EdgePS: Selective Parameter Aggregation for Distributed Machine Learning in Edge Computing
1
Citation
32
Reference
10
Related Paper
Citation Trend
Abstract:
In this paper, we propose EdgePS, an advanced parameter server approach for distributed machine learning in edge computing scenarios. Different from the Conventional Parameter Server (CPS) approach, which performs parameter aggregation after every local training epoch, EdgePS synchronizes the parameters of all workers only when the local training cannot improve the global model performance. We first analyze how the local training will impact the performance of the global model, and then design algorithms to determine when the best time is to perform the parameter aggregation. Both real testbed experiments and extensive large scale simulations demonstrate that EdgePS can train a practical machine learning model, e.g., VGG-16, with up to 59.28% less time compared with the CPS approach. With the same training time, EdgePS can improve model accuracy by up to 30.19 % compared with the state-of-the-art distributed machine learning algorithm designed for edge computing scenarios.Keywords:
Testbed
This paper describes a testbed facility - the HoTDeC (HOvercraft Testbed for DEcentralized Control) - developed by the authors at the University of Illinois, consisting of multiple autonomous hovercraft vehicles which are wirelessly networked. This facility provides a flexible and state-of-the-art testbed for experimentation with inter-networked vehicles and sensors for decentralized and cooperative control, in a dynamically nontrivial setting.
Testbed
Command and Control
Cite
Citations (0)
Edge device
Transfer of learning
Cite
Citations (0)
Edge device
Cite
Citations (2)
This paper is a method for efficient data processing trend prediction in edge computing, where many studies have recently been conducted. For distributed processing in edge computing, offloading method from each edge must be processed within the limited computing power of the edge. Thus, in the user devices, it needs to efficiently select the edge in consideration of edge performance and data processing trend with MDP. This paper provides an efficient offloading scheme by selecting edges and distributing the traffic in the edge. As a result, our method is to offload effectively to the edge because it considers both data processing trends based on MDP and the performance of the edge compared with the existing offloading methods.
Edge device
Data Processing
Cite
Citations (0)
This tutorial will introduce the participants to the DeterLab testbed and demonstrate how to use it in research and in education. DeterLab is publicly available and free network testbed hosted by U...
Testbed
Cite
Citations (0)
本解説では，ブレインモルフィックコンピューティングハードウェア（Brainmorphic Computing HW）およびHuman-centric Edge AIの2つの話題と，それらの相互関係について述べる．ブレインモルフィックコンピューティングは，脳の情報処理原理に構成論的に迫ることを目指した新しいコンピューティングパラダイムである．これを実現するためには，現状のノイマン型デジタルコンピューティングHWとは異なり，脳科学的なアプローチにより，脳に特異的な機能群に対応する脳型アーキテクチャ群を，脳型HWに適した新規機能デバイスの物理・ダイナミクスを活用して効率的に実装する，ボトムアップ的アプローチを取る必要がある．一方，近未来のHuman-centric Edge AIパラダイムとは，現在のEdge AIあるいはEdgeコンピューティングの先にあるAIパラダイムであり，それぞれの個人の，個人的な問題に対し，個人的なデバイスにより，必要な時（だけ）に，できる限りローカルに対処するEdge AIパラダイムである．このためには，非常に小型で，非常に低消費電力なEdge HWが必要となる．ただし，HWに汎用性は求められず，必要な機能をいかに効率的に実装するかがカギとなる．最後に，これら2つの間の密接な相互関係について紹介する．
Edge device
Cite
Citations (1)
Testbed management tool is developed for the purpose of supporting user's experiments. On the other hand, there are researches that developed a tool to enable experiments that testbed does not targeted. Although there are requests from users for using these tools, many are not provided to users in Testbed. Therefore, it is necessary to provide a framework for users to easily use or implement such tools in Testbed. We aimed to make it possible to use new tools easily without modifying of conventional Testbed management tool by implementing subsystem on testbed. In this paper, we describe the result of a case study for building the environment on StarBED resources using TopDL description which is the topology description language proposed by DeterLab.
Testbed
Cite
Citations (0)
This paper describes a testbed facility - the HoTDeC (hovercraft testbed for decentralized control) - developed by the authors at the University of Illinois, consisting of multiple autonomous hovercraft vehicles which are wirelessly networked. This facility provides a flexible and state-of-the-art testbed for experimentation with inter-networked vehicles and sensors for decentralized and cooperative control, in a dynamically nontrivial setting.
Testbed
Decentralised system
Command and Control
Cite
Citations (29)
Edge computing (EC) has recently emerged as a novel computing paradigm that offers users low-latency services. Suffering from constrained computing resources due to their limited physical sizes, edge servers cannot always handle all the incoming computation tasks timely when they operate independently. They often need to cooperate through peer-offloading. Deployed and managed by different stakeholders, edge servers operate in a distrusted environment. Trust and incentive are the two main issues that challenge cooperative computing between them. Another unique challenge in the EC environment is to facilitate trust and incentive in a decentralized manner. To tackle these challenges systematically, this paper proposes CoopEdge, a novel blockchain-based decentralized platform, to drive and support cooperative edge computing. On CoopEdge, an edge server can publish a computation task for other edge servers to contend for. A winner is selected from candidate edge servers based on their reputations. After that, a consensus is reached among edge servers to record the performance in task execution on blockchain. We implement CoopEdge based on Hyperledger Sawtooth and evaluate it experimentally against a baseline and two state-of-the-art implementations in a simulated EC environment. The results validate the usefulness of CoopEdge and demonstrate its performance.
Implementation
Cite
Citations (110)
The evolution of the Internet of Things (IoT) has augmented the necessity for Cloud, edge and fog platforms. The chief benefit of cloud-based schemes is they allow data to be collected from numerous services and sites, which is reachable from any place of the world. The organizations will be benefited by merging the cloud platform with the on-site fog networks and edge devices and as result, this will increase the utilization of the IoT devices and end users too. The network traffic will reduce as data will be distributed and this will also improve the operational efficiency. The impact of monitoring in edge and fog computing can play an important role to efficiently utilize the resources available at these layers. This paper discusses various techniques involved for monitoring for edge and fog computing and its advantages. The paper ends with a case study to demonstarte the need of monitoring in fog and edge in the healthcare system.
Fog Computing
Edge device
Cite
Citations (61)