Pub Date : 2024-06-20DOI: 10.26599/TST.2024.9010022
Qihua Zhou;Zhili Zhou;Zhipeng Bao;Weina Niu;Yuling Liu
Since different kinds of face forgeries leave similar forgery traces in videos, learning the common features from different kinds of forged faces would achieve promising generalization ability of forgery detection. Therefore, to accurately detect known forgeries while ensuring high generalization ability of detecting unknown forgeries, we propose an intra-inter network (IIN) for face forgery detection (FFD) in videos with continual learning. The proposed IIN mainly consists of three modules, i.e., intra-module, inter-module, and forged trace masking module (FTMM). Specifically, the intra-module is trained for each kind of face forgeries by supervised learning to extract special features, while the inter-module is trained by self-supervised learning to extract the common features. As a result, the common and special features of the different forgeries are decoupled by the two feature learning modules, and then the decoupled common features can be utlized to achieve high generalization ability for FFD. Moreover, the FTMM is deployed for contrastive learning to further improve detection accuracy. The experimental results on FaceForensic++ dataset demonstrate that the proposed IIN outperforms the state-of-the-arts in FFD. Also, the generalization ability of the IIN verified on DFDC and Celeb-DF datasets demonstrates that the proposed IIN significantly improves the generalization ability for FFD.
{"title":"IIN-FFD: Intra-Inter Network for Face Forgery Detection","authors":"Qihua Zhou;Zhili Zhou;Zhipeng Bao;Weina Niu;Yuling Liu","doi":"10.26599/TST.2024.9010022","DOIUrl":"https://doi.org/10.26599/TST.2024.9010022","url":null,"abstract":"Since different kinds of face forgeries leave similar forgery traces in videos, learning the common features from different kinds of forged faces would achieve promising generalization ability of forgery detection. Therefore, to accurately detect known forgeries while ensuring high generalization ability of detecting unknown forgeries, we propose an intra-inter network (IIN) for face forgery detection (FFD) in videos with continual learning. The proposed IIN mainly consists of three modules, i.e., intra-module, inter-module, and forged trace masking module (FTMM). Specifically, the intra-module is trained for each kind of face forgeries by supervised learning to extract special features, while the inter-module is trained by self-supervised learning to extract the common features. As a result, the common and special features of the different forgeries are decoupled by the two feature learning modules, and then the decoupled common features can be utlized to achieve high generalization ability for FFD. Moreover, the FTMM is deployed for contrastive learning to further improve detection accuracy. The experimental results on FaceForensic++ dataset demonstrate that the proposed IIN outperforms the state-of-the-arts in FFD. Also, the generalization ability of the IIN verified on DFDC and Celeb-DF datasets demonstrates that the proposed IIN significantly improves the generalization ability for FFD.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"29 6","pages":"1839-1850"},"PeriodicalIF":6.6,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10566006","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141435251","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-20DOI: 10.26599/TST.2023.9010152
Hang Luo;Lu Han;Tianping Shuai;Fengmin Wang
In this paper, we propose the Priority Facility Location Problem with Outliers (PFLPO), which is a generalization of both the Facility Location Problem with Outliers (FLPO) and Priority Facility Location Problem (PFLP). As our main contribution, we use the technique of primal-dual to provide a 3-approximation algorithm for the PFLPO. We also give two heuristic algorithms. One of them is a greedy-based algorithm and the other is a local search algorithm. Moreover, we compare the experimental results of all the proposed algorithms in order to illustrate their performance.
{"title":"Approximation and Heuristic Algorithms for the Priority Facility Location Problem with Outliers","authors":"Hang Luo;Lu Han;Tianping Shuai;Fengmin Wang","doi":"10.26599/TST.2023.9010152","DOIUrl":"https://doi.org/10.26599/TST.2023.9010152","url":null,"abstract":"In this paper, we propose the Priority Facility Location Problem with Outliers (PFLPO), which is a generalization of both the Facility Location Problem with Outliers (FLPO) and Priority Facility Location Problem (PFLP). As our main contribution, we use the technique of primal-dual to provide a 3-approximation algorithm for the PFLPO. We also give two heuristic algorithms. One of them is a greedy-based algorithm and the other is a local search algorithm. Moreover, we compare the experimental results of all the proposed algorithms in order to illustrate their performance.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"29 6","pages":"1694-1702"},"PeriodicalIF":6.6,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10566010","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141435248","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-20DOI: 10.26599/TST.2024.9010039
Maojie Sun;Fang Dong;Zhaowu Huang;Junzhou Luo
The steel plate is one of the main products in steel industries, and its surface quality directly affects the final product performance. How to detect surface defects of steel plates in real time during the production process is a challenging problem. The single or fixed model compression method cannot be directly applied to the detection of steel surface defects, because it is difficult to consider the diversity of production tasks, the uncertainty caused by environmental factors, such as communication networks, and the influence of process and working conditions in steel plate production. In this paper, we propose an adaptive model compression method for steel surface defect online detection based on expert knowledge and working conditions. First, we establish an expert system to give lightweight model parameters based on the correlation between defect types and manufacturing processes. Then, lightweight model parameters are adaptively adjusted according to working conditions, which improves detection accuracy while ensuring real-time performance. The experimental results show that compared with the detection method of constant lightweight parameter model, the proposed method makes the total detection time cut down by 23.1%, and the deadline satisfaction ratio increased by 36.5%, while upgrading the accuracy by 4.2% and reducing the false detection rate by 4.3%.
{"title":"Adaptive Model Compression for Steel Plate Surface Defect Detection: An Expert Knowledge and Working Condition-Based Approach","authors":"Maojie Sun;Fang Dong;Zhaowu Huang;Junzhou Luo","doi":"10.26599/TST.2024.9010039","DOIUrl":"https://doi.org/10.26599/TST.2024.9010039","url":null,"abstract":"The steel plate is one of the main products in steel industries, and its surface quality directly affects the final product performance. How to detect surface defects of steel plates in real time during the production process is a challenging problem. The single or fixed model compression method cannot be directly applied to the detection of steel surface defects, because it is difficult to consider the diversity of production tasks, the uncertainty caused by environmental factors, such as communication networks, and the influence of process and working conditions in steel plate production. In this paper, we propose an adaptive model compression method for steel surface defect online detection based on expert knowledge and working conditions. First, we establish an expert system to give lightweight model parameters based on the correlation between defect types and manufacturing processes. Then, lightweight model parameters are adaptively adjusted according to working conditions, which improves detection accuracy while ensuring real-time performance. The experimental results show that compared with the detection method of constant lightweight parameter model, the proposed method makes the total detection time cut down by 23.1%, and the deadline satisfaction ratio increased by 36.5%, while upgrading the accuracy by 4.2% and reducing the false detection rate by 4.3%.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"29 6","pages":"1851-1871"},"PeriodicalIF":6.6,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10565998","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141435224","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-04DOI: 10.26599/TST.2024.9010007
Zeyu Xu;Jiamin Cui;Kai Hu;Meiqin Wang
FUTURE is a recently proposed lightweight block cipher that achieved a remarkable hardware performance due to careful design decisions. FUTURE is an Advanced Encryption Standard (AES)-like Substitution-Permutation Network (SPN) with 10 rounds, whose round function consists of four components, i.e., SubCell, MixColumn, ShiftRow, and AddRoundKey. Unlike AES, it is a 64-bit-size block cipher with a 128-bit secret key, and the state can be arranged into 16 cells. Therefore, the operations of FUTURE including its S-box is defined over $boldsymbol{F}_{2}^{4}$ . The previous studies have shown that the integral properties of 4-bit S-boxes are usually weaker than larger-size S-boxes, thus the number of rounds of FUTURE, i.e., 10 rounds only, might be too aggressive to provide enough resistance against integral cryptanalysis. In this paper, we mount the integral cryptanalysis on FUTURE. With state-of-the-art detection techniques, we identify several integral distinguishers of 7 rounds of FUTURE. By extending this 7-round distinguisher by 3 forward rounds, we manage to recover all the 128 bits secret keys from the full FUTURE cipher without the full codebook for the first time. To further achieve better time complexity, we also present a key recovery attack on full FUTURE with full codebook. Both attacks have better time complexity than existing results.
{"title":"Integral Attack on the Full FUTURE Block Cipher","authors":"Zeyu Xu;Jiamin Cui;Kai Hu;Meiqin Wang","doi":"10.26599/TST.2024.9010007","DOIUrl":"https://doi.org/10.26599/TST.2024.9010007","url":null,"abstract":"FUTURE is a recently proposed lightweight block cipher that achieved a remarkable hardware performance due to careful design decisions. FUTURE is an Advanced Encryption Standard (AES)-like Substitution-Permutation Network (SPN) with 10 rounds, whose round function consists of four components, i.e., SubCell, MixColumn, ShiftRow, and AddRoundKey. Unlike AES, it is a 64-bit-size block cipher with a 128-bit secret key, and the state can be arranged into 16 cells. Therefore, the operations of FUTURE including its S-box is defined over \u0000<tex>$boldsymbol{F}_{2}^{4}$</tex>\u0000. The previous studies have shown that the integral properties of 4-bit S-boxes are usually weaker than larger-size S-boxes, thus the number of rounds of FUTURE, i.e., 10 rounds only, might be too aggressive to provide enough resistance against integral cryptanalysis. In this paper, we mount the integral cryptanalysis on FUTURE. With state-of-the-art detection techniques, we identify several integral distinguishers of 7 rounds of FUTURE. By extending this 7-round distinguisher by 3 forward rounds, we manage to recover all the 128 bits secret keys from the full FUTURE cipher without the full codebook for the first time. To further achieve better time complexity, we also present a key recovery attack on full FUTURE with full codebook. Both attacks have better time complexity than existing results.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 1","pages":"161-170"},"PeriodicalIF":6.6,"publicationDate":"2024-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10547705","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142169650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-04-23DOI: 10.26599/TST.2024.9010012
Pan Li;Kai Di;Xinlei Bai;Fulin Chen;Yuanshuang Jiang;Xiping Fu;Yichuan Jiang
With the advancement of electronic information technology and the growth of the intelligent industry, the industrial sector has undergone a shift from simplex, linear, and vertical chains to complex, multi-level, and multi-dimensional networked industrial chains. In order to enhance energy efficiency in multiplex networked industrial chains under time-of-use price, a coarse time granularity task scheduling approach has been adopted. This approach adjusts the distribution of electricity supply based on task deadlines, dividing it into longer periods to facilitate batch access to task information. However, traditional simplex-network task assignment optimization methods are unable to achieve a globally optimal solution for cross-layer links in multiplex networked industrial chains. Existing solutions struggle to balance execution costs and completion efficiency in time-of-use price scenarios. Therefore, this paper presents a mixed-integer linear programming model for solving the problem scenario and two algorithms: an exact algorithm based on the branch-and-bound method and a multi-objective heuristic algorithm based on cross-layer policy propagation. These algorithms are designed to adapt to small-scale and large-scale problem scenarios under coarse time granularity. Through extensive simulation experiments and theoretical analysis, the proposed methods effectively optimize the energy and time costs associated with the task execution.
{"title":"Time-of-Use Price Resource Scheduling in Multiplex Networked Industrial Chains","authors":"Pan Li;Kai Di;Xinlei Bai;Fulin Chen;Yuanshuang Jiang;Xiping Fu;Yichuan Jiang","doi":"10.26599/TST.2024.9010012","DOIUrl":"https://doi.org/10.26599/TST.2024.9010012","url":null,"abstract":"With the advancement of electronic information technology and the growth of the intelligent industry, the industrial sector has undergone a shift from simplex, linear, and vertical chains to complex, multi-level, and multi-dimensional networked industrial chains. In order to enhance energy efficiency in multiplex networked industrial chains under time-of-use price, a coarse time granularity task scheduling approach has been adopted. This approach adjusts the distribution of electricity supply based on task deadlines, dividing it into longer periods to facilitate batch access to task information. However, traditional simplex-network task assignment optimization methods are unable to achieve a globally optimal solution for cross-layer links in multiplex networked industrial chains. Existing solutions struggle to balance execution costs and completion efficiency in time-of-use price scenarios. Therefore, this paper presents a mixed-integer linear programming model for solving the problem scenario and two algorithms: an exact algorithm based on the branch-and-bound method and a multi-objective heuristic algorithm based on cross-layer policy propagation. These algorithms are designed to adapt to small-scale and large-scale problem scenarios under coarse time granularity. Through extensive simulation experiments and theoretical analysis, the proposed methods effectively optimize the energy and time costs associated with the task execution.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 1","pages":"303-317"},"PeriodicalIF":6.6,"publicationDate":"2024-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10507222","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142169606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-04-09DOI: 10.26599/TST.2023.9010158
Xiangyu Wang;Jianfeng Ma
The core goal of network security is to protect the security of data sharing. Traditional wireless network security technology is committed to guaranteeing end-to-end data transmission security. However, with the advancement of mobile networks, cloud computing, and Internet of Things, communication-computing integration and cloud-network integration have been important technical routes. As a result, the main application requirements of wireless networks have changed from data transmission to cloud-based information services. Traditional data transmission security technology cannot overcome the security requirements of cloud-network-end collaborative services in the new era, and secure semantic communication has become an important model. To address this issue, we propose a cloud-network-end collaborative security architecture. Firstly, we clarify security mechanisms for end system security, network connection security, and cloud services security, respectively. Next, based on the above three aspects, we elaborate on the connotation of cloud-network-end collaborative security. By giving example applications, including heterogeneous network secure convergence framework, unmanned system collaborative operations security framework, and space-air-ground integrated network security framework, we demonstrate the universality of the proposed architecture. Finally, we review the current research on end system security, network connection security, and cloud services security, respectively.
{"title":"Cloud-Network-End Collaborative Security for Wireless Networks: Architecture, Mechanisms, and Applications","authors":"Xiangyu Wang;Jianfeng Ma","doi":"10.26599/TST.2023.9010158","DOIUrl":"https://doi.org/10.26599/TST.2023.9010158","url":null,"abstract":"The core goal of network security is to protect the security of data sharing. Traditional wireless network security technology is committed to guaranteeing end-to-end data transmission security. However, with the advancement of mobile networks, cloud computing, and Internet of Things, communication-computing integration and cloud-network integration have been important technical routes. As a result, the main application requirements of wireless networks have changed from data transmission to cloud-based information services. Traditional data transmission security technology cannot overcome the security requirements of cloud-network-end collaborative services in the new era, and secure semantic communication has become an important model. To address this issue, we propose a cloud-network-end collaborative security architecture. Firstly, we clarify security mechanisms for end system security, network connection security, and cloud services security, respectively. Next, based on the above three aspects, we elaborate on the connotation of cloud-network-end collaborative security. By giving example applications, including heterogeneous network secure convergence framework, unmanned system collaborative operations security framework, and space-air-ground integrated network security framework, we demonstrate the universality of the proposed architecture. Finally, we review the current research on end system security, network connection security, and cloud services security, respectively.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 1","pages":"18-33"},"PeriodicalIF":6.6,"publicationDate":"2024-04-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10495800","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142169564","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recently, with the increasing complexity of multiplex Unmanned Aerial Vehicles (multi-UAVs) collaboration in dynamic task environments, multi-UAVs systems have shown new characteristics of inter-coupling among multiplex groups and intra-correlation within groups. However, previous studies often overlooked the structural impact of dynamic risks on agents among multiplex UAV groups, which is a critical issue for modern multi-UAVs communication to address. To address this problem, we integrate the influence of dynamic risks on agents among multiplex UAV group structures into a multi-UAVs task migration problem and formulate it as a partially observable Markov game. We then propose a Hybrid Attention Multi-agent Reinforcement Learning (HAMRL) algorithm, which uses attention structures to learn the dynamic characteristics of the task environment, and it integrates hybrid attention mechanisms to establish efficient intra- and inter-group communication aggregation for information extraction and group collaboration. Experimental results show that in this comprehensive and challenging model, our algorithm significantly outperforms state-of-the-art algorithms in terms of convergence speed and algorithm performance due to the rational design of communication mechanisms.
{"title":"Optimizing Risk-Aware Task Migration Algorithm Among Multiplex UAV Groups Through Hybrid Attention Multi-Agent Reinforcement Learning","authors":"Yuanshuang Jiang;Kai Di;Ruiyi Qian;Xingyu Wu;Fulin Chen;Pan Li;Xiping Fu;Yichuan Jiang","doi":"10.26599/TST.2024.9010013","DOIUrl":"https://doi.org/10.26599/TST.2024.9010013","url":null,"abstract":"Recently, with the increasing complexity of multiplex Unmanned Aerial Vehicles (multi-UAVs) collaboration in dynamic task environments, multi-UAVs systems have shown new characteristics of inter-coupling among multiplex groups and intra-correlation within groups. However, previous studies often overlooked the structural impact of dynamic risks on agents among multiplex UAV groups, which is a critical issue for modern multi-UAVs communication to address. To address this problem, we integrate the influence of dynamic risks on agents among multiplex UAV group structures into a multi-UAVs task migration problem and formulate it as a partially observable Markov game. We then propose a Hybrid Attention Multi-agent Reinforcement Learning (HAMRL) algorithm, which uses attention structures to learn the dynamic characteristics of the task environment, and it integrates hybrid attention mechanisms to establish efficient intra- and inter-group communication aggregation for information extraction and group collaboration. Experimental results show that in this comprehensive and challenging model, our algorithm significantly outperforms state-of-the-art algorithms in terms of convergence speed and algorithm performance due to the rational design of communication mechanisms.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 1","pages":"318-330"},"PeriodicalIF":6.6,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10488078","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142169603","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Multipath signal recognition is crucial to the ability to provide high-precision absolute-position services by the BeiDou Navigation Satellite System (BDS). However, most existing approaches to this issue involve supervised machine learning (ML) methods, and it is difficult to move to unsupervised multipath signal recognition because of the limitations in signal labeling. Inspired by an autoencoder with powerful unsupervised feature extraction, we propose a new deep learning (DL) model for BDS signal recognition that places a long short-term memory (LSTM) module in series with a convolutional sparse autoencoder to create a new autoencoder structure. First, we propose to capture the temporal correlations in long-duration BeiDou satellite time-series signals by using the LSTM module to mine the temporal change patterns in the time series. Second, we develop a convolutional sparse autoencoder method that learns a compressed representation of the input data, which then enables downscaled and unsupervised feature extraction from long-duration BeiDou satellite series signals. Finally, we add an l 1/2 regularizer to the objective function of our DL model to remove redundant neurons from the neural network while ensuring recognition accuracy. We tested our proposed approach on a real urban canyon dataset, and the results demonstrated that our algorithm could achieve better classification performance than two ML-based methods (e.g., 11% better than a support vector machine) and two existing DL-based methods (e.g., 7.26% better than convolutional neural networks).
{"title":"Non-Line-of-Sight Multipath Classification Method for BDS Using Convolutional Sparse Autoencoder with LSTM","authors":"Yahang Qin;Zhenni Li;Shengli Xie;Bo Li;Ming Liu;Victor Kuzin","doi":"10.26599/TST.2024.9010004","DOIUrl":"https://doi.org/10.26599/TST.2024.9010004","url":null,"abstract":"Multipath signal recognition is crucial to the ability to provide high-precision absolute-position services by the BeiDou Navigation Satellite System (BDS). However, most existing approaches to this issue involve supervised machine learning (ML) methods, and it is difficult to move to unsupervised multipath signal recognition because of the limitations in signal labeling. Inspired by an autoencoder with powerful unsupervised feature extraction, we propose a new deep learning (DL) model for BDS signal recognition that places a long short-term memory (LSTM) module in series with a convolutional sparse autoencoder to create a new autoencoder structure. First, we propose to capture the temporal correlations in long-duration BeiDou satellite time-series signals by using the LSTM module to mine the temporal change patterns in the time series. Second, we develop a convolutional sparse autoencoder method that learns a compressed representation of the input data, which then enables downscaled and unsupervised feature extraction from long-duration BeiDou satellite series signals. Finally, we add an l\u0000<inf>1/2</inf>\u0000 regularizer to the objective function of our DL model to remove redundant neurons from the neural network while ensuring recognition accuracy. We tested our proposed approach on a real urban canyon dataset, and the results demonstrated that our algorithm could achieve better classification performance than two ML-based methods (e.g., 11% better than a support vector machine) and two existing DL-based methods (e.g., 7.26% better than convolutional neural networks).","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 1","pages":"68-86"},"PeriodicalIF":6.6,"publicationDate":"2024-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10480322","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142169736","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-19DOI: 10.26599/TST.2023.9010093
Xuehan Wang;Xu Shi;Jintao Wang
Orthogonal Time Frequency Space (OTFS) modulation has exhibited significant potential to further promote the performance of future wireless communication networks especially in high-mobility scenarios. In practical OTFS systems, the subcarrier-dependent Doppler shift which is referred to as the Doppler Squint Effect (DSE) plays an important role due to the assistance of time-frequency modulation. Unfortunately, most existing works on OTFS channel estimation ignore DSE, which leads to severe performance degradation. In this letter, OTFS systems taking DSE into consideration are investigated. Inspired by the input-output analysis with DSE and the embedded pilot pattern, the sparse Bayesian learning based parameter estimation scheme is adopted to recover the delay-Doppler channel. Simulation results verify the excellent performance of the proposed off-grid estimation approach considering DSE.
{"title":"Sparse Bayesian Learning Based Off-Grid Estimation of OTFS Channels with Doppler Squint","authors":"Xuehan Wang;Xu Shi;Jintao Wang","doi":"10.26599/TST.2023.9010093","DOIUrl":"https://doi.org/10.26599/TST.2023.9010093","url":null,"abstract":"Orthogonal Time Frequency Space (OTFS) modulation has exhibited significant potential to further promote the performance of future wireless communication networks especially in high-mobility scenarios. In practical OTFS systems, the subcarrier-dependent Doppler shift which is referred to as the Doppler Squint Effect (DSE) plays an important role due to the assistance of time-frequency modulation. Unfortunately, most existing works on OTFS channel estimation ignore DSE, which leads to severe performance degradation. In this letter, OTFS systems taking DSE into consideration are investigated. Inspired by the input-output analysis with DSE and the embedded pilot pattern, the sparse Bayesian learning based parameter estimation scheme is adopted to recover the delay-Doppler channel. Simulation results verify the excellent performance of the proposed off-grid estimation approach considering DSE.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"29 6","pages":"1821-1828"},"PeriodicalIF":6.6,"publicationDate":"2024-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10475894","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141435218","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-05DOI: 10.26599/TST.2023.9010100
Liwei Dong;Ni Li;Guanghong Gong;Xin Lin
Reinforcement Learning (RL) has emerged as a promising data-driven solution for wargaming decision-making. However, two domain challenges still exist: (1) dealing with discrete-continuous hybrid wargaming control and (2) accelerating RL deployment with rich offline data. Existing RL methods fail to handle these two issues simultaneously, thereby we propose a novel offline RL method targeting hybrid action space. A new constrained action representation technique is developed to build a bidirectional mapping between the original hybrid action space and a latent space in a semantically consistent way. This allows learning a continuous latent policy with offline RL with better exploration feasibility and scalability and reconstructing it back to a needed hybrid policy. Critically, a novel offline RL optimization objective with adaptively adjusted constraints is designed to balance the alleviation and generalization of out-of-distribution actions. Our method demonstrates superior performance and generality across different tasks, particularly in typical realistic wargaming scenarios.
{"title":"Offline Reinforcement Learning with Constrained Hybrid Action Implicit Representation Towards Wargaming Decision-Making","authors":"Liwei Dong;Ni Li;Guanghong Gong;Xin Lin","doi":"10.26599/TST.2023.9010100","DOIUrl":"https://doi.org/10.26599/TST.2023.9010100","url":null,"abstract":"Reinforcement Learning (RL) has emerged as a promising data-driven solution for wargaming decision-making. However, two domain challenges still exist: (1) dealing with discrete-continuous hybrid wargaming control and (2) accelerating RL deployment with rich offline data. Existing RL methods fail to handle these two issues simultaneously, thereby we propose a novel offline RL method targeting hybrid action space. A new constrained action representation technique is developed to build a bidirectional mapping between the original hybrid action space and a latent space in a semantically consistent way. This allows learning a continuous latent policy with offline RL with better exploration feasibility and scalability and reconstructing it back to a needed hybrid policy. Critically, a novel offline RL optimization objective with adaptively adjusted constraints is designed to balance the alleviation and generalization of out-of-distribution actions. Our method demonstrates superior performance and generality across different tasks, particularly in typical realistic wargaming scenarios.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"29 5","pages":"1422-1440"},"PeriodicalIF":6.6,"publicationDate":"2024-03-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10460410","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140820229","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","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：