Pub Date : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00023
Praveen Venkateswaran, Cheng-Hsin Hsu, S. Mehrotra, N. Venkatasubramanian
An increasing number of community spaces are being instrumented with heterogeneous IoT sensors and actuators that enable continuous monitoring of the surrounding environments. Data streams generated from the devices are analyzed using a range of analytics operators and transformed into meaningful information for community monitoring applications. To ensure high quality results, timely monitoring, and application reliability, we argue that these operators must be hosted at edge servers located in close proximity to the community space. In this paper, we present a Resource Efficient Adaptive Monitoring (REAM) framework at the edge that adaptively selects workflows of devices and operators to maintain adequate quality of information for the application at hand while judiciously consuming the limited resources available on edge servers. IoT deployments in community spaces are in a state of continuous flux that are dictated by the nature of activities and events within the space. Since these spaces are complex and change dynamically, and events can take place under different environmental contexts, developing a one-size-fits-all model that works for all types of spaces is infeasible. The REAM framework utilizes deep reinforcement learning agents that learn by interacting with each individual community spaces and take decisions based on the state of the environment in each space and other contextual information. We evaluate our framework on two real-world testbeds in Orange County, USA and NTHU, Taiwan. The evaluation results show that community spaces using REAM can achieve > 90% monitoring accuracy while incurring ~ 50% less resource consumption costs compared to existing static monitoring and Machine Learning driven approaches.
{"title":"REAM: Resource Efficient Adaptive Monitoring of Community Spaces at the Edge Using Reinforcement Learning","authors":"Praveen Venkateswaran, Cheng-Hsin Hsu, S. Mehrotra, N. Venkatasubramanian","doi":"10.1109/SMARTCOMP50058.2020.00023","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00023","url":null,"abstract":"An increasing number of community spaces are being instrumented with heterogeneous IoT sensors and actuators that enable continuous monitoring of the surrounding environments. Data streams generated from the devices are analyzed using a range of analytics operators and transformed into meaningful information for community monitoring applications. To ensure high quality results, timely monitoring, and application reliability, we argue that these operators must be hosted at edge servers located in close proximity to the community space. In this paper, we present a Resource Efficient Adaptive Monitoring (REAM) framework at the edge that adaptively selects workflows of devices and operators to maintain adequate quality of information for the application at hand while judiciously consuming the limited resources available on edge servers. IoT deployments in community spaces are in a state of continuous flux that are dictated by the nature of activities and events within the space. Since these spaces are complex and change dynamically, and events can take place under different environmental contexts, developing a one-size-fits-all model that works for all types of spaces is infeasible. The REAM framework utilizes deep reinforcement learning agents that learn by interacting with each individual community spaces and take decisions based on the state of the environment in each space and other contextual information. We evaluate our framework on two real-world testbeds in Orange County, USA and NTHU, Taiwan. The evaluation results show that community spaces using REAM can achieve > 90% monitoring accuracy while incurring ~ 50% less resource consumption costs compared to existing static monitoring and Machine Learning driven approaches.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115986238","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00094
Takuya Suzuki, Yu Ishimaki, H. Yamana
Smart computing aims to improve the quality of life by utilizing Internet-of-Things devices and cloud computing. Typically, this computing handles private and/or personal information so concealing such sensitive information is a challenge. Adopting fully homomorphic encryption (FHE) is one approach for handling such sensitive information safely; that is, we can calculate the encrypted data without decryption. However, the time and space complexity of the FHE operation is high. Thus, its computation takes a long time. In this study, we aim to shorten FHE execution time by adopting our new scheduling algorithm, which divides a task into several macro-tasks and then assigns a set of threads. We assume a cloud computing system that is equipped with a many-core CPU. Thus, we propose the dynamic allocation method of computation resource to macro-tasks (DAMCREM), which dynamically allocates a certain number of threads (selected from pre-defined candidates) to each macro-task of every given job. In the evaluation, we compared DAMCREM to naive methods that allocate a pre-defined number of threads to each macro-task. The result shows that the average latency and maximum latency of job execution is less than those of naive methods, even when the average interval of job arrival is short.
{"title":"DAMCREM: Dynamic Allocation Method of Computation REsource to Macro-Tasks for Fully Homomorphic Encryption Applications","authors":"Takuya Suzuki, Yu Ishimaki, H. Yamana","doi":"10.1109/SMARTCOMP50058.2020.00094","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00094","url":null,"abstract":"Smart computing aims to improve the quality of life by utilizing Internet-of-Things devices and cloud computing. Typically, this computing handles private and/or personal information so concealing such sensitive information is a challenge. Adopting fully homomorphic encryption (FHE) is one approach for handling such sensitive information safely; that is, we can calculate the encrypted data without decryption. However, the time and space complexity of the FHE operation is high. Thus, its computation takes a long time. In this study, we aim to shorten FHE execution time by adopting our new scheduling algorithm, which divides a task into several macro-tasks and then assigns a set of threads. We assume a cloud computing system that is equipped with a many-core CPU. Thus, we propose the dynamic allocation method of computation resource to macro-tasks (DAMCREM), which dynamically allocates a certain number of threads (selected from pre-defined candidates) to each macro-task of every given job. In the evaluation, we compared DAMCREM to naive methods that allocate a pre-defined number of threads to each macro-task. The result shows that the average latency and maximum latency of job execution is less than those of naive methods, even when the average interval of job arrival is short.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"117 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116183316","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00085
Carlo Scaffidi, Giuseppe Tricomi, S. Distefano, A. Puliafito
Global warming and climate changes are due to several factors, not least vehicle and transportation emissions. Smart City technologies can provide mechanisms for emission (greenhouse gases, particles) containment that may significantly impact on the environment. This paper proposes a solution, based on an intelligent cruise control system, allowing a vehicle to interact with the Smart City infrastructure facilities for cutting down its emissions on the planned route while saving fuel. The proposed approach aims at implementing a (virtually) continuous green wave for a vehicle lowering its emissions by modulating the speed only considering local traffic congestion and traffic light information provided by the Smart City infrastructure, without actuating on the latter. A green-green (green2) wave also reducing the fuel consumption and ensuring a good trade off with travel time. To demonstrate the effectiveness of the proposed solution, a power train model of a c-segment car traveling on while interacting with Smart City facilities has been implemented and evaluated, providing significant insights.
{"title":"Continuous Green2 Waves for Surfin Smart Cities","authors":"Carlo Scaffidi, Giuseppe Tricomi, S. Distefano, A. Puliafito","doi":"10.1109/SMARTCOMP50058.2020.00085","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00085","url":null,"abstract":"Global warming and climate changes are due to several factors, not least vehicle and transportation emissions. Smart City technologies can provide mechanisms for emission (greenhouse gases, particles) containment that may significantly impact on the environment. This paper proposes a solution, based on an intelligent cruise control system, allowing a vehicle to interact with the Smart City infrastructure facilities for cutting down its emissions on the planned route while saving fuel. The proposed approach aims at implementing a (virtually) continuous green wave for a vehicle lowering its emissions by modulating the speed only considering local traffic congestion and traffic light information provided by the Smart City infrastructure, without actuating on the latter. A green-green (green2) wave also reducing the fuel consumption and ensuring a good trade off with travel time. To demonstrate the effectiveness of the proposed solution, a power train model of a c-segment car traveling on while interacting with Smart City facilities has been implemented and evaluated, providing significant insights.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127969237","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00073
Eugenio Cesario, Paschal I. Uchubilo, Andrea Vinci, Xiaotian Zhu
Leveraged by a large-scale diffusion of sensing networks and scanning devices in modern cities, huge volumes of geo-referenced urban data are collected every day. Such amount of information is analyzed to discover data-driven models, which can be exploited to tackle the major issues that cities face, including air pollution, virus diffusion, human mobility, traffic flows. In particular, the detection of city hotspots is becoming a valuable organization technique for framing detailed knowledge of a metropolitan area, providing high-level summaries for spatial datasets, which are valuable for planners, scientists, and policymakers. However, while classic density-based clustering algorithms show to be suitable to discover hotspots characterized by homogeneous density, their application on multi-density data can produce inaccurate results. For such a reason, since metropolitan cities are heavily characterized by variable densities, multi-density clustering seems to be more appropriate to discover city hotspots. This paper presents a study about how density-based clustering algorithms are suitable for discovering urban hotspots in a city, by showing a comparative analysis of single-density and multi-density clustering on both state-of-the-art data and real-world data. The experimental evaluation shows that, in an urban scenario, multi-density clustering achieves higher quality hotspots than a single-density approach.
{"title":"Discovering Multi-density Urban Hotspots in a Smart City","authors":"Eugenio Cesario, Paschal I. Uchubilo, Andrea Vinci, Xiaotian Zhu","doi":"10.1109/SMARTCOMP50058.2020.00073","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00073","url":null,"abstract":"Leveraged by a large-scale diffusion of sensing networks and scanning devices in modern cities, huge volumes of geo-referenced urban data are collected every day. Such amount of information is analyzed to discover data-driven models, which can be exploited to tackle the major issues that cities face, including air pollution, virus diffusion, human mobility, traffic flows. In particular, the detection of city hotspots is becoming a valuable organization technique for framing detailed knowledge of a metropolitan area, providing high-level summaries for spatial datasets, which are valuable for planners, scientists, and policymakers. However, while classic density-based clustering algorithms show to be suitable to discover hotspots characterized by homogeneous density, their application on multi-density data can produce inaccurate results. For such a reason, since metropolitan cities are heavily characterized by variable densities, multi-density clustering seems to be more appropriate to discover city hotspots. This paper presents a study about how density-based clustering algorithms are suitable for discovering urban hotspots in a city, by showing a comparative analysis of single-density and multi-density clustering on both state-of-the-art data and real-world data. The experimental evaluation shows that, in an urban scenario, multi-density clustering achieves higher quality hotspots than a single-density approach.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121773688","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00076
Francesco Alongi, Nicolò Ghielmetti, D. Pau, F. Terraneo, W. Fornaciari
Collecting vast amount of data and performing complex calculations to feed modern Numerical Weather Prediction (NWP) algorithms require to centralize intelligence into some of the most powerful energy and resource hungry supercomputers in the world. This is due to the chaotic complex nature of the atmosphere which interpretation require virtually unlimited computing and storage resources. With Machine Learning (ML) techniques, a statistical approach can be designed in order to perform weather forecasting activity. Moreover, the recently growing interest in Edge Computing Tiny Intelligent architectures is proposing a shift towards the deployment of ML algorithms on Tiny Embedded Systems (ES). This paper describes how Deep but Tiny Neural Networks (DTNN) can be designed to be parsimonious and can be automatically converted into a STM32 microcontroller-optimized C-library through X-CUBE-AI toolchain; we propose the integration of the obtained library with Miosix, a Real Time Operating System (RTOS) tailored for resource constrained and tiny processors, which is an enabling factor for system scalability and multi tasking. With our experiments we demonstrate that it is possible to deploy a DTNN, with a FLASH and RAM occupation of 45,5 KByte and 480 Byte respectively, for atmospheric pressure forecasting in an affordable cost effective system. We deployed the system in a real context, obtaining the same prediction quality as the same DNN model deployed on the cloud but with the advantage of processing all the necessary data to perform the prediction close to environmental sensors, avoiding raw data traffic to the cloud.
收集大量的数据和执行复杂的计算来提供现代数值天气预报(NWP)算法需要将智能集中到世界上一些最强大的能源和资源饥渴的超级计算机中。这是由于大气混乱复杂的性质，解释需要几乎无限的计算和存储资源。利用机器学习(ML)技术，可以设计统计方法来执行天气预报活动。此外，最近对边缘计算微型智能架构的兴趣日益浓厚，这提出了在微型嵌入式系统(ES)上部署机器学习算法的转变。本文介绍了如何通过X-CUBE-AI工具链将深度但微小的神经网络(Deep but Tiny Neural Networks, DTNN)设计得简洁，并可自动转换为STM32微控制器优化的c库;我们建议将获得的库与Miosix集成，Miosix是为资源受限和微型处理器量身定制的实时操作系统(RTOS)，这是系统可扩展性和多任务处理的有利因素。通过我们的实验，我们证明可以部署DTNN, FLASH和RAM分别为45,5 KByte和480字节，用于经济实惠的大气压力预测系统。我们将系统部署在真实环境中，获得与部署在云上的相同DNN模型相同的预测质量，但具有处理所有必要数据以接近环境传感器执行预测的优势，避免了原始数据流量到云。
{"title":"Tiny Neural Networks for Environmental Predictions: An Integrated Approach with Miosix","authors":"Francesco Alongi, Nicolò Ghielmetti, D. Pau, F. Terraneo, W. Fornaciari","doi":"10.1109/SMARTCOMP50058.2020.00076","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00076","url":null,"abstract":"Collecting vast amount of data and performing complex calculations to feed modern Numerical Weather Prediction (NWP) algorithms require to centralize intelligence into some of the most powerful energy and resource hungry supercomputers in the world. This is due to the chaotic complex nature of the atmosphere which interpretation require virtually unlimited computing and storage resources. With Machine Learning (ML) techniques, a statistical approach can be designed in order to perform weather forecasting activity. Moreover, the recently growing interest in Edge Computing Tiny Intelligent architectures is proposing a shift towards the deployment of ML algorithms on Tiny Embedded Systems (ES). This paper describes how Deep but Tiny Neural Networks (DTNN) can be designed to be parsimonious and can be automatically converted into a STM32 microcontroller-optimized C-library through X-CUBE-AI toolchain; we propose the integration of the obtained library with Miosix, a Real Time Operating System (RTOS) tailored for resource constrained and tiny processors, which is an enabling factor for system scalability and multi tasking. With our experiments we demonstrate that it is possible to deploy a DTNN, with a FLASH and RAM occupation of 45,5 KByte and 480 Byte respectively, for atmospheric pressure forecasting in an affordable cost effective system. We deployed the system in a real context, obtaining the same prediction quality as the same DNN model deployed on the cloud but with the advantage of processing all the necessary data to perform the prediction close to environmental sensors, avoiding raw data traffic to the cloud.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125227064","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00078
J. Fink
The collection and use of digital data by “smart city” programs raise complex operational and ethical questions that can best be addressed through carefully-monitored pilot studies before urban innovations are more widely adopted. We have created a network of single-owner campuses (academic, government, corporate and nonprofit) in the Cascadia megaregion that connects Portland (OR), Seattle (WA) and Vancouver (BC), where smart city products and services can be evaluated before deployment in neighborhoods and business districts. On the five initial campuses, we are co-locating assemblages of up to a dozen technologies through which issues of data interoperability, management, privacy and monopolization can be explored. The initial research and policy goals of this network are to educate the public about smart cities, improve accessibility for populations with disabilities, prepare city residents for natural disasters, and monitor urban tree canopies so they can better mitigate the urban heat island effect. If replicated in other regions, this testing approach can accelerate cities' responsible integration of data science solutions that can address both local and global problems.
{"title":"Digital City Testbed Center: Using campuses as smart city testbeds in the binational Cascadia region","authors":"J. Fink","doi":"10.1109/SMARTCOMP50058.2020.00078","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00078","url":null,"abstract":"The collection and use of digital data by “smart city” programs raise complex operational and ethical questions that can best be addressed through carefully-monitored pilot studies before urban innovations are more widely adopted. We have created a network of single-owner campuses (academic, government, corporate and nonprofit) in the Cascadia megaregion that connects Portland (OR), Seattle (WA) and Vancouver (BC), where smart city products and services can be evaluated before deployment in neighborhoods and business districts. On the five initial campuses, we are co-locating assemblages of up to a dozen technologies through which issues of data interoperability, management, privacy and monopolization can be explored. The initial research and policy goals of this network are to educate the public about smart cities, improve accessibility for populations with disabilities, prepare city residents for natural disasters, and monitor urban tree canopies so they can better mitigate the urban heat island effect. If replicated in other regions, this testing approach can accelerate cities' responsible integration of data science solutions that can address both local and global problems.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128159105","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00022
C. Puliafito, A. Virdis, E. Mingozzi
The integration between fog computing and the Internet of Things (IoT) creates plenty of new opportunities. Fog computing nodes run complex tasks on behalf of IoT devices, and the topological proximity of fog computing to the IoT enables several advantages (e.g., low latency). However, some IoT devices are mobile, and mobility may compromise the fog advantages. When a device moves, the communication path to the corresponding fog service may increase, with an impact on the fog advantages (which are a consequence of fog proximity) and overall performance. To overcome this issue, the fog service may be migrated across the fog computing infrastructure and maintained close enough to the served IoT device(s). It is worth noting, though, that service migration comes at a cost and may affect application Quality of Service (QoS). In this paper, we consider a fog service to be implemented as multiple containers, having one of them encapsulating an MQTT broker. Our contribution is the evaluation of the impact of container migration, which is considered in various flavours, on application QoS as perceived by mobile things. To this purpose, we consider an augmented reality application based on the MQTT protocol and conduct a set of experiments over a real fog computing testbed. Results show how migrating the fog service gives some benefits on the experienced QoS with respect to a case where no migration is performed.
{"title":"The Impact of Container Migration on Fog Services as Perceived by Mobile Things","authors":"C. Puliafito, A. Virdis, E. Mingozzi","doi":"10.1109/SMARTCOMP50058.2020.00022","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00022","url":null,"abstract":"The integration between fog computing and the Internet of Things (IoT) creates plenty of new opportunities. Fog computing nodes run complex tasks on behalf of IoT devices, and the topological proximity of fog computing to the IoT enables several advantages (e.g., low latency). However, some IoT devices are mobile, and mobility may compromise the fog advantages. When a device moves, the communication path to the corresponding fog service may increase, with an impact on the fog advantages (which are a consequence of fog proximity) and overall performance. To overcome this issue, the fog service may be migrated across the fog computing infrastructure and maintained close enough to the served IoT device(s). It is worth noting, though, that service migration comes at a cost and may affect application Quality of Service (QoS). In this paper, we consider a fog service to be implemented as multiple containers, having one of them encapsulating an MQTT broker. Our contribution is the evaluation of the impact of container migration, which is considered in various flavours, on application QoS as perceived by mobile things. To this purpose, we consider an augmented reality application based on the MQTT protocol and conduct a set of experiments over a real fog computing testbed. Results show how migrating the fog service gives some benefits on the experienced QoS with respect to a case where no migration is performed.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131789807","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00053
M. Cococcioni, Federico Rossi, E. Ruffaldi, S. Saponara
Nowadays, real-time applications are exploiting DNNs more and more for computer vision and image recognition tasks. Such kind of applications are posing strict constraints in terms of both fast and efficient information representation and processing. New formats for representing real numbers have been proposed and among them the Posit format appears to be very promising, providing means to implement fast approximated version of widely used activation functions in DNNs. Moreover, information processing performance are continuously improved thanks to advanced vectorized SIMD (single-instruction multiple-data) processor architectures and instructions like ARM SVE (Scalable Vector Extension). This paper explores both approaches (Posit-based implementation of activation functions and vectorized SIMD processor architectures) to obtain faster DNNs. The two proposed techniques are able to speed up both DNN training and inference steps.
{"title":"A Novel Posit-based Fast Approximation of ELU Activation Function for Deep Neural Networks","authors":"M. Cococcioni, Federico Rossi, E. Ruffaldi, S. Saponara","doi":"10.1109/SMARTCOMP50058.2020.00053","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00053","url":null,"abstract":"Nowadays, real-time applications are exploiting DNNs more and more for computer vision and image recognition tasks. Such kind of applications are posing strict constraints in terms of both fast and efficient information representation and processing. New formats for representing real numbers have been proposed and among them the Posit format appears to be very promising, providing means to implement fast approximated version of widely used activation functions in DNNs. Moreover, information processing performance are continuously improved thanks to advanced vectorized SIMD (single-instruction multiple-data) processor architectures and instructions like ARM SVE (Scalable Vector Extension). This paper explores both approaches (Posit-based implementation of activation functions and vectorized SIMD processor architectures) to obtain faster DNNs. The two proposed techniques are able to speed up both DNN training and inference steps.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132974129","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00087
P. Bellini, D. Cenni, P. Nesi, M. Soderi
Smart Cities are probably on the more complex environment for IOT data collection. IOT data could have different producers, sample rates, periodic and aperiodic, typical trends, structures and stacks, faults, etc. Thus, a strongly flexible and scalable solution is needed to avoid investing huge amount of resources in anomaly detection that has to be done in real time and has to be agnostic to the above-mentioned problems. This paper presents a solution for automatic detection of anomalies. The proposed approach scales seamlessly and integrates in different contexts, featuring different sensor types, protocols, and data formats, and computationally cheap. The research has been developed in the context of Snap4City PCP Select4Cities project and is presently implemented in the Https://www.snap4city.org solution adopted in several cities and regions.
{"title":"Anomaly Detection on IOT Data for Smart City","authors":"P. Bellini, D. Cenni, P. Nesi, M. Soderi","doi":"10.1109/SMARTCOMP50058.2020.00087","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00087","url":null,"abstract":"Smart Cities are probably on the more complex environment for IOT data collection. IOT data could have different producers, sample rates, periodic and aperiodic, typical trends, structures and stacks, faults, etc. Thus, a strongly flexible and scalable solution is needed to avoid investing huge amount of resources in anomaly detection that has to be done in real time and has to be agnostic to the above-mentioned problems. This paper presents a solution for automatic detection of anomalies. The proposed approach scales seamlessly and integrates in different contexts, featuring different sensor types, protocols, and data formats, and computationally cheap. The research has been developed in the context of Snap4City PCP Select4Cities project and is presently implemented in the Https://www.snap4city.org solution adopted in several cities and regions.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"128 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128104998","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 : 2020-09-01DOI: 10.1109/SMARTCOMP50058.2020.00049
Olli Väänänen, T. Hämäläinen
The escalation of the Internet of Things applications has put on display the different sensor data processing methods. The sensor data compression is one of the fundamental methods to reduce the amount of data needed to transmit from the sensor node which is often battery powered and operates wirelessly. Reducing the amount of data in wireless transmission is an effective way to reduce overall energy consumption in wireless sensor nodes. The methods presented and tested are suitable for constrained sensor nodes with limited computational power and limited energy resources. The methods presented are compared with each other using compression ratio and inherent latency. Latency is an important parameter in on-line applications. The improved variation of the linear regression-based method called RT-LRbTC is tested and it has proved to be a potential method to be used in a wireless sensor node with a fixed and predictable latency. The compression efficiency of the compression algorithms is tested with real measurement data sets.
{"title":"Sensor Data Stream on-line Compression with Linearity-based Methods","authors":"Olli Väänänen, T. Hämäläinen","doi":"10.1109/SMARTCOMP50058.2020.00049","DOIUrl":"https://doi.org/10.1109/SMARTCOMP50058.2020.00049","url":null,"abstract":"The escalation of the Internet of Things applications has put on display the different sensor data processing methods. The sensor data compression is one of the fundamental methods to reduce the amount of data needed to transmit from the sensor node which is often battery powered and operates wirelessly. Reducing the amount of data in wireless transmission is an effective way to reduce overall energy consumption in wireless sensor nodes. The methods presented and tested are suitable for constrained sensor nodes with limited computational power and limited energy resources. The methods presented are compared with each other using compression ratio and inherent latency. Latency is an important parameter in on-line applications. The improved variation of the linear regression-based method called RT-LRbTC is tested and it has proved to be a potential method to be used in a wireless sensor node with a fixed and predictable latency. The compression efficiency of the compression algorithms is tested with real measurement data sets.","PeriodicalId":346827,"journal":{"name":"2020 IEEE International Conference on Smart Computing (SMARTCOMP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129031678","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：