovery in Massive MIMO

Machine Learning-Enabled Joint Antenna Selection and Precoding Design: From Offline Complexity to Online Performance

Recursive CSI Quantization of Time-Correlated MIMO Channels by Deep Learning Classification

Contact Info

Product

Resources

About