2021 IEEE/ACM Symposium on Edge Computing (SEC)最新文献

{"title":"Federated Learning with Infrastructure Resource Limitations in Vehicular Object Detection","authors":"Yiyue Chen, Chianing Johnny Wang, Baekgyu Kim","doi":"10.1145/3453142.3491412","DOIUrl":"https://doi.org/10.1145/3453142.3491412","url":null,"abstract":"Object detection plays an essential role in many vehicular applications such as Advanced Driver Assistance System(ADAS), Dynamic Map, and Obstacle Detection. However, object detection under the traditional centralized machine learning framework, where images transmission utilization of infrastructure resources and privacy concerns about sensitive image content leakage. We introduce Federated Learning, a practical framework that enables machine learning to be conducted in a distributed manner and potentially addresses the traditional centralized machine learning issues by avoiding raw data transmission. However, Federated Learning distributes the pieces of training to the client, which relies on client communication in Vehicular Networks heavily, and not all the clients have the same resources in the real world. Therefore, we study communication and client resource limitation issues where clients have different amounts of local images and compute resources in the Vehicular Federated Learning framework, propose an algorithm to deal with these issues, and design the experiments to prove it. The experimental results show the efficacy of the proposed algorithm, which maintains the object detection precision while improving the 66% training time and reducing 35% communication cost.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88728561","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}

{"title":"iBranchy: An Accelerated Edge Inference Platform for loT Devices◊","authors":"S. Nukavarapu, Mohammed Ayyat, T. Nadeem","doi":"10.1145/3453142.3493517","DOIUrl":"https://doi.org/10.1145/3453142.3493517","url":null,"abstract":"With the phenomenal growth of IoT devices at the network edge, many new applications have emerged, including remote health monitoring, augmented reality, and video analytics. However, se-curing these devices from different network attacks has remained a major challenge. To enable more secure services for IoT devices, threats must be discovered quickly in the network edge and effi-ciently dealt with within device resource constraints. Deep Neural Networks (DNN) have emerged as solution to provide both security and high performance. However, existing edge-based IoT DNN clas-sifiers are neither lightweight nor flexible to perform conditional computation based on device types to save edge resources. Dynamic deep neural networks have recently emerged as a technique that can accelerate inference by performing conditional computation and, therefore, save computational resources. In this work, we de-sign and develop an accelerated IoT classifier iBranchy based on a dynamic neural network that can perform quick inference with fewer edge resources while also providing flexibility to adapt to different hardware and network conditions. CCS CONCEPTS • Security and privacy → Mobile and wireless security; • Com-puting methodologies → Neural networks.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82701895","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}

{"title":"AggNet: Cost-Aware Aggregation Networks for Geo-distributed Streaming Analytics","authors":"Dhruv Kumar, Sohaib Ahmad, A. Chandra","doi":"10.1145/3453142.3491276","DOIUrl":"https://doi.org/10.1145/3453142.3491276","url":null,"abstract":"Large-scale real-time analytics services continuously collect and analyze data from end-user applications and devices distributed around the globe. Such analytics requires data to be transferred over the wide-area network (WAN) to data centers (DCs) capable of processing the data. Since WAN bandwidth is expensive and scarce, it is beneficial to reduce WAN traffic by partially aggregating the data closer to end-users. We propose aggregation networks for performing aggregation on a geo-distributed edge-cloud infrastructure consisting of edge servers, transit and destination DCs. We identify a rich set of research questions aimed at reducing the traffic costs in an aggregation network. We present an optimization formulation for solving these questions in a principled manner, and use insights from the optimization solutions to propose an efficient, near-optimal practical heuristic. We implement the heuristic in AggNet, built on top of Apache Flink. We evaluate our approach using a geo-distributed deployment on Amazon EC2 as well as a WAN-emulated local testbed. Our evaluation using real-world traces from Twitter and Akamai shows that our approach is able to achieve 47% to 83% reduction in traffic cost over existing baselines without any compromise in timeliness.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88513480","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}

{"title":"Data-Driven End-to-End Delay Violation Probability Prediction with Extreme Value Mixture Models","authors":"S. Mostafavi, G. Dán, James Gross","doi":"10.1145/3453142.3493506","DOIUrl":"https://doi.org/10.1145/3453142.3493506","url":null,"abstract":"With the advent of edge computing, there is increasing interest in wireless latency-critical services. Such applications require the end-to-end delay of the network infrastructure (communication and computation) to be less than a target delay with a certain probability, e.g., 10-2-10-5. To deal with this guarantee level, the first step is to predict the transient delay violation probability (DVP) of the packets traversing the network. The guarantee level puts a threshold on the tail of the end-to-end delay distribution; thus, it makes data-driven DVP prediction a challenging task. We propose to use the extreme value mixture model in the mixture density network (MDN) method for this task. We implemented it in a multi-hop queuing-theoretic system to predict the DVP of each packet from the network state variables. This work is a first step toward utilizing the DVP predictions, possibly in the resource allocation scheme or queuing discipline. Numerically, we show that our proposed approach outperforms state-of-the-art Gaussian mixture model-based predictors by orders of magnitude, in particular for scenarios with guarantee levels above 10−2.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88117577","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}

{"title":"Edge-Assisted Collaborative Perception in Autonomous Driving: A Reflection on Communication Design","authors":"Ruozhou Yu, Dejun Yang, Hao Zhang","doi":"10.1145/3453142.3491413","DOIUrl":"https://doi.org/10.1145/3453142.3491413","url":null,"abstract":"Collaborative perception enables autonomous driving vehicles to share sensing or perception data via broadcast-based vehicle-to-everything (V2X) communication technologies such as Cellular-V2X (C-V2X), hoping to enable accurate perception in face of inaccurate perception results by each individual vehicle. Nevertheless, the V2X communication channel remains a significant bottleneck to the performance and usefulness of collaborative perception due to limited bandwidth and ad hoc communication scheduling. In this paper, we explore challenges and design choices for V2X-based collaborative perception, and propose an architecture that lever-ages the power of edge computing such as road-side units for central communication scheduling. Using NS-3 simulations, we show the performance gap between distributed and centralized C-V2X scheduling in terms of achievable throughput and communication efficiency, and explore scenarios where edge assistance is beneficial or even necessary for collaborative perception.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83206344","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}

{"title":"Scheduling Real-Time Applications on Edge Computing Platforms with Remote Attestation for Security","authors":"Niklas Reusch, P. Pop","doi":"10.1145/3453142.3493510","DOIUrl":"https://doi.org/10.1145/3453142.3493510","url":null,"abstract":"Edge Computing Platforms (ECP) increasingly integrate applications with mixed-criticality requirements. In this paper, we consider that critical applications and Edge applications share an ECP. Critical applications are implemented as periodic hard real-time tasks and messages and have stringent timing and security requirements. Edge applications are implemented as aperiodic tasks and messages, and are not critical. We assume that the critical tasks are scheduled using static cyclic scheduling, Time-Sensitive Networking (TSN) is used for dependable communication, and Remote Attestation (RA) is employed to check that the platform components are secure. We formulate an optimization problem for the joint scheduling of critical and Edge applications, such that (i) the deadlines of the critical applications are guaranteed at design-time, (ii) the platform has resources to perform RA, and (iii) we can successfully accommodate multiple dynamic responsive Edge applications at runtime. We evaluate our approach on a realistic use case. The results show that our approach generates dependable schedules that can meet the timing constraints of the critical applications, have enough periodic slack to perform RA for security, and can accommodate Edge applications with a shorter response time.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88261367","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}

{"title":"EDDL: A Distributed Deep Learning System for Resource-limited Edge Computing Environment","authors":"Pengzhan Hao, Yifan Zhang","doi":"10.1145/3453142.3491286","DOIUrl":"https://doi.org/10.1145/3453142.3491286","url":null,"abstract":"This paper investigates the problem of performing distributed deep learning (DDL) to train machine learning (ML) models at the edge with resource-constrained embedded devices. Existing solutions mostly focus on data center environments, where powerful serverclass machines are interconnected with ultra-high-speed Ethernet, and are not suitable for edge environments where much less powerful computing devices and networks are used. Due to the resource constraint on computing devices and the network connecting them, there are three main challenges for performing edge-based DDL: (1) susceptibility to struggling workers, (2) difficulty of scaling up to a large training cluster, and (3) frequent changes in training device availability and capability. To address these challenges, we design and implement EDDL, an edge-based DDL system, with ARM-based ODROID-XU4 and Raspberry Pi 3 Model B boards. We evaluate the prototype EDDL system by performing edge-based mobile malware detection and classification on a large Android APK dataset. The evaluation results show that EDDL can efficiently train deep learning models with consumer-grade embedded devices and wireless networks while incurring small overhead.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76675008","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}

{"title":"Edge Intelligence for Beyond-5G through Federated Learning","authors":"Shashank Jere, Y. Yi","doi":"10.1145/3453142.3493519","DOIUrl":"https://doi.org/10.1145/3453142.3493519","url":null,"abstract":"The computational capabilities of mobile devices have been advancing at a rapid pace in recent times, leading to a growing interest in deploying machine learning applications on such devices. In parallel, Mobile Edge Computing (MEC) has gained traction as a potential enabler for many applications in 5G and Beyond-5G networks, paving the path for making edge devices more intelligent through distributed learning strategies. In this article, we overview the application of federated learning (FL), a novel privacy-preserving distributed learning strategy, within the context of MEC. Minimizing communications latency involved in FL tasks as well as optimizing FL tasks for resource-constrained Internet of Things (IoT) devices are investigated.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80427989","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}

{"title":"LotteryFL: Empower Edge Intelligence with Personalized and Communication-Efficient Federated Learning","authors":"Ang Li, Jingwei Sun, Binghui Wang, Lin Duan, Sicheng Li, Yiran Chen, H. Li","doi":"10.1145/3453142.3492909","DOIUrl":"https://doi.org/10.1145/3453142.3492909","url":null,"abstract":"With the proliferation of mobile computing and Internet of Things (IoT), massive mobile and IoT devices are connected to the Internet. These devices are generating a huge amount of data every second at the network edge. Many artificial intelligence applications and ser-vices have been proposed for edge devices based on the distributed data. Federated learning (FL) proves to be an extremely viable option for distributed machine learning with enhanced privacy, which can help artificial intelligence applications unleash the potential of data residing at the network edge. Its primary goal is learning a global model that offers good performance for the participants as many as possible. However, the data residing across devices is intrinsically statistically heterogeneous (i.e., non-IID data distribution) and edge devices usually have limited communication resources to transfer data. Such statistical heterogeneity (i.e., non-IID) and communication efficiency are two critical bottlenecks that hinder the development of FL. In this work, we propose LotteryFL - a personalized and communication-efficient FL framework via exploiting the Lottery Ticket hypothesis. In LotteryFL, each client learns a lottery ticket network (i.e., a subnetwork of the base model) by applying the Lottery Ticket hypothesis, and only these lottery networks will be communicated between the server and clients. Rather than learning a shared global model in classic FL, each client learns a personalized model via LotteryFL; the communication cost can be significantly reduced due to the compact size of lottery networks. To support the training and evaluation of our framework, we construct non-IID) datasets based on MNIST, CIFAR-10 and EMNIST by taking feature distribution skew, label distribution skew and quantity skew into consideration. Experiments on these non-IID datasets demonstrate that compared with the state-of-the-art approaches, LotteryFL can achieve as much as 17.24% increase in inference accuracy and 2.94x reduction on communication cost. We also demonstrate the via-bility of LotteryFL, showcasing the real-time performance of the deployed models on edge devices.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73303484","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}

{"title":"A Data-Driven Optimal Control Decision-Making System for Multiple Autonomous Vehicles","authors":"Liuwang Kang, Haiying Shen","doi":"10.1145/3453142.3493686","DOIUrl":"https://doi.org/10.1145/3453142.3493686","url":null,"abstract":"With the fast development and rising popularity of autonomous vehicle (AV) technology, multiple AVs may soon be driving on the same road simultaneously. Such multi-AV coexistence driving situations will lead to new and persistent challenges. Therefore, improvements on making control decisions for multiple AVs becomes necessary for continued driving safety. In this paper, we propose a multi-AV decision making system (MADM), which considers multi-AV coexistence driving situations during the decision-making process. In MADM, we first build a policy formation method to generate policies that learn the driving behaviors of an expert based on the expert's driving trajectory data. We then develop a multi-AV decision-making method, which adjusts the formed policies through multi-agent reinforcement learning. The adjusted policies make control decisions for multiple AVs with safety guarantee. We used a real-world traffic dataset to evaluate the decision making performance of MADM in comparison with several state-of-the-art methods. Experimental results show that MADM reduces emergency rate by as high as 51% when compared with existing methods.","PeriodicalId":6779,"journal":{"name":"2021 IEEE/ACM Symposium on Edge Computing (SEC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72829051","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}