Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287713
Santiago López-Tapia, Alice Lucas, R. Molina, A. Katsaggelos
Despite the success of Recurrent Neural Networks in tasks involving temporal video processing, few works in Video Super-Resolution (VSR) have employed them. In this work we propose a new Gated Recurrent Convolutional Neural Network for VSR adapting some of the key components of a Gated Recurrent Unit. Our model employs a deformable attention module to align the features calculated at the previous time step with the ones in the current step and then uses a gated operation to combine them. This allows our model to effectively reuse previously calculated features and exploit longer temporal relationships between frames without the need of explicit motion compensation. The experimental validation shows that our approach outperforms current VSR learning based models in terms of perceptual quality and temporal consistency.
{"title":"Gated Recurrent Networks for Video Super Resolution","authors":"Santiago López-Tapia, Alice Lucas, R. Molina, A. Katsaggelos","doi":"10.23919/Eusipco47968.2020.9287713","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287713","url":null,"abstract":"Despite the success of Recurrent Neural Networks in tasks involving temporal video processing, few works in Video Super-Resolution (VSR) have employed them. In this work we propose a new Gated Recurrent Convolutional Neural Network for VSR adapting some of the key components of a Gated Recurrent Unit. Our model employs a deformable attention module to align the features calculated at the previous time step with the ones in the current step and then uses a gated operation to combine them. This allows our model to effectively reuse previously calculated features and exploit longer temporal relationships between frames without the need of explicit motion compensation. The experimental validation shows that our approach outperforms current VSR learning based models in terms of perceptual quality and temporal consistency.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"13 1","pages":"700-704"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78371605","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287801
R. Wildhaber, Elizabeth Ren, F. Waldmann, Hans-Andrea Loeliger
Local polynomial approximations represent a versatile feature space for time-domain signal analysis. The parameters of such polynomial approximations can be computed by efficient recursions using autonomous linear state space models and often allow analytical solutions for quantities of interest. The approach is illustrated by practical examples including the estimation of the delay difference between two acoustic signals and template matching in electrocardiogram signals with local variations in amplitude and time scale.
{"title":"Signal Analysis Using Local Polynomial Approximations","authors":"R. Wildhaber, Elizabeth Ren, F. Waldmann, Hans-Andrea Loeliger","doi":"10.23919/Eusipco47968.2020.9287801","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287801","url":null,"abstract":"Local polynomial approximations represent a versatile feature space for time-domain signal analysis. The parameters of such polynomial approximations can be computed by efficient recursions using autonomous linear state space models and often allow analytical solutions for quantities of interest. The approach is illustrated by practical examples including the estimation of the delay difference between two acoustic signals and template matching in electrocardiogram signals with local variations in amplitude and time scale.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"52 6 1","pages":"2239-2243"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77790220","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287673
Yao-Shan Hsiao, Mingyu Yang, Hun-Seok Kim
This paper presents a learning-based algorithm that estimates the time of arrival (ToA) of radio frequency (RF) signals from channel frequency response (CFR) measurements for wireless localization applications. A generator neural network is proposed to enhance the effective bandwidth of the narrowband CFR measurement and to produce a high-resolution estimation of channel impulse response (CIR). In addition, two regressor neural networks are introduced to perform a two-step coarsefine ToA estimation based on the enhanced CIR. For simulated channels, the proposed method achieves 9% – 58% improved root mean squared error (RMSE) for distance ranging and up to 22% improved false detection rate compared with conventional super-resolution algorithms. For real-world measured channels, the proposed method exhibits an improvement of 1.3m in distance error at 90 percentile.
{"title":"Super-Resolution Time-of-Arrival Estimation using Neural Networks","authors":"Yao-Shan Hsiao, Mingyu Yang, Hun-Seok Kim","doi":"10.23919/Eusipco47968.2020.9287673","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287673","url":null,"abstract":"This paper presents a learning-based algorithm that estimates the time of arrival (ToA) of radio frequency (RF) signals from channel frequency response (CFR) measurements for wireless localization applications. A generator neural network is proposed to enhance the effective bandwidth of the narrowband CFR measurement and to produce a high-resolution estimation of channel impulse response (CIR). In addition, two regressor neural networks are introduced to perform a two-step coarsefine ToA estimation based on the enhanced CIR. For simulated channels, the proposed method achieves 9% – 58% improved root mean squared error (RMSE) for distance ranging and up to 22% improved false detection rate compared with conventional super-resolution algorithms. For real-world measured channels, the proposed method exhibits an improvement of 1.3m in distance error at 90 percentile.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"35 1","pages":"1692-1696"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81689834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287792
Youssef Mourchid, M. Donias, Y. Berthoumieu
This paper presents a deep automatic colorization approach which avoids any manual intervention. Recently Generative Adversarial Network (GANs) approaches have proven their effectiveness for image colorization tasks. Inspired by GANs methods, we propose a novel colorization model that produces more realistic quality results. The model employs an additional discriminator which works in the feature domain. Using a feature discriminator, our generator produces structural high-frequency features instead of noisy artifacts. To achieve the required level of details in the colorization process, we incorporate non-adversarial losses from recent image style transfer techniques. Besides, the generator architecture follows the general shape of U-Net, to transfer information more effectively between distant layers. The performance of the proposed model was evaluated quantitatively as well as qualitatively with places365 dataset. Results show that the proposed model achieves more realistic colors with less artifacts compared to the state-of-the-art approaches.
{"title":"Automatic Image Colorization based on Multi-Discriminators Generative Adversarial Networks","authors":"Youssef Mourchid, M. Donias, Y. Berthoumieu","doi":"10.23919/Eusipco47968.2020.9287792","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287792","url":null,"abstract":"This paper presents a deep automatic colorization approach which avoids any manual intervention. Recently Generative Adversarial Network (GANs) approaches have proven their effectiveness for image colorization tasks. Inspired by GANs methods, we propose a novel colorization model that produces more realistic quality results. The model employs an additional discriminator which works in the feature domain. Using a feature discriminator, our generator produces structural high-frequency features instead of noisy artifacts. To achieve the required level of details in the colorization process, we incorporate non-adversarial losses from recent image style transfer techniques. Besides, the generator architecture follows the general shape of U-Net, to transfer information more effectively between distant layers. The performance of the proposed model was evaluated quantitatively as well as qualitatively with places365 dataset. Results show that the proposed model achieves more realistic colors with less artifacts compared to the state-of-the-art approaches.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"23 1","pages":"1532-1536"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78857899","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287337
Santiago Ruiz, T. Waterschoot, M. Moonen
Distributed combined acoustic echo cancellation (AEC) and noise reduction (NR) in a wireless acoustic sensor network (WASN) is tackled by using a specific version of the PK-GEVD-DANSE algorithm (cfr. [1]). Although this algorithm was initially developed for distributed NR with partial prior knowledge of the desired speech steering vector, it is shown that it can also be used for AEC combined with NR. Simulations have been carried out using centralized and distributed batch-mode implementations to verify the performance of the algorithm in terms of AEC quantified with the echo return loss enhancement (ERLE), as well as in terms of the NR quantified with the signal- to-noise ratio (SNR).
{"title":"Distributed combined acoustic echo cancellation and noise reduction using GEVD-based distributed adaptive node specific signal estimation with prior knowledge","authors":"Santiago Ruiz, T. Waterschoot, M. Moonen","doi":"10.23919/Eusipco47968.2020.9287337","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287337","url":null,"abstract":"Distributed combined acoustic echo cancellation (AEC) and noise reduction (NR) in a wireless acoustic sensor network (WASN) is tackled by using a specific version of the PK-GEVD-DANSE algorithm (cfr. [1]). Although this algorithm was initially developed for distributed NR with partial prior knowledge of the desired speech steering vector, it is shown that it can also be used for AEC combined with NR. Simulations have been carried out using centralized and distributed batch-mode implementations to verify the performance of the algorithm in terms of AEC quantified with the echo return loss enhancement (ERLE), as well as in terms of the NR quantified with the signal- to-noise ratio (SNR).","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"363 1","pages":"206-210"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76557926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287840
Tong Wei, B. Liao, Peng Xiao, Ziyang Cheng
In this paper, the problem of transmit beampattern synthesis (i.e., transmit beamforming) in multiple input multiple output (MIMO) radar which deploys one-bit digital-to-analog converts (DACs) is investigated. We aim to design appropriate transmit signal sequences, which are quantized by one-bit DACs, such that the amount of transmit energy can be focused into mainlobe region as much as possible, meanwhile, the leakage power of sidelobe region is minimized. It is shown that these requirements can be simultaneously fulfilled by minimizing the integrated sidelobe to mainlobe ratio (ISMR) of transmit beampattern with discrete binary constraints. According to this concept, we utilize the alternating direction multiplier method (ADMM) framework to solve the resulting nonconvex problem. Simulation results will demonstrate the effectiveness and improved performance of the proposed method.
{"title":"Transmit Beampattern Synthesis for MIMO Radar with One-Bit DACs","authors":"Tong Wei, B. Liao, Peng Xiao, Ziyang Cheng","doi":"10.23919/Eusipco47968.2020.9287840","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287840","url":null,"abstract":"In this paper, the problem of transmit beampattern synthesis (i.e., transmit beamforming) in multiple input multiple output (MIMO) radar which deploys one-bit digital-to-analog converts (DACs) is investigated. We aim to design appropriate transmit signal sequences, which are quantized by one-bit DACs, such that the amount of transmit energy can be focused into mainlobe region as much as possible, meanwhile, the leakage power of sidelobe region is minimized. It is shown that these requirements can be simultaneously fulfilled by minimizing the integrated sidelobe to mainlobe ratio (ISMR) of transmit beampattern with discrete binary constraints. According to this concept, we utilize the alternating direction multiplier method (ADMM) framework to solve the resulting nonconvex problem. Simulation results will demonstrate the effectiveness and improved performance of the proposed method.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"19 1","pages":"1827-1830"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87485898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287486
Sarah Itani, D. Thanou
Graph Signal Processing (GSP) addresses the analysis of data living on an irregular domain which can be modeled with a graph. This capability is of great interest for the study of brain connectomes. In this case, data lying on the nodes of the graph are considered as signals (e.g., fMRI time-series) that have a strong dependency on the graph topology (e.g., brain structural connectivity). In this paper, we adopt GSP tools to build features related to the frequency content of the signals. To make these features highly discriminative, we apply an extension of the Fukunaga-Koontz transform. We then use these new features to train a decision tree for the prediction of autism spectrum disorder. Interestingly, our framework outperforms state-of-the-art methods on the publicly available ABIDE dataset.
{"title":"A Graph Signal Processing Framework for the Classification of Temporal Brain Data","authors":"Sarah Itani, D. Thanou","doi":"10.23919/Eusipco47968.2020.9287486","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287486","url":null,"abstract":"Graph Signal Processing (GSP) addresses the analysis of data living on an irregular domain which can be modeled with a graph. This capability is of great interest for the study of brain connectomes. In this case, data lying on the nodes of the graph are considered as signals (e.g., fMRI time-series) that have a strong dependency on the graph topology (e.g., brain structural connectivity). In this paper, we adopt GSP tools to build features related to the frequency content of the signals. To make these features highly discriminative, we apply an extension of the Fukunaga-Koontz transform. We then use these new features to train a decision tree for the prediction of autism spectrum disorder. Interestingly, our framework outperforms state-of-the-art methods on the publicly available ABIDE dataset.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"38 1","pages":"1180-1184"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87088152","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287567
A. Borowicz
Adaptive Fourier decomposition (AFD) provides an expansion of an analytic function into a sum of basic signals, called mono-components. Unlike the Fourier series decomposition, the AFD is based on an adaptive rational orthogonal system, hence it is better suited for analyzing non-stationary data. The most popular algorithm for the AFD decomposes any signal in such a way that the energy of the low-frequency components is maximized. Unfortunately, this results in poor energy compaction of high-frequency components. In this paper, we develop a novel algorithm for the AFD. The key idea is to maximize the energy of any components no matter how big or small the corresponding frequencies are. A comparative evaluation was conducted of the signal reconstruction efficiency of the proposed approach and several conventional algorithms by using speech recordings. The experimental results show that with the new algorithm, it is possible to get a better performance in terms of the reconstruction quality and energy compaction property.
{"title":"Improving Energy Compaction of Adaptive Fourier Decomposition","authors":"A. Borowicz","doi":"10.23919/Eusipco47968.2020.9287567","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287567","url":null,"abstract":"Adaptive Fourier decomposition (AFD) provides an expansion of an analytic function into a sum of basic signals, called mono-components. Unlike the Fourier series decomposition, the AFD is based on an adaptive rational orthogonal system, hence it is better suited for analyzing non-stationary data. The most popular algorithm for the AFD decomposes any signal in such a way that the energy of the low-frequency components is maximized. Unfortunately, this results in poor energy compaction of high-frequency components. In this paper, we develop a novel algorithm for the AFD. The key idea is to maximize the energy of any components no matter how big or small the corresponding frequencies are. A comparative evaluation was conducted of the signal reconstruction efficiency of the proposed approach and several conventional algorithms by using speech recordings. The experimental results show that with the new algorithm, it is possible to get a better performance in terms of the reconstruction quality and energy compaction property.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"62 1","pages":"2348-2352"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86369905","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-24DOI: 10.23919/Eusipco47968.2020.9287638
P. Steiner, Simon Stone, P. Birkholz, A. Jalalvand
Currently, convolutional neural networks (CNNs) define the state of the art for multipitch tracking in music signals. Echo State Networks (ESNs), a recently introduced recurrent neural network architecture, achieved similar results as CNNs for various tasks, such as phoneme or digit recognition. However, they have not yet received much attention in the community of Music Information Retrieval. The core of ESNs is a group of unordered, randomly connected neurons, i.e., the reservoir, by which the low-dimensional input space is non-linearly transformed into a high-dimensional feature space. Because only the weights of the connections between the reservoir and the output are trained using linear regression, ESNs are easier to train than deep neural networks. This paper presents a first exploration of ESNs for the challenging task of multipitch tracking in music signals. The best results presented in this paper were achieved with a bidirectional two-layer ESN with 20 000 neurons in each layer. Although the final F-score of 0.7198 still falls below the state of the art (0.7370), the proposed ESN-based approach serves as a baseline for further investigations of ESNs in audio signal processing in the future.
目前，卷积神经网络(cnn)定义了音乐信号中多音高跟踪的最新技术。回声状态网络(Echo State Networks, ESNs)是最近引入的一种循环神经网络架构，在各种任务(如音素或数字识别)上取得了与cnn相似的结果。然而，它们在音乐信息检索界还没有得到足够的重视。ESNs的核心是一组无序、随机连接的神经元，即存储库，通过它将低维输入空间非线性转换为高维特征空间。因为只有储层和输出之间的连接权值是用线性回归训练的，所以esn比深度神经网络更容易训练。本文首次探索了ESNs用于音乐信号中多音高跟踪的挑战性任务。本文给出的最佳结果是双向双层回声状态网络，每层有20,000个神经元。虽然最终的f值0.7198仍然低于目前的水平(0.7370)，但所提出的基于esn的方法可以作为未来音频信号处理中进一步研究esn的基线。
{"title":"Multipitch tracking in music signals using Echo State Networks","authors":"P. Steiner, Simon Stone, P. Birkholz, A. Jalalvand","doi":"10.23919/Eusipco47968.2020.9287638","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287638","url":null,"abstract":"Currently, convolutional neural networks (CNNs) define the state of the art for multipitch tracking in music signals. Echo State Networks (ESNs), a recently introduced recurrent neural network architecture, achieved similar results as CNNs for various tasks, such as phoneme or digit recognition. However, they have not yet received much attention in the community of Music Information Retrieval. The core of ESNs is a group of unordered, randomly connected neurons, i.e., the reservoir, by which the low-dimensional input space is non-linearly transformed into a high-dimensional feature space. Because only the weights of the connections between the reservoir and the output are trained using linear regression, ESNs are easier to train than deep neural networks. This paper presents a first exploration of ESNs for the challenging task of multipitch tracking in music signals. The best results presented in this paper were achieved with a bidirectional two-layer ESN with 20 000 neurons in each layer. Although the final F-score of 0.7198 still falls below the state of the art (0.7370), the proposed ESN-based approach serves as a baseline for further investigations of ESNs in audio signal processing in the future.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"2012 1","pages":"126-130"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86352897","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We investigate user activity and data detection problem in a multiple-input multiple-output uplink cloud-radio access network, where the data matrix over a time-frame has overlapped burst sparsity due to sporadic user activity. We exploit this sparsity to recover data by proposing a weighted prior-sparse Bayesian learning algorithm. The proposed algorithm, due to carefully selected prior, captures not only the overlapped burst sparsity across time but also the block sparsity due to multi-user antennas. We also derive hyperparameter updates, and estimate the weight parameters using the support estimated via index-wise log-likelihood ratio test. We numerically demonstrate that the proposed algorithm has much lower bit error rate than the state-of-the-art competing algorithms.
{"title":"User Activity And Data Detection For MIMO Uplink C-RAN Using Bayesian Learning","authors":"Anupama Rajoriya, Vidushi Katiyar, Rohit Budhiraja","doi":"10.23919/Eusipco47968.2020.9287867","DOIUrl":"https://doi.org/10.23919/Eusipco47968.2020.9287867","url":null,"abstract":"We investigate user activity and data detection problem in a multiple-input multiple-output uplink cloud-radio access network, where the data matrix over a time-frame has overlapped burst sparsity due to sporadic user activity. We exploit this sparsity to recover data by proposing a weighted prior-sparse Bayesian learning algorithm. The proposed algorithm, due to carefully selected prior, captures not only the overlapped burst sparsity across time but also the block sparsity due to multi-user antennas. We also derive hyperparameter updates, and estimate the weight parameters using the support estimated via index-wise log-likelihood ratio test. We numerically demonstrate that the proposed algorithm has much lower bit error rate than the state-of-the-art competing algorithms.","PeriodicalId":6705,"journal":{"name":"2020 28th European Signal Processing Conference (EUSIPCO)","volume":"21 1","pages":"1742-1746"},"PeriodicalIF":0.0,"publicationDate":"2021-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89220420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：