{"title":"Sensor Self-Declaration of Numeric Data Reliability in Internet of Things","authors":"Sakib Shahriar Shafin, Gour Karmakar, Iven Mareels, Venki Balasubramanian, Ramachandra Rao Kolluri","doi":"10.1109/tr.2024.3416967","DOIUrl":"https://doi.org/10.1109/tr.2024.3416967","url":null,"abstract":"","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"205 1","pages":""},"PeriodicalIF":5.9,"publicationDate":"2024-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141511371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hongxi Liu;Mingzu Zhang;Sun-Yuan Hsieh;Chia-Wei Lee
The underlying topology of the interconnection network of parallel and distributed systems is usually modelled by a simple connected graph $G$. In order to quantitatively analyze the reliability and fault tolerance of these networks more accurately, this study introduces a novel topology parameter. The $g$-good $(r+1)$-component edge-connectivity $lambda _{g,r+1}(G)$ of $G$, if any, is the smallest cardinality of faulty link set, whose malfunction yields a disconnected graph with at least $r+1$ connected components, and with the neighboring edges of any vertex being at least $g$. When designing and maintaining parallel and distributed systems, the hypercube network $Q_{n}$ is one of the most attractive interconnection network models. This article offers a unified method to derive an upper bound for $g$-good $(r+1)$-component edge-connectivity $lambda _{g,r+1}(Q_{n})$ of $Q_{n}$. When $ngeq 4$, this upper bound is proved to be tight for $1leq 2^{g}cdot rleq 2^{lfloor frac{n}{2}rfloor }$ or $r=2^{k_{0}}$, $0leq k_{0}< lfloor frac{n}{2}rfloor$, $0leq gleq n-2k_{0}-1$. The conclusions for the $g$-good-neighbor edge-connectivity of $Q_{n}$ from Xu and the $(r+1)$-component edge-connectivity of $Q_{n}$ from Zhao et al. are contained as corollaries of our main results for $r=1$, $0leq gleq n-1$ and $1leq rleq 2^{lfloor frac{n}{2}rfloor }$, $g=0$, respectively.
{"title":"A Novel Links Fault Tolerant Analysis: $g$-Good $r$-Component Edge-Connectivity of Interconnection Networks With Applications to Hypercubes","authors":"Hongxi Liu;Mingzu Zhang;Sun-Yuan Hsieh;Chia-Wei Lee","doi":"10.1109/TR.2024.3410526","DOIUrl":"10.1109/TR.2024.3410526","url":null,"abstract":"The underlying topology of the interconnection network of parallel and distributed systems is usually modelled by a simple connected graph <inline-formula><tex-math>$G$</tex-math></inline-formula>. In order to quantitatively analyze the reliability and fault tolerance of these networks more accurately, this study introduces a novel topology parameter. The <inline-formula><tex-math>$g$</tex-math></inline-formula>-good <inline-formula><tex-math>$(r+1)$</tex-math></inline-formula>-component edge-connectivity <inline-formula><tex-math>$lambda _{g,r+1}(G)$</tex-math></inline-formula> of <inline-formula><tex-math>$G$</tex-math></inline-formula>, if any, is the smallest cardinality of faulty link set, whose malfunction yields a disconnected graph with at least <inline-formula><tex-math>$r+1$</tex-math></inline-formula> connected components, and with the neighboring edges of any vertex being at least <inline-formula><tex-math>$g$</tex-math></inline-formula>. When designing and maintaining parallel and distributed systems, the hypercube network <inline-formula><tex-math>$Q_{n}$</tex-math></inline-formula> is one of the most attractive interconnection network models. This article offers a unified method to derive an upper bound for <inline-formula><tex-math>$g$</tex-math></inline-formula>-good <inline-formula><tex-math>$(r+1)$</tex-math></inline-formula>-component edge-connectivity <inline-formula><tex-math>$lambda _{g,r+1}(Q_{n})$</tex-math></inline-formula> of <inline-formula><tex-math>$Q_{n}$</tex-math></inline-formula>. When <inline-formula><tex-math>$ngeq 4$</tex-math></inline-formula>, this upper bound is proved to be tight for <inline-formula><tex-math>$1leq 2^{g}cdot rleq 2^{lfloor frac{n}{2}rfloor }$</tex-math></inline-formula> or <inline-formula><tex-math>$r=2^{k_{0}}$</tex-math></inline-formula>, <inline-formula><tex-math>$0leq k_{0}< lfloor frac{n}{2}rfloor$</tex-math></inline-formula>, <inline-formula><tex-math>$0leq gleq n-2k_{0}-1$</tex-math></inline-formula>. The conclusions for the <inline-formula><tex-math>$g$</tex-math></inline-formula>-good-neighbor edge-connectivity of <inline-formula><tex-math>$Q_{n}$</tex-math></inline-formula> from Xu and the <inline-formula><tex-math>$(r+1)$</tex-math></inline-formula>-component edge-connectivity of <inline-formula><tex-math>$Q_{n}$</tex-math></inline-formula> from Zhao et al. are contained as corollaries of our main results for <inline-formula><tex-math>$r=1$</tex-math></inline-formula>, <inline-formula><tex-math>$0leq gleq n-1$</tex-math></inline-formula> and <inline-formula><tex-math>$1leq rleq 2^{lfloor frac{n}{2}rfloor }$</tex-math></inline-formula>, <inline-formula><tex-math>$g=0$</tex-math></inline-formula>, respectively.","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"74 1","pages":"2487-2496"},"PeriodicalIF":5.0,"publicationDate":"2024-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141511373","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The collected data of a pumping unit contain environmental noise, which significantly reduces the precision of fault diagnosis. The previous fault detection approach depends on manual feature extraction, which is time-consuming and laborious, and it cannot cope with high-noise conditions. Therefore, we propose a dual-branch time–frequency fusion deep learning model for fault diagnosis of the pumping unit. One branch extracts time-domain information, while the other branch extracts frequency-domain information by employing the fast Fourier transform. The branch information of these two branches is concatenated, and the gate-controlled channel transfer unit module automatically learns the competitive and cooperative relationships between each branch, making the key features more prominent in information fusion. Consequently, an accurate fault diagnosis of the pumping unit can be achieved under high-noise conditions. The results demonstrate that the proposed model outperforms the traditional schemes in terms of noise, with different signal-to-noise ratios.
{"title":"Fault Diagnosis Algorithm for Pumping Unit Based on Dual-Branch Time–Frequency Fusion","authors":"Fangfang Zhang;Yebin Li;Dongri Shan;Yuanhong Liu;Fengying Ma;Weiyong Yu","doi":"10.1109/TR.2024.3409427","DOIUrl":"10.1109/TR.2024.3409427","url":null,"abstract":"The collected data of a pumping unit contain environmental noise, which significantly reduces the precision of fault diagnosis. The previous fault detection approach depends on manual feature extraction, which is time-consuming and laborious, and it cannot cope with high-noise conditions. Therefore, we propose a dual-branch time–frequency fusion deep learning model for fault diagnosis of the pumping unit. One branch extracts time-domain information, while the other branch extracts frequency-domain information by employing the fast Fourier transform. The branch information of these two branches is concatenated, and the gate-controlled channel transfer unit module automatically learns the competitive and cooperative relationships between each branch, making the key features more prominent in information fusion. Consequently, an accurate fault diagnosis of the pumping unit can be achieved under high-noise conditions. The results demonstrate that the proposed model outperforms the traditional schemes in terms of noise, with different signal-to-noise ratios.","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"74 1","pages":"2082-2091"},"PeriodicalIF":5.0,"publicationDate":"2024-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141942534","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Smart contracts with excessive gas consumption can cause economic losses, such as black hole contracts. Actual gas consumption depends on runtime information and has a probability distribution under different runtime situations. However, existing static analysis tools (e.g., Solc) cannot define runtime information and only provide an approximate upper bound on gas consumption without explanation. To address the challenge, we propose a label named GCL, which describes the probability distribution of gas consumption, a code representation method containing domain features and a graph neural network (GNN) named attention-based graph isomorphism network (AGIN) oriented to domain feature, and SubgraphGas, a domain-oriented subgraph-level GNN explanation model. By combining AGIN and SubgraphGas, we have created a new explainable gas consumption prediction model (EGE). Our evaluations show that EGE outperforms prediction schemes based on Bi-LSTM. And EGE has similar explainability accuracy to general methods, but it is more efficient.
{"title":"GuessGas: Tell Me Fine-Grained Gas Consumption of My Smart Contract and Why","authors":"Qing Huang;Renxiong Chen;Zhenchang Xing;Jinshan Zeng;Qinghua Lu;Xiwei Xu","doi":"10.1109/TR.2024.3404238","DOIUrl":"10.1109/TR.2024.3404238","url":null,"abstract":"Smart contracts with excessive gas consumption can cause economic losses, such as black hole contracts. Actual gas consumption depends on runtime information and has a probability distribution under different runtime situations. However, existing static analysis tools (e.g., Solc) cannot define runtime information and only provide an approximate upper bound on gas consumption without explanation. To address the challenge, we propose a label named GCL, which describes the probability distribution of gas consumption, a code representation method containing domain features and a graph neural network (GNN) named attention-based graph isomorphism network (AGIN) oriented to domain feature, and SubgraphGas, a domain-oriented subgraph-level GNN explanation model. By combining AGIN and SubgraphGas, we have created a new explainable gas consumption prediction model (EGE). Our evaluations show that EGE outperforms prediction schemes based on Bi-LSTM. And EGE has similar explainability accuracy to general methods, but it is more efficient.","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"74 1","pages":"2348-2362"},"PeriodicalIF":5.0,"publicationDate":"2024-06-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141942533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The open environment presents a challenging issue for the online safety assessment of dynamic systems, which means that unknown scenarios may arise unexpectedly. These unknown scenarios can be task-related and result in the within-class distribution mismatch. Addressing this challenge in the semi-supervised safety assessment task, where unlabeled training data contain task-related unknown scenarios, has not been explored. This article investigates this new semi-supervised safety assessment problem. A novel active-labeling-aided semi-supervised learning scheme is proposed to tackle the within-class distribution mismatch between labeled and unlabeled training data. The proposed scheme begins by detecting out-of-distribution unlabeled data through the construction of a deep support vector data description network for each class. Subsequently, an active labeling approach along with its kernel extension is introduced, taking into account both distribution mismatch degree and sample representativeness. The proposed active labeling approach can be seamlessly integrated with any semi-supervised learning algorithm to enhance its performance in handling task-related unknown scenarios. The effectiveness and applicability of the proposed method are demonstrated through case studies based on a bearing dataset and operation data from an actual deep-sea manned submersible.
{"title":"Active Labeling Aided Semi-Supervised Safety Assessment With Task-Related Unknown Scenarios","authors":"Chang Liu;Xiao He;Minyue Li;Yi Zhang;Zhongjun Ding","doi":"10.1109/TR.2024.3376601","DOIUrl":"10.1109/TR.2024.3376601","url":null,"abstract":"The open environment presents a challenging issue for the online safety assessment of dynamic systems, which means that unknown scenarios may arise unexpectedly. These unknown scenarios can be task-related and result in the within-class distribution mismatch. Addressing this challenge in the semi-supervised safety assessment task, where unlabeled training data contain task-related unknown scenarios, has not been explored. This article investigates this new semi-supervised safety assessment problem. A novel active-labeling-aided semi-supervised learning scheme is proposed to tackle the within-class distribution mismatch between labeled and unlabeled training data. The proposed scheme begins by detecting out-of-distribution unlabeled data through the construction of a deep support vector data description network for each class. Subsequently, an active labeling approach along with its kernel extension is introduced, taking into account both distribution mismatch degree and sample representativeness. The proposed active labeling approach can be seamlessly integrated with any semi-supervised learning algorithm to enhance its performance in handling task-related unknown scenarios. The effectiveness and applicability of the proposed method are demonstrated through case studies based on a bearing dataset and operation data from an actual deep-sea manned submersible.","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"73 4","pages":"1792-1804"},"PeriodicalIF":5.0,"publicationDate":"2024-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141942437","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Tse-Chuan Hsu, Yao-Hong Tsai, William Cheng-Chung Chu
{"title":"Enhancing Digit Recognition for Luminous Images in Edge Computing Through Transfer Learning With Robustness and Fault Tolerance","authors":"Tse-Chuan Hsu, Yao-Hong Tsai, William Cheng-Chung Chu","doi":"10.1109/tr.2024.3393424","DOIUrl":"https://doi.org/10.1109/tr.2024.3393424","url":null,"abstract":"","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"79 1","pages":""},"PeriodicalIF":5.9,"publicationDate":"2024-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141942536","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Resistive random-access memory (ReRAM)-based processing-in-memory (PIM) accelerator is a promising platform for processing massively memory intensive matrix-vector multiplications of neural networks in parallel domain, due to its capability of analog computation, ultra-high density, near-zero leakage current, and nonvolatility. Despite many advantages, ReRAM-based accelerators are highly error-prone due to limitations of technology fabrication that lead to process variations and defects. These limitations degrade the accuracy of deep convolutional neural networks (CNNs) (Deep CNNs) running on PIM accelerators. While these CNNs accelerators are widely deployed in safety-critical systems, their vulnerability to fault is not well explored. In this article, we have developed a fault-injection framework to investigate the vulnerability of large-scale CNNs at both software- and hardware-level of inference phases. Faulty ReRAM devices are another reliability challenges due to significant degradation of classification accuracy when CNN parameters are mapped to the accelerators. To investigate this challenge, we map the CNN learning parameter to the ReRAM crossbar and inject faults into crossbar arrays. The proposed framework analyzes the impact of stuck-at high (SaH) and stuck-at low (SaL) fault models on different layers and locations of CNN learning parameters. By performing extensive fault injections, we illustrate that the vulnerability behavior of ReRAM-based PIM accelerator for CNNs is greatly impressible to the types and depth of layers, the location of the learning parameter in every layer, and the value and types of faults. Our observations show that different models have different vulnerabilities to faults. Specifically, we show that SaL further reduces classification accuracy than SaH.
{"title":"An Empirical Fault Vulnerability Exploration of ReRAM-Based Process-in-Memory CNN Accelerators","authors":"Aniseh Dorostkar;Hamed Farbeh;Hamid R. Zarandi","doi":"10.1109/TR.2024.3405825","DOIUrl":"10.1109/TR.2024.3405825","url":null,"abstract":"Resistive random-access memory (ReRAM)-based <italic>processing-in-memory</i> (PIM) accelerator is a promising platform for processing massively memory intensive matrix-vector multiplications of neural networks in parallel domain, due to its capability of analog computation, ultra-high density, near-zero leakage current, and nonvolatility. Despite many advantages, ReRAM-based accelerators are highly error-prone due to limitations of technology fabrication that lead to process variations and defects. These limitations degrade the accuracy of deep convolutional neural networks (CNNs) (Deep CNNs) running on PIM accelerators. While these CNNs accelerators are widely deployed in safety-critical systems, their vulnerability to fault is not well explored. In this article, we have developed a fault-injection framework to investigate the vulnerability of large-scale CNNs at both software- and hardware-level of inference phases. Faulty ReRAM devices are another reliability challenges due to significant degradation of classification accuracy when CNN parameters are mapped to the accelerators. To investigate this challenge, we map the CNN learning parameter to the ReRAM crossbar and inject faults into crossbar arrays. The proposed framework analyzes the impact of <italic>stuck-at high</i> (SaH) and <italic>stuck-at low</i> (SaL) fault models on different layers and locations of CNN learning parameters. By performing extensive fault injections, we illustrate that the vulnerability behavior of ReRAM-based PIM accelerator for CNNs is greatly impressible to the types and depth of layers, the location of the learning parameter in every layer, and the value and types of faults. Our observations show that different models have different vulnerabilities to faults. Specifically, we show that SaL further reduces classification accuracy than SaH.","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"74 1","pages":"2290-2304"},"PeriodicalIF":5.0,"publicationDate":"2024-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141942537","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
From the perspective of industrial production reliability, a robust event-triggered (ET) control strategy is presented for uncertain continuous stirred tank reactor (CSTR) system with asymmetric input constraints. To begin with, we propose a nonquadratic performance function to transform the robust control issue by constructing the relevant auxiliary dynamics. For effectively mitigating the pressure of data transmission and controller execution, a dynamic ET scheme (DETS) with an adjustable threshold function is adopted. Subsequently, we formulate the DETS-based Hamilton–Jacobi–Bellman (DET-HJB) equation according to optimality theory. In addition, a DETS-assisted reinforcement learning algorithm with a unique critic neural network can efficiently tackle the derived DET-HJB equation. Meanwhile, the corresponding critic weight is regulated on the basis of gradient descent technique and experience replay approach. By presenting a rigorous analysis under two situations, the uniform ultimate boundedness of auxiliary dynamics and weight approximation error can be ensured. Eventually, the feasibility of the proposed algorithm is demonstrated by experimental results of CSTR system.
{"title":"Reinforcement Learning-Boosted Event-Triggered Reliability Control for Uncertain CSTR System With Asymmetric Constraints","authors":"Jian Liu;Jiachen Ke;Jinliang Liu;Xiangpeng Xie;Engang Tian","doi":"10.1109/TR.2024.3407090","DOIUrl":"10.1109/TR.2024.3407090","url":null,"abstract":"From the perspective of industrial production reliability, a robust event-triggered (ET) control strategy is presented for uncertain continuous stirred tank reactor (CSTR) system with asymmetric input constraints. To begin with, we propose a nonquadratic performance function to transform the robust control issue by constructing the relevant auxiliary dynamics. For effectively mitigating the pressure of data transmission and controller execution, a dynamic ET scheme (DETS) with an adjustable threshold function is adopted. Subsequently, we formulate the DETS-based Hamilton–Jacobi–Bellman (DET-HJB) equation according to optimality theory. In addition, a DETS-assisted reinforcement learning algorithm with a unique critic neural network can efficiently tackle the derived DET-HJB equation. Meanwhile, the corresponding critic weight is regulated on the basis of gradient descent technique and experience replay approach. By presenting a rigorous analysis under two situations, the uniform ultimate boundedness of auxiliary dynamics and weight approximation error can be ensured. Eventually, the feasibility of the proposed algorithm is demonstrated by experimental results of CSTR system.","PeriodicalId":56305,"journal":{"name":"IEEE Transactions on Reliability","volume":"74 1","pages":"2069-2081"},"PeriodicalIF":5.0,"publicationDate":"2024-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141942570","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","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：