Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952379
Hong-Bae Jeon, C. Chae
In this paper, we propose a novel wireless backhaul architecture, mounted on an aerial platform, and enabled by an active reconfigurable intelligent surface (RIS). We assume that an immediate traffic increase occurs in urban area, and to serve the ground users therein authorities rapidly deploy unmanned-aerial-vehicle base-stations (UAV-BSs). We efficiently optimize the phase and the 3D placement of aerial active RIS, which leads to an increase of energy-efficiency under guaranteeing the reliable link quality. We verify the performance of the proposed algorithm via extensive numerical evaluations.
{"title":"Energy-Efficient Aerial-RIS Deployment for 6G","authors":"Hong-Bae Jeon, C. Chae","doi":"10.1109/ICTC55196.2022.9952379","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952379","url":null,"abstract":"In this paper, we propose a novel wireless backhaul architecture, mounted on an aerial platform, and enabled by an active reconfigurable intelligent surface (RIS). We assume that an immediate traffic increase occurs in urban area, and to serve the ground users therein authorities rapidly deploy unmanned-aerial-vehicle base-stations (UAV-BSs). We efficiently optimize the phase and the 3D placement of aerial active RIS, which leads to an increase of energy-efficiency under guaranteeing the reliable link quality. We verify the performance of the proposed algorithm via extensive numerical evaluations.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133767013","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952597
Junghoon Ha, Seunghoon Lee, Jinkyu Lee
Amongst densely populated areas such as New York, Hong Kong, Seoul and etc., apartments and condominiums grew to be the most commonly used type of housing units. Throughout the years we have witnessed many issues within the residents living in such units suffer from apartment noise. Lousy neighbors are not the only cause of such problems but the way houses were built, more specifically type and thickness of the flooring materials, determines the amount of noise transfer from one household to the other. In this paper, we show feasibility of classifying flooring materials using COTS (Commercial Off-The-Shelf) mobile devices. We exert each flooring materials propagating synthetic chirps in various ways. Our approach makes it possible to distinguish three types of target flooring materials (wood, polystyrene and concrete) and can also give similar results when the target materials are piled on top of one another. To this end we i) design an acoustic signal which effectively differentiates each target flooring materials, ii) gather sound samples propagated from the target materials, and iii) provide a classification methodology using methods such as SVM and KNN. Finally, we discuss the necessities to achieve the final goal of identifying flooring materials and their thickness.
{"title":"Towards Flooring Material Classification Using Acoustic Signal from COTS Mobile Devices","authors":"Junghoon Ha, Seunghoon Lee, Jinkyu Lee","doi":"10.1109/ICTC55196.2022.9952597","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952597","url":null,"abstract":"Amongst densely populated areas such as New York, Hong Kong, Seoul and etc., apartments and condominiums grew to be the most commonly used type of housing units. Throughout the years we have witnessed many issues within the residents living in such units suffer from apartment noise. Lousy neighbors are not the only cause of such problems but the way houses were built, more specifically type and thickness of the flooring materials, determines the amount of noise transfer from one household to the other. In this paper, we show feasibility of classifying flooring materials using COTS (Commercial Off-The-Shelf) mobile devices. We exert each flooring materials propagating synthetic chirps in various ways. Our approach makes it possible to distinguish three types of target flooring materials (wood, polystyrene and concrete) and can also give similar results when the target materials are piled on top of one another. To this end we i) design an acoustic signal which effectively differentiates each target flooring materials, ii) gather sound samples propagated from the target materials, and iii) provide a classification methodology using methods such as SVM and KNN. Finally, we discuss the necessities to achieve the final goal of identifying flooring materials and their thickness.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133789882","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the outbreak of the covid-19 pandemic in recent years, Video Stream Analytics technology quickly became a hot topic of discussion across technology forums. As it has appeared, in the pandemic situation in recent years, the use of masks when interacting with the community is a must, that's why the research works on mask identification today and more. receiving more and more attention. Understanding the situation, the team conducted facial recognition analysis inside the video to determine if the people appearing in the video were wearing masks. to then apply the trained model into practice. After a period of research, the team has also successfully built a mask recognition system that can generate images and can display the results as real-time video. Especially, the model is trained successful using systemml machine learning system. This is considered a positive result with real-time masked face recognition analysis.
{"title":"Face Mask Wearing Recognition System for Big Data Video Streaming","authors":"Van-Phuc-Le, Thuy-Ngoc Bui, Thanh-Truc Nguyen, Cao-Tien Do, Trong-Hop Do, Arooj Masood, Sungrae Cho","doi":"10.1109/ICTC55196.2022.9952941","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952941","url":null,"abstract":"With the outbreak of the covid-19 pandemic in recent years, Video Stream Analytics technology quickly became a hot topic of discussion across technology forums. As it has appeared, in the pandemic situation in recent years, the use of masks when interacting with the community is a must, that's why the research works on mask identification today and more. receiving more and more attention. Understanding the situation, the team conducted facial recognition analysis inside the video to determine if the people appearing in the video were wearing masks. to then apply the trained model into practice. After a period of research, the team has also successfully built a mask recognition system that can generate images and can display the results as real-time video. Especially, the model is trained successful using systemml machine learning system. This is considered a positive result with real-time masked face recognition analysis.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115435311","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952506
Sanjaya Samuel Ady, Suyoto
Covid-19 began in 2019 until now in 2022. It is still taking place in various countries, including Asia. It is necessary to lock down or limit citizens' activities to prevent the virus's spread. The imposition of lockdowns for several countries causes the mobility of citizens to be modified in many sectors. However, with vaccination in stages starting in 2021, people will become more resistant to the risk of being exposed to or dying from Covid-19. The process of transforming the return of activity before the pandemic needs to be observed, especially how countries' readiness, especially in Asia, to accept Covid-19 as endemic. The data used in this study are data on Covid-19 cases and Vaccinations by WHO (World Health Organization) and also Mobility Data from Google. This data will be used to conduct observations and Spatio-temporal analysis to see the development of cases, vaccinations, and mobility of citizens for various sectors. This research will include several stages of study, such as descriptive statistics, correlation analysis, and the last is trend analysis. The results of this study for countries in Asia are 21.87% have mobility above the baseline since 2021. Next, 46.87% of countries in Asia have mobility, with an increasing trend in 2022. Only 18.75% of countries in Asia have stable mobility. Below the baseline, there are also about 12.5% of countries in Asia that are around the baseline. In general, countries in Asia have an excellent response to the imposition of an endemic status for Covid-19.
{"title":"The transformation from Pandemic to Endemic of Covid-19: Spatio-temporal Analysis of Citizen Mobility in Asia Countries","authors":"Sanjaya Samuel Ady, Suyoto","doi":"10.1109/ICTC55196.2022.9952506","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952506","url":null,"abstract":"Covid-19 began in 2019 until now in 2022. It is still taking place in various countries, including Asia. It is necessary to lock down or limit citizens' activities to prevent the virus's spread. The imposition of lockdowns for several countries causes the mobility of citizens to be modified in many sectors. However, with vaccination in stages starting in 2021, people will become more resistant to the risk of being exposed to or dying from Covid-19. The process of transforming the return of activity before the pandemic needs to be observed, especially how countries' readiness, especially in Asia, to accept Covid-19 as endemic. The data used in this study are data on Covid-19 cases and Vaccinations by WHO (World Health Organization) and also Mobility Data from Google. This data will be used to conduct observations and Spatio-temporal analysis to see the development of cases, vaccinations, and mobility of citizens for various sectors. This research will include several stages of study, such as descriptive statistics, correlation analysis, and the last is trend analysis. The results of this study for countries in Asia are 21.87% have mobility above the baseline since 2021. Next, 46.87% of countries in Asia have mobility, with an increasing trend in 2022. Only 18.75% of countries in Asia have stable mobility. Below the baseline, there are also about 12.5% of countries in Asia that are around the baseline. In general, countries in Asia have an excellent response to the imposition of an endemic status for Covid-19.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115727444","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9953024
Mohtasin Golam, Rubina Akter, Esmot Ara Tuli, Dong‐Seong Kim, Jae-Min Lee
Unmanned aerial vehicles (UAVs) offer important strategic advantages in the Internet of Military Things (IoMT). On the other hand, the use of the UAV by an untrusted party may threaten security and perhaps endanger the vital operation of the IoMT network. The IoMT network additionally faces significant challenges from data falsification and manipulation through unauthorised access. For the purpose of addressing these challenges, this research presents a blockchain-based architecture and user authentication mechanism for the purpose of identifying and keeping track of unauthorized UAVs in the IoMT network. Blockchain technology not only stores data on the central control server but also forbids unauthorized access, data manipulation, and illegal intrusions. Additionally, a unique consensus model for UAV communication has been created. By streamlining the resource-constrained mining process, the created consensus can improve the efficiency of blockchain transactions between UAVs.
{"title":"Lightweight Blockchain Assisted Unauthorized UAV Access Prevention in the Internet of Military Things","authors":"Mohtasin Golam, Rubina Akter, Esmot Ara Tuli, Dong‐Seong Kim, Jae-Min Lee","doi":"10.1109/ICTC55196.2022.9953024","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9953024","url":null,"abstract":"Unmanned aerial vehicles (UAVs) offer important strategic advantages in the Internet of Military Things (IoMT). On the other hand, the use of the UAV by an untrusted party may threaten security and perhaps endanger the vital operation of the IoMT network. The IoMT network additionally faces significant challenges from data falsification and manipulation through unauthorised access. For the purpose of addressing these challenges, this research presents a blockchain-based architecture and user authentication mechanism for the purpose of identifying and keeping track of unauthorized UAVs in the IoMT network. Blockchain technology not only stores data on the central control server but also forbids unauthorized access, data manipulation, and illegal intrusions. Additionally, a unique consensus model for UAV communication has been created. By streamlining the resource-constrained mining process, the created consensus can improve the efficiency of blockchain transactions between UAVs.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124333865","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952667
Jinwuk Seok, Chang-Sik Cho
We propose a stochastic differential equation for a quantization-based optimization algorithm. A fundamental differential equation describes the state transition by an algorithm to analyze the dynamics of an optimization algorithm. According to the White Noise Hypothesis of quantization error with dense and uniform distribution, we can regard the quantization error as an identically independent distribution(i.i.d.) white noise. It leads that we can obtain a stochastic equation about the quantization-based optimization algorithm to analyze the global dynamics. Numerical experiments show that the proposed algorithm involves better performance than the conventional optimization algorithm, such as simulated annealing.
{"title":"Stochastic Differential Equation of the Quantization based Optimization","authors":"Jinwuk Seok, Chang-Sik Cho","doi":"10.1109/ICTC55196.2022.9952667","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952667","url":null,"abstract":"We propose a stochastic differential equation for a quantization-based optimization algorithm. A fundamental differential equation describes the state transition by an algorithm to analyze the dynamics of an optimization algorithm. According to the White Noise Hypothesis of quantization error with dense and uniform distribution, we can regard the quantization error as an identically independent distribution(i.i.d.) white noise. It leads that we can obtain a stochastic equation about the quantization-based optimization algorithm to analyze the global dynamics. Numerical experiments show that the proposed algorithm involves better performance than the conventional optimization algorithm, such as simulated annealing.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124420574","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952888
Won Gi Choi, Sohyeon Kim, Jeehyeong Kim, Minhwan Song, Sangshin Lee
According to the interest in digital transformation, the requirements for data processing to handle real-world data in real-time steadily raised. Though the efforts for standardization to achieve interoperability among IoT ecosystems have emerged and the opportunity for digital transformation is provided, it is still difficult for service applications to handle the digital values from sensors only based on a resource model optimized for device management. We proposed a user-friendly real-time processing framework that supports oneM2M IoT resources transformation to effectively provide data of time and spatial characteristics to the services. We also suggested two types of demo applications to verify usability.
{"title":"Real-Time Data Processing Framework for Things with time-series and spatial features","authors":"Won Gi Choi, Sohyeon Kim, Jeehyeong Kim, Minhwan Song, Sangshin Lee","doi":"10.1109/ICTC55196.2022.9952888","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952888","url":null,"abstract":"According to the interest in digital transformation, the requirements for data processing to handle real-world data in real-time steadily raised. Though the efforts for standardization to achieve interoperability among IoT ecosystems have emerged and the opportunity for digital transformation is provided, it is still difficult for service applications to handle the digital values from sensors only based on a resource model optimized for device management. We proposed a user-friendly real-time processing framework that supports oneM2M IoT resources transformation to effectively provide data of time and spatial characteristics to the services. We also suggested two types of demo applications to verify usability.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114491341","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952400
Vivian Ukamaka Ihekoronye, S. Ajakwe, Dong‐Seong Kim, Jae-Min Lee
The synchronization of swarms of drones (also known as unmanned aerial vehicles (UAV)) in a network can be attributed to their high mobility and maneuverability capabilities, making them deployable for time-critical operations such as security surveillance, disaster management, and search and rescue operations. However, the resource constraints of these flying robots are limitations to their functionalities. Likewise, the neglect of the security status of this network significantly promotes attacks by invaders, thus, thwarting the mission of this network. In this study, mobile edge computing (MEC) technology and anomaly-based intrusion detection scheme are leveraged to curb these challenges using an optimized Random Forest (RCSV) model embedded in dedicated UAV-MEC servers. The selection of prominent features and hyperparameters for modeling an optimized attack predictor is enabled by Pearson correlation coefficient (PCC) and randomized search cross-validation techniques. Also, the training and evaluation of the proposed model were achieved using intrusion detection data set (CICIDS2017 data set) comprised of complex network attack types. The simulation results obtained by the model in the detection and classification of the different attacks in the network (accuracy = 99.87%, precision = 99.32%, recall = 98.81 % and F1-score = 99.06%) shows its superiority over other optimized machine learning models and some existing models utilized in previous research.
{"title":"Cyber Edge Intelligent Intrusion Detection Framework For UAV Network Based on Random Forest Algorithm","authors":"Vivian Ukamaka Ihekoronye, S. Ajakwe, Dong‐Seong Kim, Jae-Min Lee","doi":"10.1109/ICTC55196.2022.9952400","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952400","url":null,"abstract":"The synchronization of swarms of drones (also known as unmanned aerial vehicles (UAV)) in a network can be attributed to their high mobility and maneuverability capabilities, making them deployable for time-critical operations such as security surveillance, disaster management, and search and rescue operations. However, the resource constraints of these flying robots are limitations to their functionalities. Likewise, the neglect of the security status of this network significantly promotes attacks by invaders, thus, thwarting the mission of this network. In this study, mobile edge computing (MEC) technology and anomaly-based intrusion detection scheme are leveraged to curb these challenges using an optimized Random Forest (RCSV) model embedded in dedicated UAV-MEC servers. The selection of prominent features and hyperparameters for modeling an optimized attack predictor is enabled by Pearson correlation coefficient (PCC) and randomized search cross-validation techniques. Also, the training and evaluation of the proposed model were achieved using intrusion detection data set (CICIDS2017 data set) comprised of complex network attack types. The simulation results obtained by the model in the detection and classification of the different attacks in the network (accuracy = 99.87%, precision = 99.32%, recall = 98.81 % and F1-score = 99.06%) shows its superiority over other optimized machine learning models and some existing models utilized in previous research.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114863458","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952986
Yeunwoong Kyung, Haneul Ko
Initially, users use internet of things (IoT) services to remotely monitor or control things through the internet. However, according to the development of IoT technologies, there have been various IoT services which require some strict requirements such as real-time response or high computing processing. In addition, the intelligent computing for the services has been needed in these days. To cover these services, the communication networks and computing paradigms have been changed. Conventionally, IoT services have been usually supported in the cloud. Then, the edge cloud or computing has been considered for the real-time response. Finally, the computing in the device by itself such as a tiny machine learning concept has been recently introduced. Therefore, this paper reviews the changes in communications networks and computing paradigms from the cloud to the edge and device computing for the intelligent IoT services.
{"title":"Changes in Communication Networks and Computing for Intelligent IoT Services from Cloud to Edge","authors":"Yeunwoong Kyung, Haneul Ko","doi":"10.1109/ICTC55196.2022.9952986","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952986","url":null,"abstract":"Initially, users use internet of things (IoT) services to remotely monitor or control things through the internet. However, according to the development of IoT technologies, there have been various IoT services which require some strict requirements such as real-time response or high computing processing. In addition, the intelligent computing for the services has been needed in these days. To cover these services, the communication networks and computing paradigms have been changed. Conventionally, IoT services have been usually supported in the cloud. Then, the edge cloud or computing has been considered for the real-time response. Finally, the computing in the device by itself such as a tiny machine learning concept has been recently introduced. Therefore, this paper reviews the changes in communications networks and computing paradigms from the cloud to the edge and device computing for the intelligent IoT services.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114911868","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-19DOI: 10.1109/ICTC55196.2022.9952987
Suhwan Kim, Sehun Jung, Hyang-Won Lee
With the development of artificial intelligence (AI) technology, many applications are providing AI services. The key part of these AI services is the Deep Neural Networks(DNNs) requiring a lot of computation. However, it is usually time-consuming to provide an inference process on end devices that lack resources. Because of these limitations, distributed computing, which can perform large amounts of calculations using the processing power of various computers connected to the Internet, is emerging. We develop how to efficiently distribute DNN inference jobs in distributed computing environments and quickly process large amounts of DNN computations. In this paper, we will introduce the learning method and the results of the Deep Reinforcement Learning(DRL) model to reduce end-to-end latency by observing the state of the distributed computing environment and scheduling the DNN job using DRL.
{"title":"Reducing DNN inference latency using DRL","authors":"Suhwan Kim, Sehun Jung, Hyang-Won Lee","doi":"10.1109/ICTC55196.2022.9952987","DOIUrl":"https://doi.org/10.1109/ICTC55196.2022.9952987","url":null,"abstract":"With the development of artificial intelligence (AI) technology, many applications are providing AI services. The key part of these AI services is the Deep Neural Networks(DNNs) requiring a lot of computation. However, it is usually time-consuming to provide an inference process on end devices that lack resources. Because of these limitations, distributed computing, which can perform large amounts of calculations using the processing power of various computers connected to the Internet, is emerging. We develop how to efficiently distribute DNN inference jobs in distributed computing environments and quickly process large amounts of DNN computations. In this paper, we will introduce the learning method and the results of the Deep Reinforcement Learning(DRL) model to reduce end-to-end latency by observing the state of the distributed computing environment and scheduling the DNN job using DRL.","PeriodicalId":441404,"journal":{"name":"2022 13th International Conference on Information and Communication Technology Convergence (ICTC)","volume":"15 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116875787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：