Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking最新文献

Scheduling Continuous Operators for IoT Edge Analytics 为物联网边缘分析调度连续运营商

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-26 DOI: 10.1145/3434770.3459738

Patient Ntumba, N. Georgantas, V. Christophides

引用次数: 3

Accelerated Training via Device Similarity in Federated Learning 联邦学习中基于设备相似性的加速训练

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-26 DOI: 10.1145/3434770.3459734

Yuanli Wang, Joel Wolfrath, N. Sreekumar, Dhruv Kumar, A. Chandra

{"title":"Accelerated Training via Device Similarity in Federated Learning","authors":"Yuanli Wang, Joel Wolfrath, N. Sreekumar, Dhruv Kumar, A. Chandra","doi":"10.1145/3434770.3459734","DOIUrl":"https://doi.org/10.1145/3434770.3459734","url":null,"abstract":"Federated Learning is a privacy-preserving, machine learning technique that generates a globally shared model with in-situ model training on distributed devices. These systems are often comprised of millions of user devices and only a subset of available devices can be used for training in each epoch. Designing a device selection strategy is challenging, given that devices are highly heterogeneous in both their system resources and training data. This heterogeneity makes device selection very crucial for timely model convergence and sufficient model accuracy. Existing approaches have addressed system heterogeneity for device selection but have largely ignored the data heterogeneity. In this work, we analyze the impact of data heterogeneity on device selection, model convergence, model accuracy, and fault tolerance in a federated learning setting. Based on our analysis, we propose that clustering devices with similar data distributions followed by selecting the devices with the best processing capacity from each cluster can significantly improve the model convergence without compromising model accuracy. This clustering also guides us in designing policies for fault tolerance in the system. We propose three methods for identifying groups of devices with similar data distributions. We also identify and discuss rich trade-offs between privacy, bandwidth consumption, and computation overhead for each of these proposed methods. Our preliminary experiments show that the proposed methods can provide a 46% - 58% reduction in training time compared to existing approaches in reaching the same accuracy.","PeriodicalId":389020,"journal":{"name":"Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking","volume":"131 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123211801","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}

引用次数: 10

Towards Federated Learning with Attention Transfer to Mitigate System and Data Heterogeneity of Clients 基于注意力转移的联邦学习缓解客户端系统和数据异构性

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-26 DOI: 10.1145/3434770.3459739

Hongrui Shi, Valentin Radu

{"title":"Towards Federated Learning with Attention Transfer to Mitigate System and Data Heterogeneity of Clients","authors":"Hongrui Shi, Valentin Radu","doi":"10.1145/3434770.3459739","DOIUrl":"https://doi.org/10.1145/3434770.3459739","url":null,"abstract":"Federated learning is a method of training a global model on the private data of many devices. With a growing spectrum of devices, some slower than smartphones, such as IoT devices, and others faster, such as home data boxes, the standard Federated Learning (FL) method of distributing the same model to all clients is starting to break down-- slow clients inevitably become strugglers. We propose a FL approach that spores different size models, each matching the computational capacity of the client system. There is still a global model, but for the edge tasks, the server trains different size student models with attention transfer, each chosen for a target client. This allows clients to perform enough local updates and still meet the round cut-off time. Client models are used as the source of attention transfer after their local update, to refine the global model on the server. We evaluate our approach on non-IID data to find that attention transfer can be paired with training on metadata brought from the client side to boost the performance of the server model even on previously unseen classes. Our FL with attention transfer opens the opportunity for smaller devices to be included in the Federated Learning training rounds and to integrate even more extreme data distributions.","PeriodicalId":389020,"journal":{"name":"Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122875130","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}

引用次数: 7

Privacy-Preserving Crowd-Monitoring Using Bloom Filters and Homomorphic Encryption 利用布隆过滤器和同态加密保护隐私的人群监控

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-26 DOI: 10.1145/3434770.3459735

V. Stanciu, M. Steen, C. Dobre, Andreas Peter

{"title":"Privacy-Preserving Crowd-Monitoring Using Bloom Filters and Homomorphic Encryption","authors":"V. Stanciu, M. Steen, C. Dobre, Andreas Peter","doi":"10.1145/3434770.3459735","DOIUrl":"https://doi.org/10.1145/3434770.3459735","url":null,"abstract":"This paper introduces an architecture for crowd-monitoring which allows statistical counting for pedestrian dynamics while considering privacy-preservation for the individuals being sensed. Monitoring crowds of pedestrians has been an interesting area of study for many years. The recent prevalence of mobile devices paved the way for wide-scale deployments of infrastructures which perform automated sensing. Suddenly, people could be discreetly monitored by leveraging radio signals such as Wi-Fi probe requests periodically sent by their devices. However, this monitoring process implies dealing with sensitive data which is prone to privacy infringement by nature. While routinely performing their tasks, parties involved in this process can try to infer private information about individuals from the data they handle. Following privacy by design principles, we envision a construction which protects the short-term storage and processing of the collected privacy-sensitive sensor readings with strong cryptographic guarantees such that only the end-result (i.e. a statistical count) becomes available in the clear. We combine Bloom filters, to facilitate set membership testing for counting, with homomorphic encryption, to allow the oblivious performance of operations under encryption. We carry out an implementation of our solution using a resource-constrained device as a sensor and perform experiments which demonstrate its feasibility in practice.","PeriodicalId":389020,"journal":{"name":"Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking","volume":"38 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114041959","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}

引用次数: 9

Snowflakes at the Edge: A Study of Variability among NVIDIA Jetson AGX Xavier Boards 边缘的雪花:NVIDIA Jetson AGX Xavier板的可变性研究

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-26 DOI: 10.1145/3434770.3459729

Hazem A. Abdelhafez, Hassan Halawa, K. Pattabiraman, M. Ripeanu

引用次数: 10

Increasing Traffic Safety with Real-Time Edge Analytics and 5G 通过实时边缘分析和5G提高交通安全

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-26 DOI: 10.1145/3434770.3459732

Ivan Lujic, Vincenzo De Maio, Klaus Pollhammer, Ivan Bodrozic, J. Lasic, I. Brandić

{"title":"Increasing Traffic Safety with Real-Time Edge Analytics and 5G","authors":"Ivan Lujic, Vincenzo De Maio, Klaus Pollhammer, Ivan Bodrozic, J. Lasic, I. Brandić","doi":"10.1145/3434770.3459732","DOIUrl":"https://doi.org/10.1145/3434770.3459732","url":null,"abstract":"Despite advances in vehicle technology and road modernization, traffic accidents are a huge global issue, causing deaths and injuries, especially among pedestrians and cyclists. This often happens due to pedestrians and cyclists in drivers' blind spots or distractions delaying drivers' reactions. Therefore, timely warning drivers about critical situations is important to increase traffic safety. New edge computing and communication technologies have been proposed to reduce latency in critical IoT systems. However, state-of-the-art solutions either do not focus on traffic safety or do not consider low-latency requirements in this context. We propose InTraSafEd5G (Increasing Traffic Safety with Edge and 5G) to address these issues. InTraSafEd5G performs real-time edge analytics to detect critical situations and deliver early warnings to drivers. After describing our design choices, we provide a prototype implementation and evaluate its performance in a real-world setup. The evaluation shows that InTraSafEd5G can (i) detect critical situations in real-time and (ii) notify affected drivers in 108.73ms on average using 5G, which is within expected latency requirements of road safety IoT applications. Our solution shows a promising step towards increasing overall traffic safety and supporting decision-making in critical situations.","PeriodicalId":389020,"journal":{"name":"Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking","volume":"474 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123283170","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}

引用次数: 22

Towards a Computing Platform for the LEO Edge 面向低轨道边缘的计算平台

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-06 DOI: 10.1145/3434770.3459736

Tobias Pfandzelter, Jonathan Hasenburg, David Bermbach

引用次数: 23

Detecting and Mitigating Network Packet Overloads on Real-Time Devices in IoT Systems 检测和减轻物联网系统中实时设备上的网络数据包过载

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-06 DOI: 10.1145/3434770.3459733

Robert Danicki, M. Haug, Ilja Behnke, Laurenz Mädje, L. Thamsen

{"title":"Detecting and Mitigating Network Packet Overloads on Real-Time Devices in IoT Systems","authors":"Robert Danicki, M. Haug, Ilja Behnke, Laurenz Mädje, L. Thamsen","doi":"10.1145/3434770.3459733","DOIUrl":"https://doi.org/10.1145/3434770.3459733","url":null,"abstract":"Manufacturing, automotive, and aerospace environments use embedded systems for control and automation and need to fulfill strict real-time guarantees. To facilitate more efficient business processes and remote control, such devices are being connected to IP networks. Due to the difficulty in predicting network packets and the interrelated workloads of interrupt handlers and drivers, devices controlling time critical processes stand under the risk of missing process deadlines when under high network loads. Additionally, devices at the edge of large networks and the internet are subject to a high risk of load spikes and network packet overloads. In this paper, we investigate strategies to detect network packet overloads in real-time and present four approaches to adaptively mitigate local deadline misses. In addition to two strategies mitigating network bursts with and without hysteresis, we present and discuss two novel mitigation algorithms, called Budget and Queue Mitigation. In an experimental evaluation, all algorithms showed mitigating effects, with the Queue Mitigation strategy enabling most packet processing while preventing lateness of critical tasks.","PeriodicalId":389020,"journal":{"name":"Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking","volume":"40 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125010285","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}

引用次数: 3

Rearchitecting Kubernetes for the Edge 为Edge重新构建Kubernetes

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 2021-04-06 DOI: 10.1145/3434770.3459730

A. Jeffery, H. Howard, R. Mortier

引用次数: 31

Edgedancer Edgedancer

Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking Pub Date : 1900-01-01 DOI: 10.1145/3434770.3459731

Manuel Nieke, Lennart Almstedt, Rüdiger Kapitza

{"title":"Edgedancer","authors":"Manuel Nieke, Lennart Almstedt, Rüdiger Kapitza","doi":"10.1145/3434770.3459731","DOIUrl":"https://doi.org/10.1145/3434770.3459731","url":null,"abstract":"Edge computing brings services and data closer to their users. It targets applications where low latency is key, such as cloud-based gaming, augmented reality, and connected cars. In these scenarios mobile users are often the norm, and services need to follow their users to continuously ensure the benefits of edge computing. Mobile edge computing addresses this demand, however, we identified three open challenges: First, edge computing facilities are sparse so far and achieving good connectivity requires a seamless transfer of workloads between different edge providers. Second, migration of edge services needs to be lightweight to make relocation transparent to users. Third, using geo-distributed edge resources of various providers demands for security measures to ensure service integrity and confidentiality.; B@We present E'dgedancer, a platform that offers infrastructure support for portable, provider-independent, and secure migration of edge services. E'dgedancer offers a lightweight and generic execution environment by utilising WebAssembly, which features an efficient, easy to transfer bytecode format. To make edge services self-migratable, and independent from the provider, E'dgedancer supports the notion of mobile agents offering weak and strong migration support. It utilises trusted execution to ensure the security of edge services during execution and migration. Finally, we shows that E'dgedancer features a lower migration time compared to previously proposed virtual machine migration while offering better security and platform independence.","PeriodicalId":389020,"journal":{"name":"Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124237364","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}

引用次数: 12