東華大學圖書館 |

Synchronization Schemes for Internet of Things and Edge Intelligence Applications.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Synchronization Schemes for Internet of Things and Edge Intelligence Applications./
作者:	Olaniyan, Richard.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
面頁冊數:	165 p.
附註:	Source: Dissertations Abstracts International, Volume: 83-10, Section: B.
Contained By:	Dissertations Abstracts International83-10B.
標題:	Scheduling. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29043398
ISBN:	9798209932055

Synchronization Schemes for Internet of Things and Edge Intelligence Applications.
Olaniyan, Richard.

Synchronization Schemes for Internet of Things and Edge Intelligence Applications. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 165 p.

Source: Dissertations Abstracts International, Volume: 83-10, Section: B.

Thesis (Ph.D.)--McGill University (Canada), 2021.

This item must not be sold to any third party vendors.

Devices controlled by cloud or edge resident coordinators are becoming an important trend for creating Internet of Things (IoT) systems and smart systems. The cloud provides a global perspective while the edge provides low latency and localized service to the devices. Coordinating these devices to work collectively to solve problems with strict timing requirements in the presence of disconnections is a challenge. An important coordination task is to have devices perform the same action at the same point in time. With the large number of devices and data being generated by millions of edge devices, there is a need for a synchronization scheme to orchestrate the actions of multiple devices such that they can line up their start times for tasks that require strict coordination. Clock synchronization is necessary but not sufficient for such a system. A synchronization scheme for an edge-basedIoT system needs to handle issues such as network disconnections, faults, failures, and mobility all of which are attributes expected from an edge-based system.With the recent intersection of edge computing and artificial intelligence (AI) applications, a new Edge AI paradigm has sufficed. Real-time AI applications mapped on edge computing need to perform data capture, data processing/intelligence extraction and device actuation within some given time bounds. Synchronization across devices is an important problem that needs to be solved at the different stages of an AI application. Synchronized data capture reduces the amount of time required in preprocessing data (data aggregation, data cleaning, missing data handling, etc.). In the data processing phase, synchronization is key in ensuring convergence, accuracy and speed of the distributed training process across multiple edge devices. The actuation phase in some cases requires some actions to be performed at the same time at different devices, thus, the need for synchronization.To solve the problem of synchronization in edge-based IoT, we propose three task-based and two redundancy-based algorithms. We present a tree-hierarchical architecture with controllers at different levels and devices at the leaf for synchronizing the operations of large collections of Internet of Things (IoT) such as drones, Internet of Vehicles, etc. Tasks are differentiated into three categories - synchronous (remote call with strict timing requirements), asynchronous (remote call with relaxed timing requirement) and local (self-call of a device on itself with relaxed timing requirement) depending on the execution mode and requirements. We evaluate the performance of the algorithms using trace-driven simulations and compare them to existing solutions. We identify the specific IoT application domains and runtime conditions in which each algorithm adapts best.We further propose a fast edge-based synchronization scheme that can time align the execution of input-output tasks as well as compute tasks. The primary idea of the fast synchronizer is to cluster devices into groups that are highly synchronized in their task execution and statically determine synchronization points using a game-theoretic solver. The cluster of devices uses a late notification protocol to select the best point among the pre-computed synchronization points to reach a time aligned task execution as quickly as possible. We evaluate the performance of our synchronization scheme using trace-driven simulations as well as an implementation in Ray - a Python framework for programming distributed applications, and we compare the performance with existing distributed synchronization schemes for real-time AI application tasks. We show that our fast synchronizer delivers significant performance improvements over existing solutions.

ISBN: 9798209932055Subjects--Topical Terms:

750729
Scheduling.

Synchronization Schemes for Internet of Things and Edge Intelligence Applications.
LDR:08249nmm a2200349 4500 001 2350387
005 20221020125750.5
008 241004s2021 ||||||||||||||||| ||eng d
020 $a 9798209932055
035 $a (MiAaPQ)AAI29043398
035 $a (MiAaPQ)McGill_3x816s56k
035 $a AAI29043398
040 $a MiAaPQ $c MiAaPQ
100 1 $a Olaniyan, Richard. $3 3689865
245 1 0 $a Synchronization Schemes for Internet of Things and Edge Intelligence Applications.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 165 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-10, Section: B.
500 $a Advisor: Maheswaran, Muthucumaru.
502 $a Thesis (Ph.D.)--McGill University (Canada), 2021.
506 $a This item must not be sold to any third party vendors.
520 $a Devices controlled by cloud or edge resident coordinators are becoming an important trend for creating Internet of Things (IoT) systems and smart systems. The cloud provides a global perspective while the edge provides low latency and localized service to the devices. Coordinating these devices to work collectively to solve problems with strict timing requirements in the presence of disconnections is a challenge. An important coordination task is to have devices perform the same action at the same point in time. With the large number of devices and data being generated by millions of edge devices, there is a need for a synchronization scheme to orchestrate the actions of multiple devices such that they can line up their start times for tasks that require strict coordination. Clock synchronization is necessary but not sufficient for such a system. A synchronization scheme for an edge-basedIoT system needs to handle issues such as network disconnections, faults, failures, and mobility all of which are attributes expected from an edge-based system.With the recent intersection of edge computing and artificial intelligence (AI) applications, a new Edge AI paradigm has sufficed. Real-time AI applications mapped on edge computing need to perform data capture, data processing/intelligence extraction and device actuation within some given time bounds. Synchronization across devices is an important problem that needs to be solved at the different stages of an AI application. Synchronized data capture reduces the amount of time required in preprocessing data (data aggregation, data cleaning, missing data handling, etc.). In the data processing phase, synchronization is key in ensuring convergence, accuracy and speed of the distributed training process across multiple edge devices. The actuation phase in some cases requires some actions to be performed at the same time at different devices, thus, the need for synchronization.To solve the problem of synchronization in edge-based IoT, we propose three task-based and two redundancy-based algorithms. We present a tree-hierarchical architecture with controllers at different levels and devices at the leaf for synchronizing the operations of large collections of Internet of Things (IoT) such as drones, Internet of Vehicles, etc. Tasks are differentiated into three categories - synchronous (remote call with strict timing requirements), asynchronous (remote call with relaxed timing requirement) and local (self-call of a device on itself with relaxed timing requirement) depending on the execution mode and requirements. We evaluate the performance of the algorithms using trace-driven simulations and compare them to existing solutions. We identify the specific IoT application domains and runtime conditions in which each algorithm adapts best.We further propose a fast edge-based synchronization scheme that can time align the execution of input-output tasks as well as compute tasks. The primary idea of the fast synchronizer is to cluster devices into groups that are highly synchronized in their task execution and statically determine synchronization points using a game-theoretic solver. The cluster of devices uses a late notification protocol to select the best point among the pre-computed synchronization points to reach a time aligned task execution as quickly as possible. We evaluate the performance of our synchronization scheme using trace-driven simulations as well as an implementation in Ray - a Python framework for programming distributed applications, and we compare the performance with existing distributed synchronization schemes for real-time AI application tasks. We show that our fast synchronizer delivers significant performance improvements over existing solutions.
520 $a Les appareils controles par les coordinateurs residents du cloud ou de peripherie deviennent une tendance importante pour la creation de systemes Internet des objets (IoT) et de systemes intelligents. Le cloud offre une perspective globale tandis que la peripherie fournit une faible latence et un service localise aux appareils. Coordonner ces dispositifs pour qu'ils fonctionnent collectivement pour resoudre des problemes avec des exigences de synchronisation strictes en presence de deconnexions est un defi. Une tache de coordination importante consiste a demander aux appareils d'effectuer la meme action au meme moment. Avec le grand nombre dappareils et de donnees generes par des millions dappareils de peripherie, il est necessaire de disposer dun schema de synchronisation pour orchestrer les actions de plusieurs appareils afin qu'ils puissent aligner leurs heures de debut pour les taches qui necessitent une coordination stricte. La synchronisation dhorloge est necessaire, mais pas suffisante pour un tel systeme. Un schema de synchronisation pour un systeme IoT base sur la peripherie doit gerer des problemes tels que les deconnexions reseau, les pannes, les pannes et la mobilite, qui sont tous des attributs attendus dun systeme base sur la peripherie.Avec l'intersection recente des applications de Edge computing et d'intelligence artificielle (IA), un nouveau paradigme Edge Al a suffi. Les applications dIA en temps reel mappees a linformatique de pointe doivent effectuer la capture de donnees, le traitement des donnees / extraction de lintelligence et lactivation de lappareil dans des limites de temps donnees. La synchronisation entre les appareils est un probleme important qui doit etre resolu aux differentes etapes dune application dIA. La capture de donnees synchronisee reduit le temps necessaire au pretraitement des donnees (agregation de donnees, nettoyage des donnees, traitement des donnees manquantes, etc.). Dans la phase de traitement des donnees, la synchronisation est essentielle pour garantir la convergence, la precision et la vitesse du processus de formation distribue sur plusieurs peripheriques de peripherie. La phase dactionnement dans certains cas necessite que certaines actions soient effectuees en meme temps sur differents appareils, dou la necessite dune synchronisation.Pour resoudre le probleme de la synchronisation dans l'loT base sur la peripherie, nous proposons trois algorithmes bases sur des taches et deux algorithmes bases sur la redondance. Nous presentons une architecture arborescente avec des controleurs a differents niveaux et des dispositifs a la feuille pour synchroniser les operations de grandes collections d'Internet des objets (IoT) tels que les drones, Internet des vehicules, etc. Les taches sont differenciees en trois categories - synchrones ( appel a distance avec exigence de synchronisation stricte), asynchrone (appel a distance avec exigence de synchronisation assouplie) et local (appel automatique dun appareil sur lui-meme avec exigence de synchronisation assouplie) en fonction du mode dexecution et des exigences. Nous evaluons les performances des algorithmes a laide de simulations basees sur les traces et les comparons aux solutions existantes. Nous identifions les domaines dapplication IoT specifiques et les conditions d'execution dans lesquels chaque algorithme sadapte le mieux.
590 $a School code: 0781.
650 4 $a Scheduling. $3 750729
650 4 $a Internet of Things. $3 3538511
650 4 $a Taxonomy. $3 3556303
650 4 $a Drones. $3 3549538
650 4 $a Algorithms. $3 536374
650 4 $a Communication. $3 524709
650 4 $a Artificial intelligence. $3 516317
650 4 $a Computer science. $3 523869
650 4 $a Web studies. $3 2122754
690 $a 0459
690 $a 0800
690 $a 0984
690 $a 0646
710 2 $a McGill University (Canada). $3 1018122
773 0 $t Dissertations Abstracts International $g 83-10B.
790 $a 0781
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29043398