Pub Date : 2022-04-01DOI: 10.1177/15501329221090458
Zeyu Sun, G. Liao, Cao Zeng, Lan Lan, Guozeng Zhao
Load balancing is of great significance to extend the longevity of wireless sensor networks, due to the inherent imbalanced energy overhead in such networks. However, existing solutions cannot balance the load distribution in partially disconnected wireless sensor networks. For example, if a network is partitioned into several segments with different area sizes, some areas have much more traffic load than other areas. In this article, we propose a load-balanced routing scheme, which aims to balance energy consumption within each segment and among different segments. First, we adopt unequal transmission distances to build initial routing for intrasegment load balancing. Second, we adopt the genetic algorithm to build extra routing between different segments for intersegment load balancing. The unique character of our work is twofold. On one hand, we investigate partitioned wireless sensor networks where there are several isolated segments. On the other hand, we pursue load balancing from a global perspective rather than from a local one. Some simulations verify the effectiveness and the advantages of our scheme in terms of extra deployment cost, system longevity, and load balancing degree.
{"title":"GLBR: A novel global load balancing routing scheme based on intelligent computing in partially disconnected wireless sensor networks","authors":"Zeyu Sun, G. Liao, Cao Zeng, Lan Lan, Guozeng Zhao","doi":"10.1177/15501329221090458","DOIUrl":"https://doi.org/10.1177/15501329221090458","url":null,"abstract":"Load balancing is of great significance to extend the longevity of wireless sensor networks, due to the inherent imbalanced energy overhead in such networks. However, existing solutions cannot balance the load distribution in partially disconnected wireless sensor networks. For example, if a network is partitioned into several segments with different area sizes, some areas have much more traffic load than other areas. In this article, we propose a load-balanced routing scheme, which aims to balance energy consumption within each segment and among different segments. First, we adopt unequal transmission distances to build initial routing for intrasegment load balancing. Second, we adopt the genetic algorithm to build extra routing between different segments for intersegment load balancing. The unique character of our work is twofold. On one hand, we investigate partitioned wireless sensor networks where there are several isolated segments. On the other hand, we pursue load balancing from a global perspective rather than from a local one. Some simulations verify the effectiveness and the advantages of our scheme in terms of extra deployment cost, system longevity, and load balancing degree.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42341092","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-04-01DOI: 10.1177/15501329221091505
Xiaobo Zhu, Yunlong Ye
At present, intravenous infusion in clinical medicine is a very important treatment method in medical care, and it is the main work content of medical staff to ensure the safety and effectiveness of intravenous infusion process. The monitoring system of infusion patient’s physical signs is integrated with sensor technology, bio-electronic technology, and computer network technology. By means of sensor detection, program control and data processing, the automatic detection and control of intravenous infusion process and real-time monitoring of the pulse information are realized. The monitoring equipment has been unable to adapt to the rapid development of medical technology. In this article, based on ZigBee module, automatic control of drip rate, abnormal alarm, real-time liquid crystal display of infusion progress, and other functions are realized without manual intervention. At the same time, the hardware part of the lower computer combines the detection of ECG and pulse to design a multi-parameter monitoring system of vital signs, and realizes the parameter synchronization management requirements of the monitoring terminal based on the local area network. The test proves that the research content in this article is higher than the traditional clinical management mode in terms of monitoring data consistency and data synchronization management accuracy, which can not only reduce the labor intensity of nursing staff, improve work efficiency, but also greatly reduce the potential risks in the nursing process, which is the future application trend of clinical medical nursing work.
{"title":"Monitoring of clinical signs of intravenous infusion patients with ZigBee wireless technology","authors":"Xiaobo Zhu, Yunlong Ye","doi":"10.1177/15501329221091505","DOIUrl":"https://doi.org/10.1177/15501329221091505","url":null,"abstract":"At present, intravenous infusion in clinical medicine is a very important treatment method in medical care, and it is the main work content of medical staff to ensure the safety and effectiveness of intravenous infusion process. The monitoring system of infusion patient’s physical signs is integrated with sensor technology, bio-electronic technology, and computer network technology. By means of sensor detection, program control and data processing, the automatic detection and control of intravenous infusion process and real-time monitoring of the pulse information are realized. The monitoring equipment has been unable to adapt to the rapid development of medical technology. In this article, based on ZigBee module, automatic control of drip rate, abnormal alarm, real-time liquid crystal display of infusion progress, and other functions are realized without manual intervention. At the same time, the hardware part of the lower computer combines the detection of ECG and pulse to design a multi-parameter monitoring system of vital signs, and realizes the parameter synchronization management requirements of the monitoring terminal based on the local area network. The test proves that the research content in this article is higher than the traditional clinical management mode in terms of monitoring data consistency and data synchronization management accuracy, which can not only reduce the labor intensity of nursing staff, improve work efficiency, but also greatly reduce the potential risks in the nursing process, which is the future application trend of clinical medical nursing work.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41494960","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-04-01DOI: 10.1177/15501329221092932
Xin Yan, Yiming Qin, Xiao Hu, Xiaoling Xiao
Federated learning framework facilitates more applications of deep learning algorithms on the existing network architectures, where the model parameters are aggregated in a centralized manner. However, some of federated learning participants are often inaccessible, such as in a power shortage or dormant state. That will force us to explore the possibility that the parameter aggregation is operated in an ad hoc manner, which is based on consensus computing. On the contrary, since caching mechanism is indispensable to any federated learning mobile node, it is necessary to investigate the connection between it and consensus computing. In this article, we first propose a novel federated learning paradigm, which supports an ad hoc operation mode for federated learning participants. Second, a discrete-time dynamic equation and its control law are formulated to satisfy the demands from federated learning framework, with a quantized caching scheme designed to mask the uncertainties from both asynchronous updates and measurement noises. Then, the consensus conditions and the convergence of the consensus protocol are deduced analytically, and a quantized caching strategy to optimize the convergence speed is provided. Our major contribution is to give the basic theories of distributed consensus problem for federated learning framework, and the theoretical results are validated by numerical simulations.
{"title":"Distributed consensus problem with caching on federated learning framework","authors":"Xin Yan, Yiming Qin, Xiao Hu, Xiaoling Xiao","doi":"10.1177/15501329221092932","DOIUrl":"https://doi.org/10.1177/15501329221092932","url":null,"abstract":"Federated learning framework facilitates more applications of deep learning algorithms on the existing network architectures, where the model parameters are aggregated in a centralized manner. However, some of federated learning participants are often inaccessible, such as in a power shortage or dormant state. That will force us to explore the possibility that the parameter aggregation is operated in an ad hoc manner, which is based on consensus computing. On the contrary, since caching mechanism is indispensable to any federated learning mobile node, it is necessary to investigate the connection between it and consensus computing. In this article, we first propose a novel federated learning paradigm, which supports an ad hoc operation mode for federated learning participants. Second, a discrete-time dynamic equation and its control law are formulated to satisfy the demands from federated learning framework, with a quantized caching scheme designed to mask the uncertainties from both asynchronous updates and measurement noises. Then, the consensus conditions and the convergence of the consensus protocol are deduced analytically, and a quantized caching strategy to optimize the convergence speed is provided. Our major contribution is to give the basic theories of distributed consensus problem for federated learning framework, and the theoretical results are validated by numerical simulations.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48346434","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-04-01DOI: 10.1177/15501329221092940
Cai-xue Zhou, Lihua Wang, Lingmin Wang
The smart grid is considered to be the next-generation electric power network. In a smart grid, there are massive data to be processed, so cloud computing is introduced into it to form a cloud-based smart grid data management system. However, with data no longer being stored locally, how to ensure the integrity of data stored in the cloud in the smart grid has become an urgent problem awaiting solution. Provable data possession has been proposed to solve this problem. With the development of quantum computer technology, quantum attacks-resistant cryptographic schemes are gradually entering people’s horizons. Lattice cryptography can resist quantum attacks. In this article, a lattice-based provable data possession scheme is proposed for cloud-based smart grid data management systems. The scheme is proved unforgeable under the small integer solution hard assumption in the standard model. Compared with other two efficient lattice-based provable data possession schemes in the standard model, our scheme also shows efficiency.
{"title":"Lattice-based provable data possession in the standard model for cloud-based smart grid data management systems","authors":"Cai-xue Zhou, Lihua Wang, Lingmin Wang","doi":"10.1177/15501329221092940","DOIUrl":"https://doi.org/10.1177/15501329221092940","url":null,"abstract":"The smart grid is considered to be the next-generation electric power network. In a smart grid, there are massive data to be processed, so cloud computing is introduced into it to form a cloud-based smart grid data management system. However, with data no longer being stored locally, how to ensure the integrity of data stored in the cloud in the smart grid has become an urgent problem awaiting solution. Provable data possession has been proposed to solve this problem. With the development of quantum computer technology, quantum attacks-resistant cryptographic schemes are gradually entering people’s horizons. Lattice cryptography can resist quantum attacks. In this article, a lattice-based provable data possession scheme is proposed for cloud-based smart grid data management systems. The scheme is proved unforgeable under the small integer solution hard assumption in the standard model. Compared with other two efficient lattice-based provable data possession schemes in the standard model, our scheme also shows efficiency.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43346102","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This article considers the load frequency control of multi-area power system-based multi-agent system method under false data injection attacks. The research can provide better solutions for multi-area power system load frequency control under false data injection attacks. First, an event-triggered mechanism is introduced to decide which data should be transmitted in the controller to save the limited network bandwidth. Besides, a model of cyberattacks is built using the Bernoulli random variables. Then, conditions are given for maintaining the system asymptotic stability under attack. Finally, simulations are performed to demonstrate the validity of the theory proposed in this article.
{"title":"Load frequency control under false data inject attacks based on multi-agent system method in multi-area power systems","authors":"Tengfei Weng, Yan Xie, Guorong Chen, Qi Han, Yuan Tian, Liping Feng, Yangjun Pei","doi":"10.1177/15501329221090469","DOIUrl":"https://doi.org/10.1177/15501329221090469","url":null,"abstract":"This article considers the load frequency control of multi-area power system-based multi-agent system method under false data injection attacks. The research can provide better solutions for multi-area power system load frequency control under false data injection attacks. First, an event-triggered mechanism is introduced to decide which data should be transmitted in the controller to save the limited network bandwidth. Besides, a model of cyberattacks is built using the Bernoulli random variables. Then, conditions are given for maintaining the system asymptotic stability under attack. Finally, simulations are performed to demonstrate the validity of the theory proposed in this article.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46551462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-04-01DOI: 10.1177/15501329221084223
Huan Wang, Zhongyuan Jiang, Qing Qian, Hong Wang
This article proposes an efficient XOR-based visual cryptography scheme with lossless reconfigurable algorithms, which includes an encryption and a decryption schemes for gray-scale and color secret images. The encryption scheme aims to encrypt secret images by mapping them to several shared images that do not suffer any pixel expansion problem and do not reveal any information of the secret images by considering a random column selection algorithm based on the 0-mapping and 1-mapping matrices, where the two matrices are dynamically generated in the encryption scheme. Moreover, the proposed decryption scheme can reconstruct the secret images with only a series of XOR operations. The reconstructed images do not suffer any contrast distortion and pixel expansion problems comparing with their original secret images. Finally, the proposed two schemes are illustrated with the practical examples.
{"title":"An efficient XOR-based visual cryptography scheme with lossless reconfigurable algorithms","authors":"Huan Wang, Zhongyuan Jiang, Qing Qian, Hong Wang","doi":"10.1177/15501329221084223","DOIUrl":"https://doi.org/10.1177/15501329221084223","url":null,"abstract":"This article proposes an efficient XOR-based visual cryptography scheme with lossless reconfigurable algorithms, which includes an encryption and a decryption schemes for gray-scale and color secret images. The encryption scheme aims to encrypt secret images by mapping them to several shared images that do not suffer any pixel expansion problem and do not reveal any information of the secret images by considering a random column selection algorithm based on the 0-mapping and 1-mapping matrices, where the two matrices are dynamically generated in the encryption scheme. Moreover, the proposed decryption scheme can reconstruct the secret images with only a series of XOR operations. The reconstructed images do not suffer any contrast distortion and pixel expansion problems comparing with their original secret images. Finally, the proposed two schemes are illustrated with the practical examples.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49166971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-04-01DOI: 10.1177/15501329221084882
Liguo Zhao, D. Zhu, Wasswa Shafik, S. Matinkhah, Zubair Ahmad, Lule Sharif, Alisa Craig
The application of Big Data Analytics is identified through the Cyber Research Alliance for cybersecurity as the foremost preference for future studies and advancement in the field of cybersecurity. In this study, we develop a repeatable procedure for detecting cyber-attacks in an accurate, scalable, and timely manner. An in-depth learning algorithm is utilized for training a neural network for detecting suspicious user activities. The proposed system architecture was implemented with the help of Splunk Enterprise Edition 6.42. A data set of average feature counts has been executed through a Splunk search command in 1-min intervals. All the data sets consisted of a minute trait total derived from a sparkling file. The attack patterns that were not anonymized or were indicative of the vulnerability of cyber-attack were denoted with yellow. The rule-based method dispensed a low quantity of irregular illustrations in contrast with the Partitioning Around Medoids method. The results in this study demonstrated that using a proportional collection of instances trained with the deep learning algorithm, a classified data set can accurately detect suspicious behavior. This method permits for the allocation of multiple log source types through a sliding time window and provides a scalable solution, which is a much-needed function.
{"title":"Artificial intelligence analysis in cyber domain: A review","authors":"Liguo Zhao, D. Zhu, Wasswa Shafik, S. Matinkhah, Zubair Ahmad, Lule Sharif, Alisa Craig","doi":"10.1177/15501329221084882","DOIUrl":"https://doi.org/10.1177/15501329221084882","url":null,"abstract":"The application of Big Data Analytics is identified through the Cyber Research Alliance for cybersecurity as the foremost preference for future studies and advancement in the field of cybersecurity. In this study, we develop a repeatable procedure for detecting cyber-attacks in an accurate, scalable, and timely manner. An in-depth learning algorithm is utilized for training a neural network for detecting suspicious user activities. The proposed system architecture was implemented with the help of Splunk Enterprise Edition 6.42. A data set of average feature counts has been executed through a Splunk search command in 1-min intervals. All the data sets consisted of a minute trait total derived from a sparkling file. The attack patterns that were not anonymized or were indicative of the vulnerability of cyber-attack were denoted with yellow. The rule-based method dispensed a low quantity of irregular illustrations in contrast with the Partitioning Around Medoids method. The results in this study demonstrated that using a proportional collection of instances trained with the deep learning algorithm, a classified data set can accurately detect suspicious behavior. This method permits for the allocation of multiple log source types through a sliding time window and provides a scalable solution, which is a much-needed function.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47198515","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-01DOI: 10.1177/15501477221075544
Marek Ruzicka, Marcel Volosin, J. Gazda, T. Maksymyuk, Longzhe Han, M. Dohler
The challenge of dynamic traffic demand in mobile networks is tackled by moving cells based on unmanned aerial vehicles. Considering the tremendous potential of unmanned aerial vehicles in the future, we propose a new heuristic algorithm for coverage optimization. The proposed algorithm is implemented based on a conditional generative adversarial neural network, with a unique multilayer sum-pooling loss function. To assess the performance of the proposed approach, we compare it with the optimal core-set algorithm and quasi-optimal spiral algorithm. Simulation results show that the proposed approach converges to the quasi-optimal solution with a negligible difference from the global optimum while maintaining a quadratic complexity regardless of the number of users.
{"title":"Fast and computationally efficient generative adversarial network algorithm for unmanned aerial vehicle–based network coverage optimization","authors":"Marek Ruzicka, Marcel Volosin, J. Gazda, T. Maksymyuk, Longzhe Han, M. Dohler","doi":"10.1177/15501477221075544","DOIUrl":"https://doi.org/10.1177/15501477221075544","url":null,"abstract":"The challenge of dynamic traffic demand in mobile networks is tackled by moving cells based on unmanned aerial vehicles. Considering the tremendous potential of unmanned aerial vehicles in the future, we propose a new heuristic algorithm for coverage optimization. The proposed algorithm is implemented based on a conditional generative adversarial neural network, with a unique multilayer sum-pooling loss function. To assess the performance of the proposed approach, we compare it with the optimal core-set algorithm and quasi-optimal spiral algorithm. Simulation results show that the proposed approach converges to the quasi-optimal solution with a negligible difference from the global optimum while maintaining a quadratic complexity regardless of the number of users.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48823921","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-01DOI: 10.1177/15501329221077922
Min Wei, Chunmeng Rong, Erxiong Liang, Yuan Zhuang
With the advancement of IPv6 technology, many nodes in wireless sensor networks realize seamless connections with the Internet via IPv6 addresses. Security issues are a significant obstacle to the widespread adoption of IPv6 technology. Resource-constrained IPv6 nodes face dual attacks: local and Internet-based. Moreover, constructing an active cyber defense system for IPv6-based wireless sensor networks is difficult. In this article, we propose a K-nearest neighbor-based intrusion detection mechanism and design a secure network framework. This mechanism trains an intrusion detection algorithm using a feature data set to create a normal profile. The intrusion detection algorithm uses the normal profile to perform real-time detection of network traffic data to achieve rapid detections connecting many devices in a wireless sensor network. In addition, we develop a test platform to verify this mechanism. Experimental results show that this mechanism is appropriate for IPv6-based wireless sensor networks and achieves a low false-positive rate and good intrusion detection accuracy at an acceptable resource cost.
{"title":"An intrusion detection mechanism for IPv6-based wireless sensor networks","authors":"Min Wei, Chunmeng Rong, Erxiong Liang, Yuan Zhuang","doi":"10.1177/15501329221077922","DOIUrl":"https://doi.org/10.1177/15501329221077922","url":null,"abstract":"With the advancement of IPv6 technology, many nodes in wireless sensor networks realize seamless connections with the Internet via IPv6 addresses. Security issues are a significant obstacle to the widespread adoption of IPv6 technology. Resource-constrained IPv6 nodes face dual attacks: local and Internet-based. Moreover, constructing an active cyber defense system for IPv6-based wireless sensor networks is difficult. In this article, we propose a K-nearest neighbor-based intrusion detection mechanism and design a secure network framework. This mechanism trains an intrusion detection algorithm using a feature data set to create a normal profile. The intrusion detection algorithm uses the normal profile to perform real-time detection of network traffic data to achieve rapid detections connecting many devices in a wireless sensor network. In addition, we develop a test platform to verify this mechanism. Experimental results show that this mechanism is appropriate for IPv6-based wireless sensor networks and achieves a low false-positive rate and good intrusion detection accuracy at an acceptable resource cost.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43772315","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-01DOI: 10.1177/15501329221080665
Pingshu Ge, Lie Guo, Danni He, Liang Huang
Vehicle detection is one of the most challenging research works on environment perception for intelligent vehicle. The commonly used object detection network is too large and can only be realized in real-time on a high-performance server. Based on YOLOv3-tiny, the feature extraction was realized using light-weighted networks such as DarkNet-19 and ResNet-18 to improve accuracy. The K-means algorithm was used to cluster nine anchor boxes to achieve multi-scale prediction, especially for small targets. For automotive applicable scenarios, the proposed vehicle detection network was executed in an embedded device. The KITTI data sets were trained and tested. Experimental results show that the average accuracy is improved by 14.09% compared with the traditional YOLOv3-tiny, reaching 93.66%, and can reach 13 fps on the embedded device.
{"title":"Light-weighted vehicle detection network based on improved YOLOv3-tiny","authors":"Pingshu Ge, Lie Guo, Danni He, Liang Huang","doi":"10.1177/15501329221080665","DOIUrl":"https://doi.org/10.1177/15501329221080665","url":null,"abstract":"Vehicle detection is one of the most challenging research works on environment perception for intelligent vehicle. The commonly used object detection network is too large and can only be realized in real-time on a high-performance server. Based on YOLOv3-tiny, the feature extraction was realized using light-weighted networks such as DarkNet-19 and ResNet-18 to improve accuracy. The K-means algorithm was used to cluster nine anchor boxes to achieve multi-scale prediction, especially for small targets. For automotive applicable scenarios, the proposed vehicle detection network was executed in an embedded device. The KITTI data sets were trained and tested. Experimental results show that the average accuracy is improved by 14.09% compared with the traditional YOLOv3-tiny, reaching 93.66%, and can reach 13 fps on the embedded device.","PeriodicalId":50327,"journal":{"name":"International Journal of Distributed Sensor Networks","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2022-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41637973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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：