As the demands of massive connections and vast coverage rapidly grow in the next wireless communication networks, rate splitting multiple access(RSMA) is considered to be the new promising access scheme since it can p...As the demands of massive connections and vast coverage rapidly grow in the next wireless communication networks, rate splitting multiple access(RSMA) is considered to be the new promising access scheme since it can provide higher efficiency with limited spectrum resources. In this paper, combining spectrum splitting with rate splitting, we propose to allocate resources with traffic offloading in hybrid satellite terrestrial networks. A novel deep reinforcement learning method is adopted to solve this challenging non-convex problem. However, the neverending learning process could prohibit its practical implementation. Therefore, we introduce the switch mechanism to avoid unnecessary learning. Additionally, the QoS constraint in the scheme can rule out unsuccessful transmission. The simulation results validates the energy efficiency performance and the convergence speed of the proposed algorithm.展开更多
High frequency(HF) communication is widely spread due to some merits like easy deployment and wide communication coverage. Spectrum prediction is a promising technique to facilitate the working frequency selection and...High frequency(HF) communication is widely spread due to some merits like easy deployment and wide communication coverage. Spectrum prediction is a promising technique to facilitate the working frequency selection and enhance the function of automatic link establishment. Most of the existing spectrum prediction algorithms focus on predicting spectrum values in a slot-by-slot manner and therefore are lack of timeliness. Deep learning based spectrum prediction is developed in this paper by simultaneously predicting multi-slot ahead states of multiple spectrum points within a period of time. Specifically, we first employ supervised learning and construct samples depending on longterm and short-term HF spectrum data. Then, advanced residual units are introduced to build multiple residual network modules to respectively capture characteristics in these data with diverse time scales. Further, convolution neural network fuses the outputs of residual network modules above for temporal-spectral prediction, which is combined with residual network modules to construct the deep temporal-spectral residual network. Experiments have demonstrated that the approach proposed in this paper has a significant advantage over the benchmark schemes.展开更多
Deterministic inversion based on deep learning has been widely utilized in model parameters estimation.Constrained by logging data,seismic data,wavelet and modeling operator,deterministic inversion based on deep learn...Deterministic inversion based on deep learning has been widely utilized in model parameters estimation.Constrained by logging data,seismic data,wavelet and modeling operator,deterministic inversion based on deep learning can establish nonlinear relationships between seismic data and model parameters.However,seismic data lacks low-frequency and contains noise,which increases the non-uniqueness of the solutions.The conventional inversion method based on deep learning can only establish the deterministic relationship between seismic data and parameters,and cannot quantify the uncertainty of inversion.In order to quickly quantify the uncertainty,a physics-guided deep mixture density network(PG-DMDN)is established by combining the mixture density network(MDN)with the deep neural network(DNN).Compared with Bayesian neural network(BNN)and network dropout,PG-DMDN has lower computing cost and shorter training time.A low-frequency model is introduced in the training process of the network to help the network learn the nonlinear relationship between narrowband seismic data and low-frequency impedance.In addition,the block constraints are added to the PG-DMDN framework to improve the horizontal continuity of the inversion results.To illustrate the benefits of proposed method,the PG-DMDN is compared with existing semi-supervised inversion method.Four synthetic data examples of Marmousi II model are utilized to quantify the influence of forward modeling part,low-frequency model,noise and the pseudo-wells number on inversion results,and prove the feasibility and stability of the proposed method.In addition,the robustness and generality of the proposed method are verified by the field seismic data.展开更多
The vector vortex beam(VVB)has attracted significant attention due to its intrinsic diversity of information and has found great applications in both classical and quantum communications.However,a VVB is unavoidably a...The vector vortex beam(VVB)has attracted significant attention due to its intrinsic diversity of information and has found great applications in both classical and quantum communications.However,a VVB is unavoidably affected by atmospheric turbulence(AT)when it propagates through the free-space optical communication environment,which results in detection errors at the receiver.In this paper,we propose a VVB classification scheme to detect VVBs with continuously changing polarization states under AT,where a diffractive deep neural network(DDNN)is designed and trained to classify the intensity distribution of the input distorted VVBs,and the horizontal direction of polarization of the input distorted beam is adopted as the feature for the classification through the DDNN.The numerical simulations and experimental results demonstrate that the proposed scheme has high accuracy in classification tasks.The energy distribution percentage remains above 95%from weak to medium AT,and the classification accuracy can remain above 95%for various strengths of turbulence.It has a faster convergence and better accuracy than that based on a convolutional neural network.展开更多
In this paper,we propose the Two-way Deep Reinforcement Learning(DRL)-Based resource allocation algorithm,which solves the problem of resource allocation in the cognitive downlink network based on the underlay mode.Se...In this paper,we propose the Two-way Deep Reinforcement Learning(DRL)-Based resource allocation algorithm,which solves the problem of resource allocation in the cognitive downlink network based on the underlay mode.Secondary users(SUs)in the cognitive network are multiplexed by a new Power Domain Sparse Code Multiple Access(PD-SCMA)scheme,and the physical resources of the cognitive base station are virtualized into two types of slices:enhanced mobile broadband(eMBB)slice and ultrareliable low latency communication(URLLC)slice.We design the Double Deep Q Network(DDQN)network output the optimal codebook assignment scheme and simultaneously use the Deep Deterministic Policy Gradient(DDPG)network output the optimal power allocation scheme.The objective is to jointly optimize the spectral efficiency of the system and the Quality of Service(QoS)of SUs.Simulation results show that the proposed algorithm outperforms the CNDDQN algorithm and modified JEERA algorithm in terms of spectral efficiency and QoS satisfaction.Additionally,compared with the Power Domain Non-orthogonal Multiple Access(PD-NOMA)slices and the Sparse Code Multiple Access(SCMA)slices,the PD-SCMA slices can dramatically enhance spectral efficiency and increase the number of accessible users.展开更多
In view of low recognition rate of complex radar intra-pulse modulation signal type by traditional methods under low signal-to-noise ratio(SNR),the paper proposes an automatic recog-nition method of complex radar intr...In view of low recognition rate of complex radar intra-pulse modulation signal type by traditional methods under low signal-to-noise ratio(SNR),the paper proposes an automatic recog-nition method of complex radar intra-pulse modulation signal type based on deep residual network.The basic principle of the recognition method is to obtain the transformation relationship between the time and frequency of complex radar intra-pulse modulation signal through short-time Fourier transform(STFT),and then design an appropriate deep residual network to extract the features of the time-frequency map and complete a variety of complex intra-pulse modulation signal type recognition.In addition,in order to improve the generalization ability of the proposed method,label smoothing and L2 regularization are introduced.The simulation results show that the proposed method has a recognition accuracy of more than 95%for complex radar intra-pulse modulation sig-nal types under low SNR(2 dB).展开更多
Blind image quality assessment(BIQA)is of fundamental importance in low-level computer vision community.Increasing interest has been drawn in exploiting deep neural networks for BIQA.Despite of the notable success ach...Blind image quality assessment(BIQA)is of fundamental importance in low-level computer vision community.Increasing interest has been drawn in exploiting deep neural networks for BIQA.Despite of the notable success achieved,there is a broad consensus that training deep convolutional neural networks(DCNN)heavily relies on massive annotated data.Unfortunately,BIQA is typically a small sample problem,resulting the generalization ability of BIQA severely restricted.In order to improve the accuracy and generalization ability of BIQA metrics,this work proposed a totally opinion-unaware BIQA in which no subjective annotations are involved in the training stage.Multiple full-reference image quality assessment(FR-IQA)metrics are employed to label the distorted image as a substitution of subjective quality annotation.A deep neural network(DNN)is trained to blindly predict the multiple FR-IQA score in absence of corresponding pristine image.In the end,a selfsupervised FR-IQA score aggregator implemented by adversarial auto-encoder pools the predictions of multiple FR-IQA scores into the final quality predicting score.Even though none of subjective scores are involved in the training stage,experimental results indicate that our proposed full reference induced BIQA framework is as competitive as state-of-the-art BIQA metrics.展开更多
The recent surge of mobile subscribers and user data traffic has accelerated the telecommunication sector towards the adoption of the fifth-generation (5G) mobile networks. Cloud radio access network (CRAN) is a promi...The recent surge of mobile subscribers and user data traffic has accelerated the telecommunication sector towards the adoption of the fifth-generation (5G) mobile networks. Cloud radio access network (CRAN) is a prominent framework in the 5G mobile network to meet the above requirements by deploying low-cost and intelligent multiple distributed antennas known as remote radio heads (RRHs). However, achieving the optimal resource allocation (RA) in CRAN using the traditional approach is still challenging due to the complex structure. In this paper, we introduce the convolutional neural network-based deep Q-network (CNN-DQN) to balance the energy consumption and guarantee the user quality of service (QoS) demand in downlink CRAN. We first formulate the Markov decision process (MDP) for energy efficiency (EE) and build up a 3-layer CNN to capture the environment feature as an input state space. We then use DQN to turn on/off the RRHs dynamically based on the user QoS demand and energy consumption in the CRAN. Finally, we solve the RA problem based on the user constraint and transmit power to guarantee the user QoS demand and maximize the EE with a minimum number of active RRHs. In the end, we conduct the simulation to compare our proposed scheme with nature DQN and the traditional approach.展开更多
To support dramatically increased traffic loads,communication networks become ultra-dense.Traditional cell association(CA)schemes are timeconsuming,forcing researchers to seek fast schemes.This paper proposes a deep Q...To support dramatically increased traffic loads,communication networks become ultra-dense.Traditional cell association(CA)schemes are timeconsuming,forcing researchers to seek fast schemes.This paper proposes a deep Q-learning based scheme,whose main idea is to train a deep neural network(DNN)to calculate the Q values of all the state-action pairs and the cell holding the maximum Q value is associated.In the training stage,the intelligent agent continuously generates samples through the trial-anderror method to train the DNN until convergence.In the application stage,state vectors of all the users are inputted to the trained DNN to quickly obtain a satisfied CA result of a scenario with the same BS locations and user distribution.Simulations demonstrate that the proposed scheme provides satisfied CA results in a computational time several orders of magnitudes shorter than traditional schemes.Meanwhile,performance metrics,such as capacity and fairness,can be guaranteed.展开更多
Infrared image recognition plays an important role in the inspection of power equipment.Existing technologies dedicated to this purpose often require manually selected features,which are not transferable and interpret...Infrared image recognition plays an important role in the inspection of power equipment.Existing technologies dedicated to this purpose often require manually selected features,which are not transferable and interpretable,and have limited training data.To address these limitations,this paper proposes an automatic infrared image recognition framework,which includes an object recognition module based on a deep self-attention network and a temperature distribution identification module based on a multi-factor similarity calculation.First,the features of an input image are extracted and embedded using a multi-head attention encoding-decoding mechanism.Thereafter,the embedded features are used to predict the equipment component category and location.In the located area,preliminary segmentation is performed.Finally,similar areas are gradually merged,and the temperature distribution of the equipment is obtained to identify a fault.Our experiments indicate that the proposed method demonstrates significantly improved accuracy compared with other related methods and,hence,provides a good reference for the automation of power equipment inspection.展开更多
Deep neural networks(DNN)are widely used in image recognition,image classification,and other fields.However,as the model size increases,the DNN hardware accelerators face the challenge of higher area overhead and ener...Deep neural networks(DNN)are widely used in image recognition,image classification,and other fields.However,as the model size increases,the DNN hardware accelerators face the challenge of higher area overhead and energy consumption.In recent years,stochastic computing(SC)has been considered a way to realize deep neural networks and reduce hardware consumption.A probabilistic compensation algorithm is proposed to solve the accuracy problem of stochastic calculation,and a fully parallel neural network accelerator based on a deterministic method is designed.The software simulation results show that the accuracy of the probability compensation algorithm on the CIFAR-10 data set is 95.32%,which is 14.98%higher than that of the traditional SC algorithm.The accuracy of the deterministic algorithm on the CIFAR-10 dataset is 95.06%,which is 14.72%higher than that of the traditional SC algorithm.The results of Very Large Scale Integration Circuit(VLSI)hardware tests show that the normalized energy efficiency of the fully parallel neural network accelerator based on the deterministic method is improved by 31%compared with the circuit based on binary computing.展开更多
A gait control method for a biped robot based on the deep Q-network (DQN) algorithm is proposed to enhance the stability of walking on uneven ground. This control strategy is an intelligent learning method of posture ...A gait control method for a biped robot based on the deep Q-network (DQN) algorithm is proposed to enhance the stability of walking on uneven ground. This control strategy is an intelligent learning method of posture adjustment. A robot is taken as an agent and trained to walk steadily on an uneven surface with obstacles, using a simple reward function based on forward progress. The reward-punishment (RP) mechanism of the DQN algorithm is established after obtaining the offline gait which was generated in advance foot trajectory planning. Instead of implementing a complex dynamic model, the proposed method enables the biped robot to learn to adjust its posture on the uneven ground and ensures walking stability. The performance and effectiveness of the proposed algorithm was validated in the V-REP simulation environment. The results demonstrate that the biped robot's lateral tile angle is less than 3° after implementing the proposed method and the walking stability is obviously improved.展开更多
The low Earth orbit(LEO)satellite networks have outstanding advantages such as wide coverage area and not being limited by geographic environment,which can provide a broader range of communication services and has bec...The low Earth orbit(LEO)satellite networks have outstanding advantages such as wide coverage area and not being limited by geographic environment,which can provide a broader range of communication services and has become an essential supplement to the terrestrial network.However,the dynamic changes and uneven distribution of satellite network traffic inevitably bring challenges to multipath routing.Even worse,the harsh space environment often leads to incomplete collection of network state data for routing decision-making,which further complicates this challenge.To address this problem,this paper proposes a state-incomplete intelligent dynamic multipath routing algorithm(SIDMRA)to maximize network efficiency even with incomplete state data as input.Specifically,we model the multipath routing problem as a markov decision process(MDP)and then combine the deep deterministic policy gradient(DDPG)and the K shortest paths(KSP)algorithm to solve the optimal multipath routing policy.We use the temporal correlation of the satellite network state to fit the incomplete state data and then use the message passing neuron network(MPNN)for data enhancement.Simulation results show that the proposed algorithm outperforms baseline algorithms regarding average end-to-end delay and packet loss rate and performs stably under certain missing rates of state data.展开更多
Track reconstruction algorithms are critical for polarization measurements.Convolutional neural networks(CNNs)are a promising alternative to traditional moment-based track reconstruction approaches.However,the hexagon...Track reconstruction algorithms are critical for polarization measurements.Convolutional neural networks(CNNs)are a promising alternative to traditional moment-based track reconstruction approaches.However,the hexagonal grid track images obtained using gas pixel detectors(GPDs)for better anisotropy do not match the classical rectangle-based CNN,and converting the track images from hexagonal to square results in a loss of information.We developed a new hexagonal CNN algorithm for track reconstruction and polarization estimation in X-ray polarimeters,which was used to extract the emission angles and absorption points from photoelectron track images and predict the uncer-tainty of the predicted emission angles.The simulated data from the PolarLight test were used to train and test the hexagonal CNN models.For individual energies,the hexagonal CNN algorithm produced 15%-30%improvements in the modulation factor compared to the moment analysis method for 100%polarized data,and its performance was comparable to that of the rectangle-based CNN algorithm that was recently developed by the Imaging X-ray Polarimetry Explorer team,but at a lower computational and storage cost for preprocessing.展开更多
In this study,an end-to-end deep learning method is proposed to improve the accuracy of continuum estimation in low-resolution gamma-ray spectra.A novel process for generating the theoretical continuum of a simulated ...In this study,an end-to-end deep learning method is proposed to improve the accuracy of continuum estimation in low-resolution gamma-ray spectra.A novel process for generating the theoretical continuum of a simulated spectrum is established,and a convolutional neural network consisting of 51 layers and more than 105 parameters is constructed to directly predict the entire continuum from the extracted global spectrum features.For testing,an in-house NaI-type whole-body counter is used,and 106 training spectrum samples(20%of which are reserved for testing)are generated using Monte Carlo simulations.In addition,the existing fitting,step-type,and peak erosion methods are selected for comparison.The proposed method exhibits excellent performance,as evidenced by its activity error distribution and the smallest mean activity error of 1.5%among the evaluated methods.Additionally,a validation experiment is performed using a whole-body counter to analyze a human physical phantom containing four radionuclides.The largest activity error of the proposed method is−5.1%,which is considerably smaller than those of the comparative methods,confirming the test results.The multiscale feature extraction and nonlinear relation modeling in the proposed method establish a novel approach for accurate and convenient continuum estimation in a low-resolution gamma-ray spectrum.Thus,the proposed method is promising for accurate quantitative radioactivity analysis in practical applications.展开更多
文摘As the demands of massive connections and vast coverage rapidly grow in the next wireless communication networks, rate splitting multiple access(RSMA) is considered to be the new promising access scheme since it can provide higher efficiency with limited spectrum resources. In this paper, combining spectrum splitting with rate splitting, we propose to allocate resources with traffic offloading in hybrid satellite terrestrial networks. A novel deep reinforcement learning method is adopted to solve this challenging non-convex problem. However, the neverending learning process could prohibit its practical implementation. Therefore, we introduce the switch mechanism to avoid unnecessary learning. Additionally, the QoS constraint in the scheme can rule out unsuccessful transmission. The simulation results validates the energy efficiency performance and the convergence speed of the proposed algorithm.
基金supported in part by the National Natural Science Foundation of China (Grants No. 61501510 and No. 61631020)Natural Science Foundation of Jiangsu Province (Grant No. BK20150717)+2 种基金China Postdoctoral Science Foundation Funded Project (Grant No. 2016M590398 and No.2018T110426)Jiangsu Planned Projects for Postdoctoral Research Funds (Grant No. 1501009A)Natural Science Foundation for Distinguished Young Scholars of Jiangsu Province (Grant No. BK20160034)
文摘High frequency(HF) communication is widely spread due to some merits like easy deployment and wide communication coverage. Spectrum prediction is a promising technique to facilitate the working frequency selection and enhance the function of automatic link establishment. Most of the existing spectrum prediction algorithms focus on predicting spectrum values in a slot-by-slot manner and therefore are lack of timeliness. Deep learning based spectrum prediction is developed in this paper by simultaneously predicting multi-slot ahead states of multiple spectrum points within a period of time. Specifically, we first employ supervised learning and construct samples depending on longterm and short-term HF spectrum data. Then, advanced residual units are introduced to build multiple residual network modules to respectively capture characteristics in these data with diverse time scales. Further, convolution neural network fuses the outputs of residual network modules above for temporal-spectral prediction, which is combined with residual network modules to construct the deep temporal-spectral residual network. Experiments have demonstrated that the approach proposed in this paper has a significant advantage over the benchmark schemes.
基金the sponsorship of Shandong Province Foundation for Laoshan National Laboratory of Science and Technology Foundation(LSKJ202203400)National Natural Science Foundation of China(42174139,42030103)Science Foundation from Innovation and Technology Support Program for Young Scientists in Colleges of Shandong Province and Ministry of Science and Technology of China(2019RA2136)。
文摘Deterministic inversion based on deep learning has been widely utilized in model parameters estimation.Constrained by logging data,seismic data,wavelet and modeling operator,deterministic inversion based on deep learning can establish nonlinear relationships between seismic data and model parameters.However,seismic data lacks low-frequency and contains noise,which increases the non-uniqueness of the solutions.The conventional inversion method based on deep learning can only establish the deterministic relationship between seismic data and parameters,and cannot quantify the uncertainty of inversion.In order to quickly quantify the uncertainty,a physics-guided deep mixture density network(PG-DMDN)is established by combining the mixture density network(MDN)with the deep neural network(DNN).Compared with Bayesian neural network(BNN)and network dropout,PG-DMDN has lower computing cost and shorter training time.A low-frequency model is introduced in the training process of the network to help the network learn the nonlinear relationship between narrowband seismic data and low-frequency impedance.In addition,the block constraints are added to the PG-DMDN framework to improve the horizontal continuity of the inversion results.To illustrate the benefits of proposed method,the PG-DMDN is compared with existing semi-supervised inversion method.Four synthetic data examples of Marmousi II model are utilized to quantify the influence of forward modeling part,low-frequency model,noise and the pseudo-wells number on inversion results,and prove the feasibility and stability of the proposed method.In addition,the robustness and generality of the proposed method are verified by the field seismic data.
基金Project supported by the National Natural Science Foundation of China(Grant Nos.62375140 and 62001249)the Open Research Fund of National Laboratory of Solid State Microstructures(Grant No.M36055).
文摘The vector vortex beam(VVB)has attracted significant attention due to its intrinsic diversity of information and has found great applications in both classical and quantum communications.However,a VVB is unavoidably affected by atmospheric turbulence(AT)when it propagates through the free-space optical communication environment,which results in detection errors at the receiver.In this paper,we propose a VVB classification scheme to detect VVBs with continuously changing polarization states under AT,where a diffractive deep neural network(DDNN)is designed and trained to classify the intensity distribution of the input distorted VVBs,and the horizontal direction of polarization of the input distorted beam is adopted as the feature for the classification through the DDNN.The numerical simulations and experimental results demonstrate that the proposed scheme has high accuracy in classification tasks.The energy distribution percentage remains above 95%from weak to medium AT,and the classification accuracy can remain above 95%for various strengths of turbulence.It has a faster convergence and better accuracy than that based on a convolutional neural network.
基金supported by the National Natural Science Foundation of China(Grant No.61971057).
文摘In this paper,we propose the Two-way Deep Reinforcement Learning(DRL)-Based resource allocation algorithm,which solves the problem of resource allocation in the cognitive downlink network based on the underlay mode.Secondary users(SUs)in the cognitive network are multiplexed by a new Power Domain Sparse Code Multiple Access(PD-SCMA)scheme,and the physical resources of the cognitive base station are virtualized into two types of slices:enhanced mobile broadband(eMBB)slice and ultrareliable low latency communication(URLLC)slice.We design the Double Deep Q Network(DDQN)network output the optimal codebook assignment scheme and simultaneously use the Deep Deterministic Policy Gradient(DDPG)network output the optimal power allocation scheme.The objective is to jointly optimize the spectral efficiency of the system and the Quality of Service(QoS)of SUs.Simulation results show that the proposed algorithm outperforms the CNDDQN algorithm and modified JEERA algorithm in terms of spectral efficiency and QoS satisfaction.Additionally,compared with the Power Domain Non-orthogonal Multiple Access(PD-NOMA)slices and the Sparse Code Multiple Access(SCMA)slices,the PD-SCMA slices can dramatically enhance spectral efficiency and increase the number of accessible users.
文摘In view of low recognition rate of complex radar intra-pulse modulation signal type by traditional methods under low signal-to-noise ratio(SNR),the paper proposes an automatic recog-nition method of complex radar intra-pulse modulation signal type based on deep residual network.The basic principle of the recognition method is to obtain the transformation relationship between the time and frequency of complex radar intra-pulse modulation signal through short-time Fourier transform(STFT),and then design an appropriate deep residual network to extract the features of the time-frequency map and complete a variety of complex intra-pulse modulation signal type recognition.In addition,in order to improve the generalization ability of the proposed method,label smoothing and L2 regularization are introduced.The simulation results show that the proposed method has a recognition accuracy of more than 95%for complex radar intra-pulse modulation sig-nal types under low SNR(2 dB).
基金supported by the Public Welfare Technology Application Research Project of Zhejiang Province,China(No.LGF21F010001)the Key Research and Development Program of Zhejiang Province,China(Grant No.2019C01002)the Key Research and Development Program of Zhejiang Province,China(Grant No.2021C03138)。
文摘Blind image quality assessment(BIQA)is of fundamental importance in low-level computer vision community.Increasing interest has been drawn in exploiting deep neural networks for BIQA.Despite of the notable success achieved,there is a broad consensus that training deep convolutional neural networks(DCNN)heavily relies on massive annotated data.Unfortunately,BIQA is typically a small sample problem,resulting the generalization ability of BIQA severely restricted.In order to improve the accuracy and generalization ability of BIQA metrics,this work proposed a totally opinion-unaware BIQA in which no subjective annotations are involved in the training stage.Multiple full-reference image quality assessment(FR-IQA)metrics are employed to label the distorted image as a substitution of subjective quality annotation.A deep neural network(DNN)is trained to blindly predict the multiple FR-IQA score in absence of corresponding pristine image.In the end,a selfsupervised FR-IQA score aggregator implemented by adversarial auto-encoder pools the predictions of multiple FR-IQA scores into the final quality predicting score.Even though none of subjective scores are involved in the training stage,experimental results indicate that our proposed full reference induced BIQA framework is as competitive as state-of-the-art BIQA metrics.
基金supported by the Universiti Tunku Abdul Rahman (UTAR) Malaysia under UTARRF (IPSR/RMC/UTARRF/2021-C1/T05)
文摘The recent surge of mobile subscribers and user data traffic has accelerated the telecommunication sector towards the adoption of the fifth-generation (5G) mobile networks. Cloud radio access network (CRAN) is a prominent framework in the 5G mobile network to meet the above requirements by deploying low-cost and intelligent multiple distributed antennas known as remote radio heads (RRHs). However, achieving the optimal resource allocation (RA) in CRAN using the traditional approach is still challenging due to the complex structure. In this paper, we introduce the convolutional neural network-based deep Q-network (CNN-DQN) to balance the energy consumption and guarantee the user quality of service (QoS) demand in downlink CRAN. We first formulate the Markov decision process (MDP) for energy efficiency (EE) and build up a 3-layer CNN to capture the environment feature as an input state space. We then use DQN to turn on/off the RRHs dynamically based on the user QoS demand and energy consumption in the CRAN. Finally, we solve the RA problem based on the user constraint and transmit power to guarantee the user QoS demand and maximize the EE with a minimum number of active RRHs. In the end, we conduct the simulation to compare our proposed scheme with nature DQN and the traditional approach.
基金This work was supported by the Fundamental Research Funds for the Central Universities of China under grant no.PA2019GDQT0012by National Natural Science Foundation of China(Grant No.61971176)by the Applied Basic Research Program ofWuhan City,China,under grand 2017010201010117.
文摘To support dramatically increased traffic loads,communication networks become ultra-dense.Traditional cell association(CA)schemes are timeconsuming,forcing researchers to seek fast schemes.This paper proposes a deep Q-learning based scheme,whose main idea is to train a deep neural network(DNN)to calculate the Q values of all the state-action pairs and the cell holding the maximum Q value is associated.In the training stage,the intelligent agent continuously generates samples through the trial-anderror method to train the DNN until convergence.In the application stage,state vectors of all the users are inputted to the trained DNN to quickly obtain a satisfied CA result of a scenario with the same BS locations and user distribution.Simulations demonstrate that the proposed scheme provides satisfied CA results in a computational time several orders of magnitudes shorter than traditional schemes.Meanwhile,performance metrics,such as capacity and fairness,can be guaranteed.
基金This work was supported by National Key R&D Program of China(2019YFE0102900).
文摘Infrared image recognition plays an important role in the inspection of power equipment.Existing technologies dedicated to this purpose often require manually selected features,which are not transferable and interpretable,and have limited training data.To address these limitations,this paper proposes an automatic infrared image recognition framework,which includes an object recognition module based on a deep self-attention network and a temperature distribution identification module based on a multi-factor similarity calculation.First,the features of an input image are extracted and embedded using a multi-head attention encoding-decoding mechanism.Thereafter,the embedded features are used to predict the equipment component category and location.In the located area,preliminary segmentation is performed.Finally,similar areas are gradually merged,and the temperature distribution of the equipment is obtained to identify a fault.Our experiments indicate that the proposed method demonstrates significantly improved accuracy compared with other related methods and,hence,provides a good reference for the automation of power equipment inspection.
文摘Deep neural networks(DNN)are widely used in image recognition,image classification,and other fields.However,as the model size increases,the DNN hardware accelerators face the challenge of higher area overhead and energy consumption.In recent years,stochastic computing(SC)has been considered a way to realize deep neural networks and reduce hardware consumption.A probabilistic compensation algorithm is proposed to solve the accuracy problem of stochastic calculation,and a fully parallel neural network accelerator based on a deterministic method is designed.The software simulation results show that the accuracy of the probability compensation algorithm on the CIFAR-10 data set is 95.32%,which is 14.98%higher than that of the traditional SC algorithm.The accuracy of the deterministic algorithm on the CIFAR-10 dataset is 95.06%,which is 14.72%higher than that of the traditional SC algorithm.The results of Very Large Scale Integration Circuit(VLSI)hardware tests show that the normalized energy efficiency of the fully parallel neural network accelerator based on the deterministic method is improved by 31%compared with the circuit based on binary computing.
基金Supported by the National Ministries and Research Funds(3020020221111)
文摘A gait control method for a biped robot based on the deep Q-network (DQN) algorithm is proposed to enhance the stability of walking on uneven ground. This control strategy is an intelligent learning method of posture adjustment. A robot is taken as an agent and trained to walk steadily on an uneven surface with obstacles, using a simple reward function based on forward progress. The reward-punishment (RP) mechanism of the DQN algorithm is established after obtaining the offline gait which was generated in advance foot trajectory planning. Instead of implementing a complex dynamic model, the proposed method enables the biped robot to learn to adjust its posture on the uneven ground and ensures walking stability. The performance and effectiveness of the proposed algorithm was validated in the V-REP simulation environment. The results demonstrate that the biped robot's lateral tile angle is less than 3° after implementing the proposed method and the walking stability is obviously improved.
文摘The low Earth orbit(LEO)satellite networks have outstanding advantages such as wide coverage area and not being limited by geographic environment,which can provide a broader range of communication services and has become an essential supplement to the terrestrial network.However,the dynamic changes and uneven distribution of satellite network traffic inevitably bring challenges to multipath routing.Even worse,the harsh space environment often leads to incomplete collection of network state data for routing decision-making,which further complicates this challenge.To address this problem,this paper proposes a state-incomplete intelligent dynamic multipath routing algorithm(SIDMRA)to maximize network efficiency even with incomplete state data as input.Specifically,we model the multipath routing problem as a markov decision process(MDP)and then combine the deep deterministic policy gradient(DDPG)and the K shortest paths(KSP)algorithm to solve the optimal multipath routing policy.We use the temporal correlation of the satellite network state to fit the incomplete state data and then use the message passing neuron network(MPNN)for data enhancement.Simulation results show that the proposed algorithm outperforms baseline algorithms regarding average end-to-end delay and packet loss rate and performs stably under certain missing rates of state data.
基金supported by the National Natural Science Foundation of China(No.12025301)the Tsinghua University Initiative Scientific Research Program.
文摘Track reconstruction algorithms are critical for polarization measurements.Convolutional neural networks(CNNs)are a promising alternative to traditional moment-based track reconstruction approaches.However,the hexagonal grid track images obtained using gas pixel detectors(GPDs)for better anisotropy do not match the classical rectangle-based CNN,and converting the track images from hexagonal to square results in a loss of information.We developed a new hexagonal CNN algorithm for track reconstruction and polarization estimation in X-ray polarimeters,which was used to extract the emission angles and absorption points from photoelectron track images and predict the uncer-tainty of the predicted emission angles.The simulated data from the PolarLight test were used to train and test the hexagonal CNN models.For individual energies,the hexagonal CNN algorithm produced 15%-30%improvements in the modulation factor compared to the moment analysis method for 100%polarized data,and its performance was comparable to that of the rectangle-based CNN algorithm that was recently developed by the Imaging X-ray Polarimetry Explorer team,but at a lower computational and storage cost for preprocessing.
基金supported by the National Natural Science Foundation of China(No.12005198).
文摘In this study,an end-to-end deep learning method is proposed to improve the accuracy of continuum estimation in low-resolution gamma-ray spectra.A novel process for generating the theoretical continuum of a simulated spectrum is established,and a convolutional neural network consisting of 51 layers and more than 105 parameters is constructed to directly predict the entire continuum from the extracted global spectrum features.For testing,an in-house NaI-type whole-body counter is used,and 106 training spectrum samples(20%of which are reserved for testing)are generated using Monte Carlo simulations.In addition,the existing fitting,step-type,and peak erosion methods are selected for comparison.The proposed method exhibits excellent performance,as evidenced by its activity error distribution and the smallest mean activity error of 1.5%among the evaluated methods.Additionally,a validation experiment is performed using a whole-body counter to analyze a human physical phantom containing four radionuclides.The largest activity error of the proposed method is−5.1%,which is considerably smaller than those of the comparative methods,confirming the test results.The multiscale feature extraction and nonlinear relation modeling in the proposed method establish a novel approach for accurate and convenient continuum estimation in a low-resolution gamma-ray spectrum.Thus,the proposed method is promising for accurate quantitative radioactivity analysis in practical applications.
