SCHEDULE AT-A-GLANCE
DAY1. Monday, April 29Day 3 of ONE Summit | |
|---|---|
2 hour on-site discussion |
Day2. Monday, April 29 (APAC time zone friendly)
21:00 – 24:00 EDT (UTC-4)
03:00 – 06:00 CEST (UTC+2) (Friday)
09:00 – 12:00 CST (UTC+8) (Friday)
Day3. Friday, Oct 20 (EMEA time zone friendly)
10:00 – 13:00 EDT (UTC-4)
16:00 – 19:00 CEST (UTC+2)
22:00 – 01:00 CST (UTC+8)
Day2. Thursday, Oct 19 (APAC time zone friendly)
Introduction to Akraino activities in 2022-2023Collaboration with other communities | SAN JOSE MCENERY CONVENTION CENTER |
Day 3 of ONE Summit
Wednesday, May 1
Zoom Link: https://zoom.us/meeting/register/tJAqfumpqDgpHtLYIi5TDpVj-k6Ymx0Uq0aYRecording: https://zoom.us/rec/share/adshMjib7SycEMDCWR019CnCjE0lY9lXqkofvSd3-jp-hCpd0cP7NQLaDbP1lwH6.sePsxcxndfFVjb6E?startTime=1697763466000j/98538301700?pwd=RXlFdHpZRDlHTzFaVFRnakw2b0F5QT09
Recording: TBD
| Time(UTC-7) | Topics | ||||||
|---|---|---|---|---|---|---|---|
| 1815:00-1815:0510 | Welcome note | ||||||
| 1815:0510-1815:35 | Announcement of akraino award 2022-2023 winners Winners of the 3rd Akraino Annual Awards 2022 - 2023 - Akraino - Akraino Confluence | ||||||
| 18:35-18:55 | Name: Colin Peters , Fujitsu Presentation title: Edge Service Enabling Platform Fall 2023 Update Abstract: The Edge Service Enabling Platform is a blueprint in development under Akraino with the goal of creating a platform for a community of device and resource providers and edge service designers and maintainers, where those individual experts can cooperate to make the design and implementation of edge services easier. This presentation will cover the goals of work on the blueprint in 2023 and what progress we have made.
Bio: Colin Peters is a software architect and business planner for edge computing at Fujitsu. He is currently leading the Edge Service Enabling Platform development project at Fujitsu, to help expand and develop the edge computing market. He has also been the PTL for the Smart Data Transactions for CPS blueprint, and continues to work with the open source community wherever he can for the spread and evolution of edge computing. | 18:55-19:15 | |||||
| 19:15-19:30 | Introduction to TSC members 2023-2024 TSC chair & co-chair nomination start(Close Oct.27) | ||||||
| 19:30-19:35 | Closing |
Waiting for reply about making presentation
・CFN BP
・LF Edge sandbox
・OSSi
DAY1 10/18 at OCP global summit(On site only)
1 hour on-site discussion
Agenda(Idea)
Collaboration with other communities
Outreach
Next Akraino release and 2024 activities
DAY2 10/19 18:00~21:00(UTC-7)
09:00~12:00(UTC+8)(10/20)
03:00~06:00(UTC+2)
DAY3 10/20 07:00~10:00(UTC-7)
22:00~01:00(UTC+8)
16:00~19:00(UTC+2)
Presentation session(Virtual)
Day3. Friday, Oct 20 (EMEA time zone friendly)
Collaboration with other open communities
| 30 | Jeff Brower, Device AI applications running at the AI Edge on very small form-factor devices (for example pico ITX), and without an online cloud connection, need to perform automatic speech recognition (ASR) under difficult conditions, including background noise, urgent or stressed voice input, and other talkers in the background. For robotics applications, background noise may also include servo motor and other mechanical noise. Under these conditions, efficient open source ASRs such as Kaldi and Whisper tend to produce "sound-alike" errors, for example: in the early days a king rolled the stake To address this issue independently of ASR model, Signalogic is developing a Small Language Model (SLM) to correct sound-alike errors, capable of running in a very small form-factor and under 10W, for example using two (2) Atom CPU cores. The SLM must run every 1/2 second and with a backwards/forwards context of 3-4 words. Unlike an LLM, a wide context window, domain knowledge, and extensive web page training are not needed. |
|
|
|
| 15: |
| 30- |
| 15: |
| 50 |
Hidetsugu Sugiyama, |
Title:Leveraging the IOWN Global Forum PoC reference cases
A technical look at AI-integrated Communications and Cyber-Physical Systems enabled by IOWN APN and Data centric infrastructure across edge, core and cloud environments
Abstraction:
Presentation title: LF edge and IOWN joint PoC plan
Abstraction:
The amount of data and computation required for computing workloads is increasing at an accelerating rate, and reducing power consumption and evolving data bandwidth are becoming challenges. At the same time, fast responding time is also required. Solving these challenges and realizing future workloads in the 2030s will require a paradigm shift in communications and computing.
Against this background, IOWN and LF Edge signed an MOU.
IOWN(Innovative Optical and Wireless Network) Global forum are focusing on next generation communications and computing infrastructure and technology to realize future workload. LF Edge are developing software and architectures to enable our workload primarily from the edge computing side.
I believe that the combination of IOWN technology and LF Edge OSS will enable future workload and improve the performance of use cases. So I'm planning a joint PoC(Proof of Concept) using two community technologies. This presentation will introduce the points to be demonstrated in the PoC and future plans.
| View file | ||||
|---|---|---|---|---|
|
Name: Alexandre Ferreira, ARM
Presentation Title: LFEdge Catalog proposal and demo
Abstraction:
LFEdge Catalog shows how LFEdge applications can be packaged in a easily deployable form. A catalog is also available allowing users to manage life-cycle of applications: deploy, update, and remove. The catalog utilizes open-source technologies like helm charts and OpenTofu scripts. SMARTER edge demo is used to demonstrate how an application can be cataloged and deployed.
Bio:
Alexandre P. Ferreira joined Arm in 2017 as a Principal Software Architect. He received his B.S. in Electrical Engineering from Universidade de Brasilia, M.S. in Computer Science from Universidade Federal de Santa Catarina and PhD in Computer Science from University of Pittsburgh in 2011. He has 10 publications and 7 patents in Computer Science. He is also a senior IEEE member since 2009. At Arm he has working on IoT related subjects like resource management.
Name: Rick Cao, Meta
Presentation Title: LF Edge AI Edge project
Abstraction:
Bio:
Rick Cao is an AI researcher in Meta focusing on Generative AI and large language models. He’s currently focusing on retrieval argumented language models. He previously worked in Ads ranking area focusing on ads targeting and delivery. Prior to Meta, he worked in the Cloud AI area at AWS SageMaker.
Name: Oleg Berzin , Equinix
Presentation title: MEC Service Federation with DevOps MEC Infra Orchestration (PCEI Blueprint R7)
Abstract:
In our solution we use Akraino Public Cloud Edge Interface (PCEI) blueprint and MEC Location API service to demonstrate orchestration of federated MEC infrastructure and services, including, Bare metal, interconnection, virtual routing for MEC and Public Cloud IaaS/SaaS, across two operators/providers (a 5G operator and a MEC provider), 5G Control and User Plane Functions
Deployment and operation of end-to-end cloud native IoT application making use of 5G access and distributed both across geographic locations and across hybrid MEC (edge cloud) and Public Cloud (SaaS) infrastructure
By orchestrating, bare metal servers and their software stack, 5G control plane and user plane functions, interconnection between the 5G provider and MEC provider, connectivity to a public cloud as well as the IoT application and the MEC Location API service, we show how it is possible for providers to enable sharing of their services in a MEC Federation environment.
| View file | ||||
|---|---|---|---|---|
|
Oleg Berzin received his Ph.D. from Drexel University in Philadelphia, PA. Dr. Berzin worked at Verizon for 20 years, where he held technology leadership roles, including development of the 4G LTE network and capabilities for enterprise mobile and Machine-to-Machine (M2M) services. Oleg is currently a Senior Distinguished Engineer, Technology and Architecture at the Office of the CTO at Equinix, where he is responsible for development of innovation strategies and architectures and prototypes in the areas of Mobile and Fixed Edge Infrastructures, Internet of Things, Next Generation Interconnection and Networking, Virtualization, and Network Automation. Oleg is proud to hold Lifetime Emeritus status for his three CCIE certifications (R&S, WAN Switching, SP). He is honored to have served as Co-chair of the Linux Foundation Edge Akraino Technical Steering Committee and as Project Technical Lead of the Linux Foundation Akraino Public Cloud Edge Interface blueprint.
Name: Jeff Brower , Signalogic, Inc.
Update on small-form factor robotics platform for SSES blueprint. We have are working on a very small language model (VSLM) uses two (2) CPU cores and updates every 1/2 sec. The objective of this model is to correct errors in Kaldi or Whisper ASR output and increase the flexibility and "conversational range" of robot commands
| View file | ||||
|---|---|---|---|---|
|
Chief Technology Strategist - Global TME, Red Hat
| |||||||
| 15:50-16:20 | Vijay Pal, Predictive Maintenance of Hardware : In the world of smart systems, encompassing 5G, IoT, and data centers uncertainty of hardware failures is very critical. Proactive maintenance of hardware can eliminate these challenges. Our device-agnostic approach, rooted in data analysis and anomaly detection using AI and ML, positions us to fortify the entire smart ecosystem, ensuring reliability and efficiency at scale.
| ||||||
16:20-17:00 | ・Discussion about Akraino 2024 activities ・Collaboration with LF Edge AI Edge and EdgeLake | ||||||
| Closing |
Call for proposal
| No | Name | Company | Presentation title | Abstract | Preferred Time Zone | Comments | |
|---|---|---|---|---|---|---|---|
| 1 | Jeff Brower | Signalogic | jbrower at signalogic dot com | Small Language Model for Device AI Applications | Device AI applications running at the AI Edge on very small form-factor devices (for example pico ITX), and without an online cloud connection, need to perform automatic speech recognition (ASR) under difficult conditions, including background noise, urgent or stressed voice input, and other talkers in the background. For robotics applications, background noise may also include servo motor and other mechanical noise. Under these conditions, efficient open source ASRs such as Kaldi and Whisper tend to produce "sound-alike" errors, for example: in the early days a king rolled the stake To address this issue independently of ASR implementation, Signalogic is developing a Small Language Model (SLM) to correct sound-alike errors, capable of running in a very small form-factor and under 10W, for example using two (2) Atom CPU cores. The SLM must run every 1/2 second and with a backwards/forwards context of 3-4 words. Unlike an LLM, a wide context window, domain knowledge, and extensive web page training are not needed. | PDT | |
| 2 | |||||||
| 3 | |||||||
| 4 | |||||||
| 5 | |||||||
| 6 | |||||||
| 7 | |||||||
| 8 | |||||||
| 9 | |||||||
| 10 | |||||||