Status of this Memo
IoT Operations Working Group F. Foukalas
Internet-Draft A. Tziouvaras
Intended status: Draft Standard March 30, 2021
Expires: September, 2019
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on September 22, 2021.
Copyright (c) 2021 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(https://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with
respect to this document.
Next generation Internet requires decentralized and distributed
intelligence in order to make available a new type of
experience to serve the user's interests. Such new services
will be enabled by deploying the intelligence
over a high volume of IoT devices in a form of distributed
protocol. Such a protocol will orchestrate the machine learning
(ML) application in order to train the aggregated data available
from the IoT devices. The training is not an easy task in such
a distributed environment, where the amount of connected IoT
devices will scale up and the needs for both interoperability
and computing are high. This draft, addresses both issues
by combining two emerging technologies known as edge AI
and fog computing. The protocol procedures aggregate the data
collected by the IoT devices into a fog node and apply edge AI
for data analysis at the edge of the infrastructure. The
analysis of the IoT requirements resulted in an end-to-end ML
protocol specification which is presented throughout this draft.
Table of Contents
1. Introduction 2
2. Background and terminology 3
3. Edge computing architecture 4
4. Protocol stages 8
4.1. Initial configuration 8
4.2. FL training 11
4.3. Cloud update 12
5. Security Considerations 14
6. IANA Considerations 15
7. Conclusions 15
8. References 15
8.1. Normative References 15
9. Acknowledgments 16
There is an evident requirement to address several challenges
to offer robust IoT services by leveraging the integration of
Edge computing with IoT known as IoT edge computing. The concept
of IoT edge computing has not been specified in detail yet
although two recent drafts described already some aspects of such
Internet architecture. Such architecture is way more useful in case
of distributed machine learning deployment to future Internet,
where the edge artificial intelligence will play an important role.
Towards this end, the proposed draft provides first the IoT edge
computing architecture, which includes the necessary elements
to deploy distributed machine learning. Second, three stages of
such a distributed intelligence are described in a sort of protocol
procedures, where the initialization, the learning and cloud updates
were devised. Details are given for all the protocol procedures
of the distributed machine learning for IoT edge computing.
2. Background and terminology
Below we list a number of terms related with the distributed
machine learning solution:
End devices: End devices  are IoT devices that collect
data while also having computing and networking capabilities.
End devices can be any type of device that can connect to the
Edge gateway and facilitate sensors for data collection.
Edge gateway: The Edge gateway is a server that is located to
the Edge of the network . It facilitates large computational
and networking capabilities and coordinates the FL process.
The Edge gateway is used to relieve the traffic from the network
backhaul as the end devices connect to the Edge instead of the
Cloud: Cloud supports very large computational capabilities 
and is geographically located far from the end devices. It provides
accessibility to the Edge gateway and remains agnostic on the amount
and type of participating end devices. As a result, the cloud does
not have an active role in the FL training process.
Show full document text