A Novel Deep Neural Network Based Approach for Sparse Code Multiple Access

Lin, Jinzhi; Feng, Shijie; Yang, Zhile; Zhang, Yun; Zhang, Yong

doi:10.48550/arxiv.1906.03169

Cited by 4 publications

(13 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For deep-learning-based approaches, however, we argue that the MPA receiver (or any other conventional receiver) can be replaced with a neural network-based multi-user decoder (NN-MUD) receiver whose complexity is no longer bounded by the constraint N. It is thus not necessary to impose any constraint on N, as in the existing deep learning-based approaches, for example, a sparse F * with N < K for SCMA [7] or dense F * with N = K, that is, all entries of one [8]. Because of the enormous complexity of (M J )!…”

Section: B Deep Learning-based Mu-mdm Design: Problem Formulationmentioning

confidence: 99%

“…Following the seminal paper by O'shea et al [5], an end-toend optimized architecture that exploits the similarity between a communication system and an autoencoder (AE) has gained considerable attention [6][7][8], [13][14][15][16][17][18][19][20][21]. One of the application areas for an AE-based optimization is a constellation (signal space diagram) design for multi-dimensional modulation (MDM).…”

Section: Introductionmentioning

confidence: 99%

“…It generates optimized multi-dimensional signal points with sparse resource mapping while realizing a deep neural network (DNN) decoder that achieves a better BER performance with a lower complexity than the MPA receiver. In [8], by contrast, dense code multiple access (DCMA) has been proposed by allowing for dense resource mapping and the use of deeplearning-based low-complexity receivers. Therefore, even if the connection is not explicitly made in the above studies, it is imperative to note that a CB design for CD-NOMA can be considered a constellation design for MU-MDM, which is a natural extension of the constellation design for SU-MDM.…”

Section: Introductionmentioning

confidence: 99%

“…In particular, we show that the BER performance can be improved by relaxing the constraint in allocating the same power to all CBs, which is however unnecessary in a receiver-agnostic design. Moreover, we attempt to maximize the resource utilization by relaxing the sparsity constraint, that is, considering the dense CD-NOMA as in [8] and [14]. In fact, the relaxation of both power and sparsity constraints enables the joint optimization of resource mapping and a constellation design that would lead to BER minimization, hopefully achieving a near single-user BER performance with the same bandwidth efficiency subject to the same order of training complexity.…”

Section: Introductionmentioning

confidence: 99%

“…More specifically, they employ L2normalization or cross-entropy with one-hot vector representation as their loss function, which minimizes the symbol error rate (SER) only, without considering a bit-to-symbol mapping function. Meanwhile, we note that binary vector representation is employed for the loss function to make their constellation achieve a better BER performance in [8], [15] and [22]. However, the optimal design of constellation signal points and its bit-to-symbol mapping varies with the signalto-noise (SNR) level.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Deep Learning-based Codebook Design for Code-domain Non-Orthogonal Multiple Access Approaching Single-User Bit Error Rate Performance

Han¹,

Seo²,

Abebe³

et al. 2021

Preprint

View full text Add to dashboard Cite

The codebook design for code-domain nonorthogonal multiple access (CD-NOMA) can be considered as a constellation design for multi-user multi-dimensional modulation (MU-MDM). This paper proposes an autoencoder (AE)-based constellation design for MU-MDM with the objective of achieving a comparable bit error rate (BER) performance to singleuser multi-dimensional modulation (SU-MDM), i.e., alleviating performance degradation in non-optimal AE design caused by overloading multiple users. Recognizing that various constraints in a receiver structure degrade a BER performance of the codebook design in the existing CD-NOMA, the MU-MDM design aims at global optimization on the common ground with SU-MDM by leveraging the agnosticism of the neural network-based multi-user decoder obtained from AE training, while mitigating the power normalization constraint and exploiting dense resource mapping in the MU-MDM AE structure. Moreover, as opposed to the existing loss function for MU-MDM which has failed to minimize BER for different levels of signal-to-noise ratio (SNR), a hyperparameterized loss function and proper training procedures are introduced to jointly optimize the signal points for MU-MDM constellation and their bit-to-symbol mapping. It has been demonstrated that the proposed design achieves a single-user BER bound with only 0.2dB loss, equivalently outperforming the existing CD-NOMA designs, while maintaining their overloading factor.

show abstract

Section: B Deep Learning-based Mu-mdm Design: Problem Formulationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Deep Learning-based Codebook Design for Code-domain Non-Orthogonal Multiple Access Approaching Single-User Bit Error Rate Performance

Han¹,

Seo²,

Abebe³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Trainable Projected Gradient Detector for Sparsely Spread Code Division Multiple Access

Takabe

Yamauchi

Wadayama

2020

2020 IEEE International Conference on Communications Workshops (ICC Workshops)

View full text Add to dashboard Cite

Sparsely spread code division multiple access (SCDMA) is a promising non-orthogonal multiple access technique for future wireless communications. In this paper, we propose a novel trainable multiuser detector called sparse trainable projected gradient (STPG) detector, which is based on the notion of deep unfolding. In the STPG detector, trainable parameters are embedded to a projected gradient descent algorithm, which can be trained by standard deep learning techniques such as back propagation and stochastic gradient descent. Advantages of the detector are its low computational cost and small number of trainable parameters, which enables us to treat massive SCDMA systems. In particular, its computational cost is smaller than a conventional belief propagation (BP) detector while the STPG detector exhibits nearly same detection performance with a BP detector. We also propose a scalable joint learning of signature sequences and the STPG detector for signature design. Numerical results show that the joint learning improves multiuser detection performance particular in the low SNR regime.

show abstract

CG-SCMA Codebook Design Based on Maximized Euclidian Distance

Mohamed¹,

Abdullah²

2023

JTIT

View full text Add to dashboard Cite

Sparse code multiple access (SCMA) is a multi-dimensional codebook based on a class of non-orthogonal multiple access (NOMA) technologies enabling the delivery of non-orthogonal resource elements to numerous users in 5G wireless communications without increasing complexity. This paper proposes a computer-generated sparse code multiple access (CG-SCMA) technique, where the minimum Euclidian distance (MED) of a star 16-point quadrature amplitude modulation is maximized by CG-SCMA, thus creating a complex SCMA codebook based on optimizing the difference between the first and other radiuses over rotated constellations. To specify the most suitable values for this constellation, it is divided into four sub-constellations using trellis coded modulation (TCM) in an effort to optimize MED. The new codebook has four sub-constellations with MED values of 3.85, 2.26, 2.26, and 3.85, respectively. Application of the message passing algorithm (MPA) ensures low complexity of the decoding process.

show abstract

A Novel Deep Neural Network Based Approach for Sparse Code Multiple Access

Cited by 4 publications

References 51 publications

Deep Learning-based Codebook Design for Code-domain Non-Orthogonal Multiple Access Approaching Single-User Bit Error Rate Performance

Deep Learning-based Codebook Design for Code-domain Non-Orthogonal Multiple Access Approaching Single-User Bit Error Rate Performance

Trainable Projected Gradient Detector for Sparsely Spread Code Division Multiple Access

CG-SCMA Codebook Design Based on Maximized Euclidian Distance

Contact Info

Product

Resources

About