Pub Date : 2018-11-01DOI: 10.1109/ICDSP.2018.8631862
Shaofei Xue, Zhijie Yan, Tao Yu, Zhang Liu
Far-field speech recognition is an essential technique for man-machine interactions. It aims to enable smart devices to recognize distant human speech. This technology is applied to many scenarios such as smart home appliances (smart loudspeaker, smart TV) and meeting transcription. Despite the significant advancement made in robust and far-field speech recognition after the introduction of deep neural network based acoustic models, the far-field speech recognition remains a challenging task due to various factors such as background noise, reverberation and even human voice interference. In this paper, we describe several technical advances for improving the performance of large-scale far-field speech recognition, including simulated data generation, improvements on front-end modules and neural network based acoustic models. Experimental results on several Mandarin Chinese speech recognition tasks have demonstrated that the combination of these technical advances can significantly outperform the conventional models.
{"title":"A Study on Improving Acoustic Model for Robust and Far-Field Speech Recognition","authors":"Shaofei Xue, Zhijie Yan, Tao Yu, Zhang Liu","doi":"10.1109/ICDSP.2018.8631862","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631862","url":null,"abstract":"Far-field speech recognition is an essential technique for man-machine interactions. It aims to enable smart devices to recognize distant human speech. This technology is applied to many scenarios such as smart home appliances (smart loudspeaker, smart TV) and meeting transcription. Despite the significant advancement made in robust and far-field speech recognition after the introduction of deep neural network based acoustic models, the far-field speech recognition remains a challenging task due to various factors such as background noise, reverberation and even human voice interference. In this paper, we describe several technical advances for improving the performance of large-scale far-field speech recognition, including simulated data generation, improvements on front-end modules and neural network based acoustic models. Experimental results on several Mandarin Chinese speech recognition tasks have demonstrated that the combination of these technical advances can significantly outperform the conventional models.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128166426","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631632
Xiaohu Zhang, Yuexian Zou
Environmental Sound Classification (ESC) plays a vital role in the field of machine auditory scene. Recently, the Highway Network CNN model has achieved the state-of-art results via solving the vanishing-gradient problem of much deeper CNN. However, carefully analyzing the Highway Network model shows that the Highway Network model lacks ability to maximize information flow between layers, which is essentially benefits the discriminative representation of acoustic events. Besides, the Highway Network model size is larger than 20MB for ESC task, which is still large for mobile applications. Regarding to these two issues, in this study, we propose a novel Densely Connected Highway Convolutional Network (DCH-Net) model for ESC task. Specifically, a densely highway module is developed which is able to ensure the maximum information flow between layers by connecting all layers directly with each other. Besides, to reduce the model size, a global average pooling layer is designed which replaces the traditional fully connection layers and the parameters of the model is greatly reduced. Experimental results show that our DCH-Net ESC model achieves accuracy of 69% and 90% on ESC50 and ESCIO dataset respectively, which is 2% and 10% higher than that of Highway Network based Highway networks ESC model. Meanwhile our model size is only 2MB.
{"title":"DCH-Net: Densely Connected Highway Convolution Neural Network for Environmental Sound Classification","authors":"Xiaohu Zhang, Yuexian Zou","doi":"10.1109/ICDSP.2018.8631632","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631632","url":null,"abstract":"Environmental Sound Classification (ESC) plays a vital role in the field of machine auditory scene. Recently, the Highway Network CNN model has achieved the state-of-art results via solving the vanishing-gradient problem of much deeper CNN. However, carefully analyzing the Highway Network model shows that the Highway Network model lacks ability to maximize information flow between layers, which is essentially benefits the discriminative representation of acoustic events. Besides, the Highway Network model size is larger than 20MB for ESC task, which is still large for mobile applications. Regarding to these two issues, in this study, we propose a novel Densely Connected Highway Convolutional Network (DCH-Net) model for ESC task. Specifically, a densely highway module is developed which is able to ensure the maximum information flow between layers by connecting all layers directly with each other. Besides, to reduce the model size, a global average pooling layer is designed which replaces the traditional fully connection layers and the parameters of the model is greatly reduced. Experimental results show that our DCH-Net ESC model achieves accuracy of 69% and 90% on ESC50 and ESCIO dataset respectively, which is 2% and 10% higher than that of Highway Network based Highway networks ESC model. Meanwhile our model size is only 2MB.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"100 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129933731","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631613
Z. Tian, Yong Li, Mu Zhou, Ze Li
Passive intrusion detection, which is an emerging technique to detect whether there exists any intruders in monitored area, is widely used in home security and smart home, etc. Up to now, various indoor fine-grained passive human intrusion detection systems using WiFi signals have been proposed. However, those existing detection systems mostly rely on elaborate off-line training process, which hampers fast deployment of wireless devices and also reduces system robustness. To response those problems, in this paper, we propose APID, a system for adaptive indoor passive intrusion detection, which enables adaptive, device-free human intrusion detection in indoor environments using channel state information (CSI) of WiFi signals. Firstly, APID evaluates dispersion of CSI amplitude, which is not affected by the mean amplitude. Secondly, APID extracts CSI amplitude dispersion ratio between two adjacent time windows as sensitive metrics for intrusion detection. Then, the hypothesis testing is utilized to achieve no-calibration human motion detection. Finally, we implement APID on the commodity WiFi devices and evaluate it in two typical indoor scenarios. The experimental results show that APID can achieve an average detection accuracy of more than 96%.
{"title":"WiFi-Based Adaptive Indoor Passive Intrusion Detection","authors":"Z. Tian, Yong Li, Mu Zhou, Ze Li","doi":"10.1109/ICDSP.2018.8631613","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631613","url":null,"abstract":"Passive intrusion detection, which is an emerging technique to detect whether there exists any intruders in monitored area, is widely used in home security and smart home, etc. Up to now, various indoor fine-grained passive human intrusion detection systems using WiFi signals have been proposed. However, those existing detection systems mostly rely on elaborate off-line training process, which hampers fast deployment of wireless devices and also reduces system robustness. To response those problems, in this paper, we propose APID, a system for adaptive indoor passive intrusion detection, which enables adaptive, device-free human intrusion detection in indoor environments using channel state information (CSI) of WiFi signals. Firstly, APID evaluates dispersion of CSI amplitude, which is not affected by the mean amplitude. Secondly, APID extracts CSI amplitude dispersion ratio between two adjacent time windows as sensitive metrics for intrusion detection. Then, the hypothesis testing is utilized to achieve no-calibration human motion detection. Finally, we implement APID on the commodity WiFi devices and evaluate it in two typical indoor scenarios. The experimental results show that APID can achieve an average detection accuracy of more than 96%.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128901287","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631794
Gustavo J. González, F. Gregorio, J. Cousseau
Narrowband internet of things (NB-IoT) considers the connection of thousands of devices to a single LTE base station (BS). To make possible the coexistence with classic LTE user equipements (UE)s, the BS allocates several IoT UEs into special physical resource blocks (PRB)s. These special PRBs reduce the IoT transmitter complexity but make the LTE signal interfere with the IoT PRBs. IoT nodes are in general low-cost and therefore prone to suffer from RF impairments. The LTE interference and the RF impairments compromise the performance of IoT nodes. In this paper, we analyze the coexistence of LTE and IoT in the multiple access uplink, considering RF impairments. We analyze the use of guard bands to reduce the interference from LTE in IoT. Also, we evaluate the allowable carrier frequency offset (CFO) and I/Q imbalance levels that ensures a reasonable system performance.
{"title":"Interference Analysis in the LTE and NB-IoT Uplink Multiple Access with RF impairments","authors":"Gustavo J. González, F. Gregorio, J. Cousseau","doi":"10.1109/ICDSP.2018.8631794","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631794","url":null,"abstract":"Narrowband internet of things (NB-IoT) considers the connection of thousands of devices to a single LTE base station (BS). To make possible the coexistence with classic LTE user equipements (UE)s, the BS allocates several IoT UEs into special physical resource blocks (PRB)s. These special PRBs reduce the IoT transmitter complexity but make the LTE signal interfere with the IoT PRBs. IoT nodes are in general low-cost and therefore prone to suffer from RF impairments. The LTE interference and the RF impairments compromise the performance of IoT nodes. In this paper, we analyze the coexistence of LTE and IoT in the multiple access uplink, considering RF impairments. We analyze the use of guard bands to reduce the interference from LTE in IoT. Also, we evaluate the allowable carrier frequency offset (CFO) and I/Q imbalance levels that ensures a reasonable system performance.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130527633","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631552
Chunrong Chen, Duo Zhao, Deqing Huang, Qichao Tang
This paper tends to use a novel perspective to suppress the lateral vibration of high-speed trains (HST), i.e., making use of the periodicity of lateral dynamics. First, the dynamics of a quarter-vehicle model are analysed and modelled. Next, a backstepping controller is designed to suppress the lateral vibration of car body based on a 3-degree-of-freedom (3-DOF) simulation model. And lateral ride comfort improvements are achieved by implementing such control strategy in comparison with passive system. Finally, under the framework of back-stepping design, a repetitive learning control (RLC) scheme is presented to reduce the lateral vibration by periodic tracking control. The learning convergence is proved rigorously in a Lyapunov way and the simulation results demonstrate the control superiority compared with the backstepping controller.
{"title":"An Improved Lateral Vibration Suppression Strategy of the High-speed Train Using Repetitive Learning Control","authors":"Chunrong Chen, Duo Zhao, Deqing Huang, Qichao Tang","doi":"10.1109/ICDSP.2018.8631552","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631552","url":null,"abstract":"This paper tends to use a novel perspective to suppress the lateral vibration of high-speed trains (HST), i.e., making use of the periodicity of lateral dynamics. First, the dynamics of a quarter-vehicle model are analysed and modelled. Next, a backstepping controller is designed to suppress the lateral vibration of car body based on a 3-degree-of-freedom (3-DOF) simulation model. And lateral ride comfort improvements are achieved by implementing such control strategy in comparison with passive system. Finally, under the framework of back-stepping design, a repetitive learning control (RLC) scheme is presented to reduce the lateral vibration by periodic tracking control. The learning convergence is proved rigorously in a Lyapunov way and the simulation results demonstrate the control superiority compared with the backstepping controller.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130580781","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631849
P. Shull, H. Xia
Harvesting energy during walking is a promising potential power source to decrease size or even eliminate batteries for wearable devices. While sliding shoes may offer a method for harvesting energy during gait, it is important to know their influence on the expected energy harvesting rate and metabolic cost rate. In this paper, we develop two multivariate linear regression models based on subject height, weight, and walking speed to predict energy harvesting rate and metabolic cost rate for walking with custom energy harvesting sliding shoes. Eight healthy subjects performed 200 meter overground walking trials at normal and fast speeds while wearing the custom sliding shoes to harvest energy and a portable gas analysis system to measure metabolic cost. The metabolic cost rate model performed well with only 6.9% error, while the energy harvesting rate model was less accurate with 29.9% error. Future research should focus on improving the models by adding additional features such as step frequency, speed of sliding and length of sliding to capture more of the variance. These findings could help to serve as a foundation to facilitate widespread adoption of wearable devices by reducing the required amount of onboard energy storage.
{"title":"Energy Harvesting Modeling and Prediction during Walking Gait for a Sliding Shoe","authors":"P. Shull, H. Xia","doi":"10.1109/ICDSP.2018.8631849","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631849","url":null,"abstract":"Harvesting energy during walking is a promising potential power source to decrease size or even eliminate batteries for wearable devices. While sliding shoes may offer a method for harvesting energy during gait, it is important to know their influence on the expected energy harvesting rate and metabolic cost rate. In this paper, we develop two multivariate linear regression models based on subject height, weight, and walking speed to predict energy harvesting rate and metabolic cost rate for walking with custom energy harvesting sliding shoes. Eight healthy subjects performed 200 meter overground walking trials at normal and fast speeds while wearing the custom sliding shoes to harvest energy and a portable gas analysis system to measure metabolic cost. The metabolic cost rate model performed well with only 6.9% error, while the energy harvesting rate model was less accurate with 29.9% error. Future research should focus on improving the models by adding additional features such as step frequency, speed of sliding and length of sliding to capture more of the variance. These findings could help to serve as a foundation to facilitate widespread adoption of wearable devices by reducing the required amount of onboard energy storage.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132200810","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}
In multitask learning (MTL) paradigm, modularity is an effective way to achieve component and parameter reuse as well as system extensibility. In this work, we introduce two enhanced modules named res-fire module (RF) and dimension reduction module(DR) to improve the performance of modular MTL network – PathNet. In addition, in order to further improve the transfer ability of the network, we apply learnable scale parameters to merge the outputs of the modules in the same layer and then scatter to the next layer. Experiments on MNIST, CIFAR, SVHN and MiniImageNet demonstrate that, with the similar scale as PathNet, our architecture achieves remarkable improvement in both transfer ability and expression ability. Our design used x5.23 fewer generations to achieve 99% accuracy on a source-to-target MNIST classification task compared with DeepMind’s PathNet. We also increase the accuracy of CIFARSVHN transfer task by x1.9. Also we get 70.75% accuracy on miniImageNet.
{"title":"Multitask Learning With Enhanced Modules","authors":"Zishuo Zheng, Yadong Wei, Zixu Zhao, Xindi Wu, Zhengcheng Li, Pengju Ren","doi":"10.1109/ICDSP.2018.8631696","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631696","url":null,"abstract":"In multitask learning (MTL) paradigm, modularity is an effective way to achieve component and parameter reuse as well as system extensibility. In this work, we introduce two enhanced modules named res-fire module (RF) and dimension reduction module(DR) to improve the performance of modular MTL network – PathNet. In addition, in order to further improve the transfer ability of the network, we apply learnable scale parameters to merge the outputs of the modules in the same layer and then scatter to the next layer. Experiments on MNIST, CIFAR, SVHN and MiniImageNet demonstrate that, with the similar scale as PathNet, our architecture achieves remarkable improvement in both transfer ability and expression ability. Our design used x5.23 fewer generations to achieve 99% accuracy on a source-to-target MNIST classification task compared with DeepMind’s PathNet. We also increase the accuracy of CIFARSVHN transfer task by x1.9. Also we get 70.75% accuracy on miniImageNet.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130963304","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631802
Xizhong Shen, Su Chenying
We examine spectral subtraction with both amplitude and phase spectra for improved speech enhancement performance by the method of maximum a posterior. Spectral subtraction is a very valid and direct denoising algorithm, but it has a vital problem, i.e., it may generate 'musical noise'. An adaptive harmonic model is utilized. Maximum a posterior is considered to derive the phase estimator, which is extra applied to amplitude spectral subtraction. Different from others, the extra parameters in our algorithm are considered as random variables, and the main extra parameters are amplitude and phase. The phase of the speech signal is assumed to have von Mises circular distribution, and the amplitude is to have normal distribution. The assumptions are applied to Bayesian theory, and we derived the update formulae of the parameters of the speech model, that is, phase estimator and amplitude estimator. Thus, we obtained the phase and amplitude of each harmonic. Simulation results show the further improvement of spectral subtraction.
{"title":"Speech Enhancement Exploiting Probabilistic Approach Using Maximum A Posterior","authors":"Xizhong Shen, Su Chenying","doi":"10.1109/ICDSP.2018.8631802","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631802","url":null,"abstract":"We examine spectral subtraction with both amplitude and phase spectra for improved speech enhancement performance by the method of maximum a posterior. Spectral subtraction is a very valid and direct denoising algorithm, but it has a vital problem, i.e., it may generate 'musical noise'. An adaptive harmonic model is utilized. Maximum a posterior is considered to derive the phase estimator, which is extra applied to amplitude spectral subtraction. Different from others, the extra parameters in our algorithm are considered as random variables, and the main extra parameters are amplitude and phase. The phase of the speech signal is assumed to have von Mises circular distribution, and the amplitude is to have normal distribution. The assumptions are applied to Bayesian theory, and we derived the update formulae of the parameters of the speech model, that is, phase estimator and amplitude estimator. Thus, we obtained the phase and amplitude of each harmonic. Simulation results show the further improvement of spectral subtraction.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127860380","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631683
Chien-An Wang, Sheng-Jui Huang, Yu-Cheng Li, Yi-Chang Lu
In this paper, we implement a Liquid Association calculator on an Altera Stratix V FPGA. Using data from high throughput microarrays, the hardware is capable of analyzing whether the presence of a third gene can affect the correlation between the existing two genes. The runtime and power consumption of our FPGA implementation is only 45.5% and 8.14%, respectively, of those required by the GPU version.
在本文中，我们在Altera Stratix V FPGA上实现了一个液体关联计算器。利用来自高通量微阵列的数据，该硬件能够分析第三个基因的存在是否会影响现有两个基因之间的相关性。FPGA实现的运行时间和功耗分别仅为GPU版本所需的45.5%和8.14%。
{"title":"An FPGA-Based Liquid Association Calculator for Genome-Wide Co-Expression Analysis","authors":"Chien-An Wang, Sheng-Jui Huang, Yu-Cheng Li, Yi-Chang Lu","doi":"10.1109/ICDSP.2018.8631683","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631683","url":null,"abstract":"In this paper, we implement a Liquid Association calculator on an Altera Stratix V FPGA. Using data from high throughput microarrays, the hardware is capable of analyzing whether the presence of a third gene can affect the correlation between the existing two genes. The runtime and power consumption of our FPGA implementation is only 45.5% and 8.14%, respectively, of those required by the GPU version.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128803375","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 : 2018-11-01DOI: 10.1109/ICDSP.2018.8631630
Zhenyue Zhang, Guan Gui, Yan Liang
In the millimeter-wave (mmWave) massive MIMO system, the accuracy of channel estimation directly affects the performance of precoding at the transmitter and detection at the receiver. Hence, it is very important to obtain accurate channel state information (CSI). Considering the channel sparsity of mmWave massive MIMO with hybrid precoding, this paper proposes a ℓ_{1/2}-regularization based sparse channel estimation scheme. The basic idea of the proposed method is to formulate the sparse channel estimation to a compressed sensing problem. Specifically, the scheme firstly constructs an objective function, which is a weighted sum of the ℓ_{1/2}-regularization and the data fitting error. Then optimizes it by means of the gradient descent method iteratively and the weight parameter in the function is also updated each time. Different from the conventional schemes, our proposed scheme can avoid the quantization error and finally achieve super-resolution performance. Simulation results verify that the proposed algorithm can achieve better performance than some recently proposed algorithms.
{"title":"ℓ1/2-Regularization Based Sparse Channel Estimation for MmWave Massive MIMO Systems","authors":"Zhenyue Zhang, Guan Gui, Yan Liang","doi":"10.1109/ICDSP.2018.8631630","DOIUrl":"https://doi.org/10.1109/ICDSP.2018.8631630","url":null,"abstract":"In the millimeter-wave (mmWave) massive MIMO system, the accuracy of channel estimation directly affects the performance of precoding at the transmitter and detection at the receiver. Hence, it is very important to obtain accurate channel state information (CSI). Considering the channel sparsity of mmWave massive MIMO with hybrid precoding, this paper proposes a ℓ_{1/2}-regularization based sparse channel estimation scheme. The basic idea of the proposed method is to formulate the sparse channel estimation to a compressed sensing problem. Specifically, the scheme firstly constructs an objective function, which is a weighted sum of the ℓ_{1/2}-regularization and the data fitting error. Then optimizes it by means of the gradient descent method iteratively and the weight parameter in the function is also updated each time. Different from the conventional schemes, our proposed scheme can avoid the quantization error and finally achieve super-resolution performance. Simulation results verify that the proposed algorithm can achieve better performance than some recently proposed algorithms.","PeriodicalId":218806,"journal":{"name":"2018 IEEE 23rd International Conference on Digital Signal Processing (DSP)","volume":"327 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125426854","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：