XHXiaoming Hou

Xiaoming Hou

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xiaoming Hou has a total of 12 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A Deep Learning-Based Codebook Design Method With MED Constraints for Uplink SCMA Systems

    by Yu Zheng, Xiaoming Hou, Hui Wang, Shengli Zhang

    Sparse code multiple access (SCMA) is a competitive candidate multiple access technology for future wireless communication systems. In recent years, SCMA systems are modeled as autoencoders to design high performance codebooks, which reduces the suboptimal problem caused by multi-stage optimization in traditional codebook design schemes. However, the current SCMA autoencoder only considers the AWGN channel, so the designed codebook can achieve excellent performance in AWGN channels but poor performance in Rayleigh fading channels. In this article, we propose a novel SCMA autoencoder scheme that incorporates minimum Euclidean Distance (MED) constraints for designing uplink SCMA codebooks. The core design of the proposed scheme is to introduce a constraint network, which aims to maximizing the MED between superimposed codewords while ensuring the distance between each user’s one-dimensional codewords. First, the framework of the proposed SCMA autoencoder model is introduced. Then, the network structures that comprise the proposed model are presented separately. Finally, the details of the proposed loss function are discussed. Simulation results show that the BER performance for the proposed codebooks is better than that of the existing codebook.


  • A Novel DCMA Codebook Design Scheme Over Downlink Rayleigh Fading Channels

    by Yu Zheng, Xiaoming Hou, Zhaolong Zhang, Shengli Zhang, Hui Wang

    Dense Code Multiple Access (DCMA) is a variant of sparse code multiple access (SCMA) that replaces sparse codewords with dense ones, enhancing diversity gain. In this letter, we propose a novel approach to DCMA codebook design for downlink Rayleigh fading channels. First, SCMA codebook with a separable structure is designed to ensure uncorrelated signals across different resource elements (REs). Next, the codebook is extended using a corresponding linear constellation precoding (LCP) matrix to produce the final DCMA codebook. By ensuring uncorrelated signals across REs, the diversity gain enhancement from LCP is maximized. Simulation results demonstrate that the proposed codebook outperforms existing DCMA codebooks.


  • High SNR SCMA Detection via Transfer Learning From Low SNR Region

    by Yu Zheng, Xiaoming Hou, Jiantao Xin, Hui Wang, Ming Jiang, Shengli Zhang

    Sparse code multiple access (SCMA) is a competitive candidate multiple access technology for future wireless communication systems. In this article, a transfer learning (TL)-based SCMA detection scheme is proposed to improve the performance of the deep neural network (DNN) detector for the downlink SCMA system. First, we propose a detection framework that preserves model parameters trained on datasets with varying signal-to-noise ratios (SNRs) and adaptively selects them during detection based on the estimated channel SNR. Then, we analyze the reason why the DNN detector, which can achieve the same BER performance as the message passing algorithm (MPA) in the low SNR region, fails to achieve MPA performance in the high SNR region. Later, a TL-based SCMA detection scheme is proposed, which consists of pretraining, fine-tuning and online detection. Simulation results show the proposed TL-based SCMA detection scheme can achieve improved BER performance compared to the deep-learning (DL)-based scheme trained from scratch. Moreover, to alleviate the gap between the source domain and the target domain, a successive TL strategy is proposed, which makes the transfer process smoother and further improves the performance by introducing intermediate states.


  • Rec-PF: Data-Driven Large-Scale Deep Learning Recommendation Model Training Optimization Based on Tensor-Train Embedding Table With Photovoltaic Forecast

    by Yunfeng Li, Zheng Wang, Chenhao Ren, Xiaoming Hou, Shengli Zhang

    Photovoltaic (PV) power forecasting is important for promoting the integration of renewable energy sources. However, neural network-based methods, particularly deep learning for PV power forecasting, face challenges with computational and memory requirements when dealing with industry-scale datasets. To address this, we introduce Rec-PF , a robust computational framework employing the tensor-train (TT) technique. This framework aims to streamline the training process of massive deep learning recommendation models (DLRMs) on constrained resources. Rec-PF employs a high-performance compressed embedding table, enhancing TT decomposition using key computing primitives. It serves as a drop-in replacement for the PyTorch API. Additionally, Rec-PF utilizes an index reordering technique to leverage local and global information from training inputs, thereby enhancing performance. Furthermore, Rec-PF adopts a pipeline training model, eliminating the need for communication between training workers and host memory. We are pioneers in applying DLRM to PV power prediction to reduce training time without compromising accuracy. Our approach demonstrates a twofold improvement in training time compared to methods that do not incorporate our approach. To better demonstrate the enhanced performance of the algorithm, we specifically compare its efficiency with other frameworks using datasets commonly employed in recommender systems. Comprehensive experiments indicate that Rec-PF is capable of processing the largest publicly accessible DLRM and PV datasets on a single GPU, offering a threefold acceleration compared to state-of-the-art DLRM and PV frameworks.


  • A Knowledge-Based Deep Learning Detection Scheme for Downlink SCMA Systems

    by Yu Zheng, Xiaoming Hou, Hui Wang, Ming Jiang, Shengli Zhang

    Sparse code multiple access (SCMA) is a promising code-domain non-orthogonal multiple access technology for future wireless communication systems. In the SCMA detector, deep learning (DL) technology has been adopted to improve the detection efficiency. However, most previous schemes are completely data-driven designed without using prior knowledge. In this letter, we propose a knowledge-based deep learning detection scheme (K-DLD) that incorporates prior knowledge into SCMA detection to lighten the neural network. Moreover, we propose to use a thinner but deeper network to further reduce the model complexity. Simulation results show that the proposed schemes significantly reduce the computation time without any performance loss.