Pub Date : 2021-12-01DOI: 10.1109/CIC52973.2021.00012
Ishan Ranasinghe, Chengping Yuan, R. Dantu, Mark V. Albert
Maintaining motivation to meet physical exercise goals is a big challenge in virtual/home-based exercise guidance systems. Lack of motivation, long-maintained bad daily routines, and fear of injury are some of the reasons that cause this hesitation. This paper proposes a reinforcement learning-based virtual exercise assistant capable of providing encouragement and customized feedback on body movement form over time. Repeated arm curls were observed and tracked using single and dual-camera systems using the Posenet pose estimation library. To accumulate enough experience across individuals, the reinforcement learning model was collaboratively trained by subjects. The proposed system is tested on 36 subjects. Behavioral changes are apparent in 31 of the 36 subjects, with 31 subjects reducing movement errors over time and 15 subjects completely eliminating the errors. The system was analyzed for which types of feedback provided the highest expected value, and feedback directly related to the previous mistake provided the highest valued feedback ($p < 0.0133$). The result showed that the Reinforcement Learning system provides meaningful feedback and positively impacts behavior progress.
{"title":"A Collaborative and Adaptive Feedback System for Physical Exercises","authors":"Ishan Ranasinghe, Chengping Yuan, R. Dantu, Mark V. Albert","doi":"10.1109/CIC52973.2021.00012","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00012","url":null,"abstract":"Maintaining motivation to meet physical exercise goals is a big challenge in virtual/home-based exercise guidance systems. Lack of motivation, long-maintained bad daily routines, and fear of injury are some of the reasons that cause this hesitation. This paper proposes a reinforcement learning-based virtual exercise assistant capable of providing encouragement and customized feedback on body movement form over time. Repeated arm curls were observed and tracked using single and dual-camera systems using the Posenet pose estimation library. To accumulate enough experience across individuals, the reinforcement learning model was collaboratively trained by subjects. The proposed system is tested on 36 subjects. Behavioral changes are apparent in 31 of the 36 subjects, with 31 subjects reducing movement errors over time and 15 subjects completely eliminating the errors. The system was analyzed for which types of feedback provided the highest expected value, and feedback directly related to the previous mistake provided the highest valued feedback ($p < 0.0133$). The result showed that the Reinforcement Learning system provides meaningful feedback and positively impacts behavior progress.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"696 19","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120882053","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 : 2021-12-01DOI: 10.1109/cic52973.2021.00008
{"title":"Steering Committee CIC 2021","authors":"","doi":"10.1109/cic52973.2021.00008","DOIUrl":"https://doi.org/10.1109/cic52973.2021.00008","url":null,"abstract":"","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125519757","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 : 2021-12-01DOI: 10.1109/CIC52973.2021.00021
Yudong Tao, Renhe Jiang, Erik Coltey, Chuang Yang, Xuan Song, R. Shibasaki, M. Shyu, Shu‐Ching Chen
Since 2019, the world has been seriously impacted by the global pandemic, COVID-19, with millions of people adversely affected. This is coupled with a trend in which the intensity and frequency of natural disasters such as hurricanes, wildfires, and earthquakes have increased over the past decades. Larger and more diverse communities have been negatively influenced by these disasters and they might encounter crises socially and/or economically, further exacerbated when the natural disasters and pandemics co-occurred. However, conventional disaster response and management rely on human surveys and case studies to identify these in-crisis communities and their problems, which might not be effective and efficient due to the scale of the impacted population. In this paper, we propose to utilize the data-driven techniques and recent advances in artificial intelligence to automate the in-crisis community identification and improve its scalability and efficiency. Thus, immediate assistance to the in-crisis communities can be provided by society and timely disaster response and management can be achieved. A novel framework of the in-crisis community identification has been presented, which can be divided into three subtasks: (1) community detection, (2) in-crisis status detection, and (3) community demand and problem identification. Furthermore, the open issues and challenges toward automated in-crisis community identification are discussed to motivate future research and innovations in the area.
{"title":"Data-Driven In-Crisis Community Identification for Disaster Response and Management","authors":"Yudong Tao, Renhe Jiang, Erik Coltey, Chuang Yang, Xuan Song, R. Shibasaki, M. Shyu, Shu‐Ching Chen","doi":"10.1109/CIC52973.2021.00021","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00021","url":null,"abstract":"Since 2019, the world has been seriously impacted by the global pandemic, COVID-19, with millions of people adversely affected. This is coupled with a trend in which the intensity and frequency of natural disasters such as hurricanes, wildfires, and earthquakes have increased over the past decades. Larger and more diverse communities have been negatively influenced by these disasters and they might encounter crises socially and/or economically, further exacerbated when the natural disasters and pandemics co-occurred. However, conventional disaster response and management rely on human surveys and case studies to identify these in-crisis communities and their problems, which might not be effective and efficient due to the scale of the impacted population. In this paper, we propose to utilize the data-driven techniques and recent advances in artificial intelligence to automate the in-crisis community identification and improve its scalability and efficiency. Thus, immediate assistance to the in-crisis communities can be provided by society and timely disaster response and management can be achieved. A novel framework of the in-crisis community identification has been presented, which can be divided into three subtasks: (1) community detection, (2) in-crisis status detection, and (3) community demand and problem identification. Furthermore, the open issues and challenges toward automated in-crisis community identification are discussed to motivate future research and innovations in the area.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130414989","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 : 2021-12-01DOI: 10.1109/CIC52973.2021.00026
Jinlai Xu, Balaji Palanisamy
The number of Internet-of-Things (IoT) devices is rapidly increasing with the growth of IoT applications in various domains. As IoT applications have a strong demand for low latency and high throughput computing, stream processing using edge computing resources is a promising approach to support low latency processing of large-scale data. Edge-based stream processing extends the capability of cloud-based stream processing by processing the data streams near the edge of the network. In this vision paper, we discuss a distributed stream processing framework that optimizes the performance of stream processing applications through a careful allocation of geo-distributed computing and network resources available in edge computing environments. The framework includes key optimizations in both the platform layer and the infrastructure layer. While the platform layer is responsible for converting the user program into a stream processing physical plan and optimizing the physical plan and operator placement, the infrastructure layer is responsible for provisioning geo-distributed resources to the platform layer. The framework optimizes the performance of stream query processing at the platform layer through its careful consideration of data locality and resource constraints during physical plan generation and operator placement and by incorporating resilience to deal with failures. The framework also includes techniques to dynamically determine the level of parallelism to adapt to changing workload conditions. At the infrastructure layer, the framework includes a novel model for allocating computing resources in edge and geo-distributed cloud computing environments by carefully considering latency and cost. End users benefit from the platform through reduced cost and improved user experience in terms of response time and latency.
{"title":"Cost-aware & Fault-tolerant Geo-distributed Edge Computing for Low-latency Stream Processing","authors":"Jinlai Xu, Balaji Palanisamy","doi":"10.1109/CIC52973.2021.00026","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00026","url":null,"abstract":"The number of Internet-of-Things (IoT) devices is rapidly increasing with the growth of IoT applications in various domains. As IoT applications have a strong demand for low latency and high throughput computing, stream processing using edge computing resources is a promising approach to support low latency processing of large-scale data. Edge-based stream processing extends the capability of cloud-based stream processing by processing the data streams near the edge of the network. In this vision paper, we discuss a distributed stream processing framework that optimizes the performance of stream processing applications through a careful allocation of geo-distributed computing and network resources available in edge computing environments. The framework includes key optimizations in both the platform layer and the infrastructure layer. While the platform layer is responsible for converting the user program into a stream processing physical plan and optimizing the physical plan and operator placement, the infrastructure layer is responsible for provisioning geo-distributed resources to the platform layer. The framework optimizes the performance of stream query processing at the platform layer through its careful consideration of data locality and resource constraints during physical plan generation and operator placement and by incorporating resilience to deal with failures. The framework also includes techniques to dynamically determine the level of parallelism to adapt to changing workload conditions. At the infrastructure layer, the framework includes a novel model for allocating computing resources in edge and geo-distributed cloud computing environments by carefully considering latency and cost. End users benefit from the platform through reduced cost and improved user experience in terms of response time and latency.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122391523","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 : 2021-12-01DOI: 10.1109/cic52973.2021.00007
{"title":"Technical Program Committee CIC 2021","authors":"","doi":"10.1109/cic52973.2021.00007","DOIUrl":"https://doi.org/10.1109/cic52973.2021.00007","url":null,"abstract":"","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"abs/2211.01143 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124213926","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 : 2021-12-01DOI: 10.1109/CIC52973.2021.00020
Sara Jafarbeiki, R. Gaire, A. Sakzad, Shabnam Kasra Kermanshahi, Ron Steinfeld
The cost of DNA sequencing has resulted in a surge of genetic data being utilised to improve scientific research, clinical procedures, and healthcare delivery in recent years. Since the human genome can uniquely identify an individual, this characteristic also raises security and privacy concerns. In order to balance the risks and benefits, governance mechanisms including regulatory and ethical controls have been established, which are prone to human errors and create hindrance for collaboration. Over the past decade, technological methods are also catching up that can support critical discoveries responsibly. In this paper, we explore regulations and ethical guidelines and propose our visions of secure/private genomic data storage/processing/sharing platforms. Then, we present some available techniques and a conceptual system model that can support our visions. Finally, we highlight the open issues that need further investigation.
{"title":"Collaborative analysis of genomic data: vision and challenges","authors":"Sara Jafarbeiki, R. Gaire, A. Sakzad, Shabnam Kasra Kermanshahi, Ron Steinfeld","doi":"10.1109/CIC52973.2021.00020","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00020","url":null,"abstract":"The cost of DNA sequencing has resulted in a surge of genetic data being utilised to improve scientific research, clinical procedures, and healthcare delivery in recent years. Since the human genome can uniquely identify an individual, this characteristic also raises security and privacy concerns. In order to balance the risks and benefits, governance mechanisms including regulatory and ethical controls have been established, which are prone to human errors and create hindrance for collaboration. Over the past decade, technological methods are also catching up that can support critical discoveries responsibly. In this paper, we explore regulations and ethical guidelines and propose our visions of secure/private genomic data storage/processing/sharing platforms. Then, we present some available techniques and a conceptual system model that can support our visions. Finally, we highlight the open issues that need further investigation.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130952872","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 : 2021-12-01DOI: 10.1109/cic52973.2021.00001
{"title":"2021 IEEE 7th International Conference on Collaboration and Internet Computing CIC 2021","authors":"","doi":"10.1109/cic52973.2021.00001","DOIUrl":"https://doi.org/10.1109/cic52973.2021.00001","url":null,"abstract":"","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121208343","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 : 2021-12-01DOI: 10.1109/CIC52973.2021.00025
Dimitrios Georgakopoulos, Dinithi Bamunuarachchi
Industry 4.0 applications integrate and analyze information from industrial machines, people involved in production, products, and processes to determine how to improve production efficiency, enhance product consistency, and reduce unplanned maintenance in industrial plants. However, Industry 4.0 application development is currently complex and expensive due to the lack of effective representations of complex industrial machines in the cyberspace, and the limited support of IoT platforms in using such representations to monitor, predict, and mange production outcomes. To simplify the complexity and reduce the cost of Industry 4.0 application development, this paper proposes Cyber Twins (CTs), which is a variant of digital twins that is specifically designed for manufacturing, for representing complex industrial machines and providing the building blocks for Industry 4.0 application development. Furthermore, the paper proposes a platform for CT-based Industry 4.0 application development that overcomes the complexity and cost limitations of existing IoT platforms in developing Industry 4.0 applications. The paper provides examples of CTs for industrial machines and Industry 4.0 applications that demonstrate the benefits of CT-based Industry 4.0 application development via the proposed platform.
{"title":"Digital Twins-based Application Development for Digital Manufacturing","authors":"Dimitrios Georgakopoulos, Dinithi Bamunuarachchi","doi":"10.1109/CIC52973.2021.00025","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00025","url":null,"abstract":"Industry 4.0 applications integrate and analyze information from industrial machines, people involved in production, products, and processes to determine how to improve production efficiency, enhance product consistency, and reduce unplanned maintenance in industrial plants. However, Industry 4.0 application development is currently complex and expensive due to the lack of effective representations of complex industrial machines in the cyberspace, and the limited support of IoT platforms in using such representations to monitor, predict, and mange production outcomes. To simplify the complexity and reduce the cost of Industry 4.0 application development, this paper proposes Cyber Twins (CTs), which is a variant of digital twins that is specifically designed for manufacturing, for representing complex industrial machines and providing the building blocks for Industry 4.0 application development. Furthermore, the paper proposes a platform for CT-based Industry 4.0 application development that overcomes the complexity and cost limitations of existing IoT platforms in developing Industry 4.0 applications. The paper provides examples of CTs for industrial machines and Industry 4.0 applications that demonstrate the benefits of CT-based Industry 4.0 application development via the proposed platform.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"249 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124740386","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 : 2021-11-24DOI: 10.1109/CIC52973.2021.00013
Deepti Gupta, O. Kayode, Smriti Bhatt, Maanak Gupta, A. Tosun
Internet of Medical Things (IoMT) is becoming ubiquitous with a proliferation of smart medical devices and applications used in smart hospitals, smart-home based care, and nursing homes. It utilizes smart medical devices and cloud computing services along with core Internet of Things (IoT) technologies to sense patients' vital body parameters, monitor health conditions and generate multivariate data to support just-in-time health services. Mostly, this large amount of data is analyzed in centralized servers. Anomaly Detection (AD) in a centralized healthcare ecosystem is often plagued by significant delays in response time with high performance overhead. Moreover, there are inherent privacy issues associated with sending patients' personal health data to a centralized server, which may also introduce several security threats to the AD model, such as possibility of data poisoning. To overcome these issues with centralized AD models, here we propose a Federated Learning (FL) based AD model which utilizes edge cloudlets to run AD models locally without sharing patients' data. Since existing FL approaches perform aggregation on a single server which restricts the scope of FL, in this paper, we introduce a hierarchical FL that allows aggregation at different levels enabling multi-party collaboration. We introduce a novel disease-based grouping mechanism where different AD models are grouped based on specific types of diseases. Furthermore, we develop a new Federated Time Distributed (FEDTIMEDIS) Long Short-Term Memory (LSTM) approach to train the AD model. We present a Remote Patient Monitoring (RPM) use case to demonstrate our model, and illustrate a proof-of-concept implementation using Digital Twin (DT) and edge cloudlets.
{"title":"Hierarchical Federated Learning based Anomaly Detection using Digital Twins for Smart Healthcare","authors":"Deepti Gupta, O. Kayode, Smriti Bhatt, Maanak Gupta, A. Tosun","doi":"10.1109/CIC52973.2021.00013","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00013","url":null,"abstract":"Internet of Medical Things (IoMT) is becoming ubiquitous with a proliferation of smart medical devices and applications used in smart hospitals, smart-home based care, and nursing homes. It utilizes smart medical devices and cloud computing services along with core Internet of Things (IoT) technologies to sense patients' vital body parameters, monitor health conditions and generate multivariate data to support just-in-time health services. Mostly, this large amount of data is analyzed in centralized servers. Anomaly Detection (AD) in a centralized healthcare ecosystem is often plagued by significant delays in response time with high performance overhead. Moreover, there are inherent privacy issues associated with sending patients' personal health data to a centralized server, which may also introduce several security threats to the AD model, such as possibility of data poisoning. To overcome these issues with centralized AD models, here we propose a Federated Learning (FL) based AD model which utilizes edge cloudlets to run AD models locally without sharing patients' data. Since existing FL approaches perform aggregation on a single server which restricts the scope of FL, in this paper, we introduce a hierarchical FL that allows aggregation at different levels enabling multi-party collaboration. We introduce a novel disease-based grouping mechanism where different AD models are grouped based on specific types of diseases. Furthermore, we develop a new Federated Time Distributed (FEDTIMEDIS) Long Short-Term Memory (LSTM) approach to train the AD model. We present a Remote Patient Monitoring (RPM) use case to demonstrate our model, and illustrate a proof-of-concept implementation using Digital Twin (DT) and edge cloudlets.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127612212","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 : 2021-04-25DOI: 10.1109/CIC52973.2021.00011
Juhyun Bae, Ling Liu, Yanzhao Wu, Gong Su, A. Iyengar
We present RDMAbox, a set of low level RDMA optimizations that provide better performance than previous approaches. The optimizations are packaged in easy-to-use kernel and user space libraries for applications and systems in data centers. We demonstrate the flexibility and effectiveness of RDMAbox by implementing a kernel remote paging system and a user space file system using RDMAbox. RDMAbox employs two optimization techniques. First, we suggest RDMA request merging and chaining to reduce the total number of I/O operations to the RDMA NIC. The I/O merge queue at the same time functions as a traffic regulator to enforce admission control and avoid overloading the NIC. Second, we propose Adaptive Polling to achieve higher efficiency of polling Work Completion than existing busy polling while maintaining the low CPU overhead of event trigger. Our implementation of a remote paging system with RDMAbox outperforms existing representative solutions with up to 4x throughput improvement and up to 83% decrease in average tail latency in bigdata workloads, and up to 83% reduction in completion time in machine learning workloads. Our implementation of a user space file system based on RDMAbox achieves up to 5.9x higher throughput over existing representative solutions.
{"title":"RDMAbox: Optimizing RDMA for Memory Intensive Workload","authors":"Juhyun Bae, Ling Liu, Yanzhao Wu, Gong Su, A. Iyengar","doi":"10.1109/CIC52973.2021.00011","DOIUrl":"https://doi.org/10.1109/CIC52973.2021.00011","url":null,"abstract":"We present RDMAbox, a set of low level RDMA optimizations that provide better performance than previous approaches. The optimizations are packaged in easy-to-use kernel and user space libraries for applications and systems in data centers. We demonstrate the flexibility and effectiveness of RDMAbox by implementing a kernel remote paging system and a user space file system using RDMAbox. RDMAbox employs two optimization techniques. First, we suggest RDMA request merging and chaining to reduce the total number of I/O operations to the RDMA NIC. The I/O merge queue at the same time functions as a traffic regulator to enforce admission control and avoid overloading the NIC. Second, we propose Adaptive Polling to achieve higher efficiency of polling Work Completion than existing busy polling while maintaining the low CPU overhead of event trigger. Our implementation of a remote paging system with RDMAbox outperforms existing representative solutions with up to 4x throughput improvement and up to 83% decrease in average tail latency in bigdata workloads, and up to 83% reduction in completion time in machine learning workloads. Our implementation of a user space file system based on RDMAbox achieves up to 5.9x higher throughput over existing representative solutions.","PeriodicalId":170121,"journal":{"name":"2021 IEEE 7th International Conference on Collaboration and Internet Computing (CIC)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-04-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121748658","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：