Skip to content
Commits on Source (84)
Expand all Show whitespace changes
Inline Side-by-side
GR_MEC-DEC_050.md
View file @ b5fb1b5d
<div align="center">
**ETSI GR LLL-LLL DDD Vm.t.e (yyyy-mm)**
**ETSI GR MEC-DEC 050 V4.0.2 (2025-06)**
![~~CAPTION~~](media/GroupReport-logo.png)
**Group REPORT**<br />
Title;<br />
Multi-access Edge Computing (MEC); ESTIMED Use Cases & Proposed Recommendations
<br />
Part #: Part element of title;<br />
......@@ -26,31 +24,24 @@ Release #
_Should you need a step-by-step guide for drafting an ETSI deliverable, please consult the "_ [_Principles for Drafting ETSI Deliverables_ ](_Principles for Drafting ETSI Deliverables_ )_" document. Otherwise you may contact us at_ [_edithelp@etsi.org_ ](mailto:edithelp@etsi.org).
<br />
<br />
<br />
<br />
<div align="center">
**Disclaimer**
The present document has been produced and approved by the <long ISGname> (<short ISGname>) ETSI Industry Specification Group (ISG) and represents the views of those members who participated in this ISG.
It does not necessarily represent the views of the entire ETSI membership.
ETSI GR LLL-LLL DDD Vm.t.e (yyyy-mm)
The present document has been produced and approved by the Multi-access Edge Computing (MEC) ETSI Industry Specification Group (ISG) and represents the views of those members who participated in this ISG. It does not necessarily represent the views of the entire ETSI membership.
Reference<br />
&lt;Workitem><br />
DGR/MEC-DEC050EstimedReq<br />
Keywords<br />
&lt;keywords><br />
M2M, MEC, oneM2M<br />
ETSI<br />
......@@ -60,7 +51,6 @@ F-06921 Sophia Antipolis Cedex - FRANCE<br />
Tel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16<br />
Siret N° 348 623 562 00017 - APE 7112B<br />
Association à but non lucratif enregistrée à la<br />
......@@ -101,8 +91,6 @@ No representation or warranty is made that this deliverable is technically accur
In no event shall ETSI be held liable for loss of profits or any other incidental or consequential damages.
Any software contained in this deliverable is provided "AS IS" with no warranties, express or implied, including but not limited to, the warranties of merchantability, fitness for a particular purpose and non-infringement of intellectual property rights and ETSI shall not be held liable in any event for any damages whatsoever (including, without limitation, damages for loss of profits, business interruption, loss of information, or any other pecuniary loss) arising out of or related to the use of or inability to use the software.
<br />
......@@ -111,9 +99,7 @@ Any software contained in this deliverable is provided "AS IS" with no warrantie
No part may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm except as authorized by written permission of ETSI. The content of the PDF version shall not be modified without the written authorization of ETSI. The copyright and the foregoing restriction extend to reproduction in all media.
&copy; ETSI yyyy.
&copy; ETSI 2025.
All rights reserved.<br />
......@@ -121,10 +107,8 @@ All rights reserved.<br />
# Contents
<br />
# Intellectual Property Rights
Essential patents
......@@ -133,7 +117,6 @@ IPRs essential or potentially essential to normative deliverables may have been
Pursuant to the ETSI Directives including the ETSI IPR Policy, no investigation regarding the essentiality of IPRs, including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become, essential to the present document.
Trademarks
The present document may include trademarks and/or tradenames which are asserted and/or registered by their owners. ETSI claims no ownership of these except for any which are indicated as being the property of ETSI, and conveys no right to use or reproduce any trademark and/or tradename. Mention of those trademarks in the present document does not constitute an endorsement by ETSI of products, services or organizations associated with those trademarks.
......@@ -141,97 +124,832 @@ The present document may include trademarks and/or tradenames which are asserted
**DECT&#8482;**, **PLUGTESTS&#8482;**, **UMTS&#8482;** and the ETSI logo are trademarks of ETSI registered for the benefit of its Members. **3GPP&#8482;**, **LTE&#8482;** and **5G&#8482;** logo are trademarks of ETSI registered for the benefit of its Members and of the 3GPP Organizational Partners. **oneM2M&#8482;** logo is a trademark of ETSI registered for the benefit of its Members and of the oneM2M Partners. **GSM&#174;** and the GSM logo are trademarks registered and owned by the GSM Association.
# Foreword
This Group Report (GR) has been produced by ETSI Industry Specification Group &lt;long ISGname> (&lt;short ISGname>).
This Group Report (GR) has been produced by ETSI Industry Specification Group (ISG) Multi-access Edge Computing (MEC).
# Modal verbs terminology
In the present document "**should** ", "**should not** ", "**may** ", "**need not** ", "**will** ", "**will not** ", "**can** " and "**cannot** " are to be interpreted as described in clause 3.2 of the [ETSI Drafting Rules](https://portal.etsi.org/Services/editHelp/How-to-start/ETSI-Drafting-Rules) (Verbal forms for the expression of provisions).
<!-- This Group Report (GR) has been produced by ETSI Industry Specification Group &lt;long ISGname> (&lt;short ISGname>). -->
"**must** " and "**must not** " are **NOT** allowed in ETSI deliverables except when used in direct citation.
<!-- <mark>akr: Who will provide the foreword?</mark> -->
# Modal verbs terminology
# Executive summary
In the present document "**should** ", "**should not** ", "**may** ", "**need not** ", "**will** ", "**will not** ", "**can** " and "**cannot** " are to be interpreted as described in clause 3.2 of the [ETSI Drafting Rules](https://portal.etsi.org/Services/editHelp/How-to-start/ETSI-Drafting-Rules) (Verbal forms for the expression of provisions).
"**must** " and "**must not** " are **NOT** allowed in ETSI deliverables except when used in direct citation.
<!-- # Executive summary -->
<!-- <mark>akr: Who will provide the executive summary?</mark> -->
# Introduction
# 1 Scope
<br />
This document provides a set of selected use cases that leverage IoT and Edge computing technologies, with a particular focus on standard-based implementations where oneM2M and ETSI MEC frameworks can be applied. Building on the joint MEC–oneM2M White Paper [i.2], it refines and expands the identified synergies between these frameworks and offers a detailed analysis of architectural evolutions, including necessary components, required mappings, and potential enhancements to facilitate interoperability and integration between oneM2M and MEC-based deployments.
# 1 Scope
The present document ...
By analyzing real-world and industry-relevant use cases, the report highlights how the convergence of IoT and Edge computing can enable scalable, efficient, and interoperable solutions across different domains. The scope extends to evaluating the applicability of existing standards, identifying gaps, and proposing recommendations to ensure alignment with ongoing standardization efforts within ETSI and oneM2M. The document serves as a reference for stakeholders to understand the practical implications of integrating MEC and oneM2M technologies in future IoT–Edge ecosystems.
The scope includes recommendations and proposals for MEC–oneM2M interworking deployment patterns across the edge–cloud continuum, covering the four deployment options to enable latency‑critical processing and service continuity, the exposure of oneM2M IoT data and control services through the MEC platform for consumption by applications that access and actuate IoT resources, and hybrid multi‑tier arrangements that distribute functions across user equipment or gateways, and centralized infrastructure for hierarchical data management and orchestration. It also encompasses cross‑edge Artificial Intelligencet approaches that are central to the stated goals, including federated learning for privacy‑preserving collaborative model training and edge inference across MEC nodes and oneM2M‑managed data spaces, and swarm computing for distributed, real‑time coordination of agents and devices supported by edge‑hosted IoT services and MEC compute. The resulting recommendations address the functional enablers, mappings, and enhancements required to operationalize these deployment patterns and intelligent capabilities across representative use cases.
# 2 References
## 2.1 Normative references
Normative references are not applicable in the present document.
Normative references are not applicable in the present document.
## 2.2 Informative references
References are either specific (identified by date of publication and/or edition number or version number) or nonspecific. For specific references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies.
> NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long term validity.
The following referenced documents may be useful in implementing an ETSI deliverable or add to the reader's understanding but are not required for conformance to the present document.
- <a name="_ref_i.1">[i.1]</a>
- [\[i.1\]](https://www.etsi.org/deliver/etsi_gs/MEC/001_099/003/04.01.01_60/gs_mec003v040101p.pdf) ETSI GS MEC 003: "Multi-access Edge Computing (MEC); Framework and Reference Architecture".
- [\[i.2\]](https://www.etsi.org/images/files/ETSIWhitePapers/ETSI-WP59-Enabling-Multi-access-Edge-Computing-in-iot.pdf) Enabling Multi-access Edge Computing in Internet-of- Things: how to deploy ETSI MEC and oneM2M".
- [\[i.3\]](https://estimed.etsi.org/). ESTIMED Project Official Website.
# 3 Definition of terms, symbols and abbreviations
Not applicable.
## 3.1 Terms
For the purposes of the present document, the [following] terms [given in ... and the following] apply:
<!-- For the purposes of the present document, the [following] terms [given in ... and the following] apply: -->
Void.
## 3.2 Symbols
<!-- For the purposes of the present document, the [following] symbols [given in ... and the following] apply: -->
Void.
## 3.2 Symbols
For the purposes of the present document, the [following] symbols [given in ... and the following] apply:
## 3.3 Abbreviations
For the purposes of the present document, the [following] abbreviations apply:
| Acronym | Meaning |
|-----------|-------------------------------------------------------------------------|
| 3D | Three-dimensional |
| 3GPP | 3rd Generation Partnership Project |
| 5G | Fifth Generation mobile networks |
| ADN | Application Dedicated Node |
| ADN-AE | Application Dedicated Node – Application Entity |
| AE | Application Entity |
| AGV | Autonomous Guided Vehicles |
| AIS | Automatic Identification System |
| AI | Artificial Intelligence |
| API | Application Programming Interface |
| API GW | API Gateway |
| AR | Augmented Reality |
| ASN | Application Service Node |
| ASN-CSE | Application Service Node – Common Services Entity |
| AV | Autonomous Vehicle |
| CPE | Customer Premises Equipment |
| CPU | Central Processing Unit |
| CSF | Common Service Function |
| CSE | Common Services Entity |
| CVD | Coordinated Vulnerability Disclosure |
| CNC | Computer Numerical Control |
| CoAP | Constrained Application Protocol |
| DT | Digital Twin |
| FL | Federated Learning |
| GPU | Graphics Processing Unit |
| GW | Gateway |
| HTTP | Hypertext Transfer Protocol |
| IDT | Industrial Digital Twin |
| IN | Infrastructure Node |
| IN-AE | Infrastructure Node – Application Entity |
| IN-CSE | Infrastructure Node – Common Services Entity |
| IoT | Internet of Things |
| LIDAR | Light Detection and Ranging |
| ROC | Remote Operation Center |
| VRU | Vulnerable Road User |
## 3.3 Abbreviations
# 4 Overview
For the purposes of the present document, the [following] abbreviations [given in ... and the following] apply:
This clause introduces the document by clearly defining its purpose, scope, and key objectives. It lays the groundwork for the integration of the ETSI MEC and oneM2M frameworks, highlighting their critical role in enabling advanced Edge and IoT deployments across a wide range of application domains.
The ESTIMED project [i.3] takes a use case-driven approach to steer the integration of ETSI MEC and oneM2M. This strategy effectively demonstrates the tangible benefits of merging edge computing capabilities with standardized IoT architectures. By anchoring the analysis in real-world scenarios, the approach not only guides integration efforts but also informs future standardization initiatives to better meet operational demands. The project emphasizes identifying new functional requirements and potential extensions to existing specifications, thereby contributing to the continuous refinement and evolution of both MEC and oneM2M frameworks in response to emerging IoT and edge computing challenges. Following this methodology, the document is organized into three main clauses, each addressing a core aspect of the analysis.
Clause 5 delivers a domain-centric analysis of carefully selected use cases, illustrating the practical application of MEC and oneM2M in Edge-IoT environments. Use cases are categorized by relevant domains—such as Mobility, Industrial, and Maritime—to reflect their specific operational contexts. Within each domain, use cases are thoroughly examined, covering the scenario context, involved stakeholders, technical and operational requirements, and key challenges. The clause also explores how MEC and oneM2M technologies contribute to overcoming these challenges. The use cases were identified through a structured engagement with stakeholders, facilitated by a dedicated form designed to capture all critical details consistently, including descriptions, actors, requirements, preconditions and postconditions, triggers, and interaction flows.
# 4 User defined clause(s) from here onwards
Building on this foundation, Clause 6 analyzes the principal architectures of MEC and oneM2M frameworks, emphasizing their core functionalities and complementarity in supporting Edge-IoT deployments. For each use case, the clause maps relevant architectural components and capabilities from both frameworks, highlighting overlaps as well as identifying new features that could enhance support for specific scenarios. This mapping lays the groundwork for a comprehensive architectural reflection and technical assessment.
## 4.1 User defined subdivisions of clause(s) from here onwards
Clause 7 includes recommendations that enable a framework for the federation and orchestration of MEC and oneM2M instances. It also discusses how these capabilities can be used in Swarm Computing and Federated Learning applications.
<br />
Overall, this document establishes a structured and cohesive foundation to understand and advance how MEC and oneM2M can jointly enable scalable, interoperable, and forward-looking Edge-IoT solutions.
# 5 Edge & IoT Domains and Use Cases
# Annex A: <br>Title of annex
## 5.1 Introduction
This clause introduces the application domains and use cases explored in the context of the ESTIMED project, focusing on how the integration of the ETSI MEC and oneM2M frameworks enables next-generation Edge-IoT solutions. The clause builds upon a use case-driven methodology that emphasizes real-world scenarios as the foundation for architectural mapping, technical analysis, and standardization recommendations.
<br />
The goal of this clause is to present selected domains where edge computing and IoT convergence are driving tangible benefits across verticals such as mobility, industry, maritime operations, digital experiences, and smart living. For each domain, representative use cases have been identified to demonstrate the practical value of combining MEC's edge capabilities with oneM2M’s standardized IoT platform. These use cases illustrate how the integration supports low-latency data processing, scalable service delivery, and intelligent, context-aware applications.
The table below summarizes the domains and the associated use cases covered in this clause:
# Annex B: <br>Title of annex
## B.1 First clause of the annex
## B.1.1 First subdivided clause of the annex
| Domain | Use Cases | Focus |
| --------------------- | ---------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------- |
| Smart City & Mobility | - Autonomous Vehicles and Edge Continuum<br>- Vulnerable Road User Detection | Real-time data processing, V2X communication, urban mobility optimization, traffic safety |
| Industrial & Robotics | - Swarm-based Autonomous Ant Delivery Optimization<br>- Smart Warehouse Automation<br>- Industrial Digital Twins | Intelligent coordination, low-latency control, industrial automation, edge-based robotics |
| Maritime | - Assisted Manoeuvring for Connected Ships | Remote vessel monitoring, mission-critical edge processing, seamless edge-cloud communication |
| Metaverse | - Smart Virtual Shopping Service | IoT-enhanced virtual environments, edge-hosted AI analytics, immersive low-latency digital experiences |
| Future Home | - Advanced Smart Home Services | Real-time media, education, health, and automation services within personalized, responsive smart environments |
Each use case is examined in detail within its domain context, describing how MEC and oneM2M contribute to solving specific challenges, enabling innovation, and supporting interoperability. The domains and use cases form the analytical foundation for the architectural mappings and technical recommendations that follow in subsequent clauses.
## 5.2 Smart City & Mobility
### 5.2.1 Description
This clause focuses on the Smart City and Mobility domain, highlighting two key use cases: Autonomous Vehicles and Edge Continuum, and Vulnerable Road User Detection. It examines how MEC and oneM2M frameworks support advanced urban mobility solutions by enabling real-time data processing, vehicle-to-infrastructure communication, and coordinated edge intelligence. These capabilities contribute to improved traffic management, enhanced road safety, and more efficient transportation systems within smart urban environments.
### 5.2.2 Autonomous Vehicle with Continuous Edge Computing
This use case demonstrates the integration of oneM2M's IoT platform with ETSI MEC's edge computing framework to enable continuous real-time processing for autonomous vehicles (AVs). The core system (schematically reported in Figure 5.2.2-1) consists of a cloud-based oneM2M IN-CSE (centralized IoT hub) for managing vehicle and environmental data, and MN-CSE instances deployed on MEC nodes that process time-sensitive data such as vehicle sensor inputs, road conditions, and traffic signals with ultra-low latency.
Service Continuity is a critical aspect of this use case. Initially, the IN-CSE in the cloud manages and stores all data related to the AV, including location, sensor data (e.g., LIDAR, radar), and external conditions such as road infrastructure. When an AV enters a zone covered by a nearby MEC node, the IN-CSE dynamically offloads data processing tasks (such as real-time object detection, collision risk analysis, and navigation updates) to the MN-CSE instances deployed at the edge. These instances not only handle locally generated data but also share additional real-time data from other MEC applications to enhance decision-making. The MN-CSE processes the data locally, providing immediate feedback and decision-making (e.g., braking, turning, lane-changing) with minimal latency, ensuring that the vehicle operates efficiently.
As the AV moves across different zones, it will seamlessly transition from one MEC node to another. The system continuously monitors the AV's location and detects when the vehicle approaches a new MEC zone. Upon detection, the system initiates the migration of the MN-CSE's offloaded data and applications to the new MEC node. This migration ensures that all real-time data and applications are immediately available at the new location, guaranteeing that the AV experiences uninterrupted service as it moves. The oneM2M system synchronizes between the cloud and edge, ensuring that all data is consistent and up-to-date. This edge-to-edge service continuity is critical for maintaining performance and safety in dynamic environments.
By integrating oneM2M’s standardized data management with MEC's edge computing capabilities, this use case shows real-time, low-latency processing, while also enabling the migration of applications across MEC zones without compromising service quality. As the AV moves, the system adjusts dynamically, ensuring continuous service delivery through the seamless orchestration between the cloud, MEC, and edge devices. This use case not only supports static services but also mobility services, ensuring that the AV receives timely updates and services as it traverses multiple zones.
![Figure 5.2.2-1: Autonomous Vehicle Edge Computing service continuity.](media/usecase_autonomouse_vehicles.png)
**Figure 5.2.2-1: Autonomous Vehicle Edge Computing service continuity.**
### 5.2.3 Vulnerable Road Users
This scenario illustrates a collaborative architecture between oneM2M’s IoT platform and ETSI MEC’s edge computing framework to enable real-time Vulnerable Road User (VRU) detection for connected vehicles. As illustrated in Figure 5.2.3-1, at the core of the system, a cloud-based oneM2M IN-CSE serves as the central IoT hub, storing and managing VRU-related data such as pedestrian locations, cyclist trajectories, and road conditions.
Meanwhile, edge-located MN-CSE instances, deployed as MEC applications on distributed edge nodes, act as localized IoT platforms that process time-sensitive data near the vehicles. When a Host Vehicle (HV) enters a zone covered by a nearby MEC node, the cloud IN-CSE dynamically offloads VRU detection tasks—such as real-time analytics and safety alert generation—to the edge-resident MN-CSE. This offloading ensures ultra-low-latency processing, as the MN-CSE can immediately analyze VRU data (e.g., from smartphones, roadside sensors) and issue warnings to the HV without relying on distant cloud servers.
The integration of oneM2M’s standardized IoT data management with MEC’s edge computing capabilities creates a responsive and location-aware safety system. For instance, when an HV prepares to make a left turn, the MN-CSE on the nearest MEC node evaluates real-time VRU positions and sends instant collision-risk alerts directly to the vehicle. The MEC platform further enhances this process by providing network-aware optimizations, such as selecting the best available connection (5G, LTE-V2X) for data exchange. As the HV moves across different zones, the system seamlessly transitions tasks between edge nodes or back to the cloud, ensuring uninterrupted service. By combining cloud scalability with edge-level agility, this approach not only improves road safety but also demonstrates how IoT and edge computing can converge to support latency-critical automotive applications.
An offloading concept locating tasks and resources to a place where close to users can be applied to this VRU detection service. In this case, a service can be provided to users with very short delay.
![Figure 5.2.3-1: Scenario of resource and tasks offloading to support VRU application.](media/usecase_vru.png)
**Figure 5.2.3-1: Scenario of resource and tasks offloading to support VRU application.**
## 5.3 Industrial & Robotics
### 5.3.1 Description
This clause covers the industrial and robotics domain with a focus on use cases such as Swarm-based Autonomous Ant Delivery Optimization, Smart Warehouse Automation, and Industrial Digital Twins. It explores how MEC and oneM2M frameworks enable intelligent coordination, real-time control, and seamless IoT integration in complex industrial and logistics environments. By leveraging edge capabilities, these use cases benefit from low-latency processing, enhanced automation, and improved operational efficiency across urban and industrial ecosystems.
### 5.3.2 Swarm-based Autonomous Ant Delivery Optimization
This scenario explores a hybrid swarm-based autonomous delivery system inspired by ant colony behavior. Swarm computing is a computational model based on the collective behavior of decentralized systems, often inspired by nature (e.g., ants, bees, or fish schools). In traditional swarm computing, multiple objects (agents) collaborate autonomously to solve problems or perform tasks by sharing information and interacting with each other without the need for centralized control. Each agent in the swarm makes decisions based on local information and interactions, leading to emergent global behaviors that solve complex tasks, such as pathfinding, resource allocation, or optimization.
However, when applied to real-world systems like autonomous delivery or energy management, fully decentralized swarm systems face several limitations. For instance, without any centralized coordination or additional guidance, the swarm may face challenges in optimizing complex tasks or adapting to dynamic, large-scale environments. Swarm systems can struggle with task allocation, resource optimization, and dealing with environmental changes when working entirely autonomously without external help. Thus, this use case explores how swarm agents, integrated with a cloud IoT platform and assisted by edge nodes, can more effectively solve given problems.
In this scenario, the edge IoT platform (e.g., MN in oneM2M), which acts as an edge node in a decentralized system, plays a critical role in overcoming these limitations. While robot (agents) in the swarm communicates and collaborates autonomously, the edge IoT platform provides indirect feedback by offering information that robots may not directly perceive, such as road conditions, elevator status, and indoor obstacles. This feedback mechanism enhances the swarm’s performance and helps in optimizing the delivery task. Specifically, the edge computing platform (e.g., MEC) allows for ultra-low latency processing and ensures that robots can receive real-time updates on the environment. By integrating the edge IoT platform for efficient resource management and the edge computing platform for real-time environmental feedback, this hybrid swarm system becomes more capable of optimizing delivery paths and avoiding congestion, while still maintaining decentralized autonomy.
As the swarm operates, it gradually evolves into a highly efficient delivery network, capable of real-time obstacle avoidance and dynamic path adaptation. This process exemplifies the power of Swarm Computing when integrated with an IoT platform deployed on Edge infrastructure. In this context (depicted also in Figure 5.3.2-1), Swarm Computing refers to a decentralized approach in which multiple autonomous agents—represented as Application Entities (AEs) within the oneM2M framework—interact and collaborate using locally available information. These agents are able to solve tasks independently, without relying on centralized control, resulting in emergent collective intelligence.
The IoT platform, implemented as an edge node (such as a Middle Node, MN, in oneM2M), plays a pivotal role by providing indirect feedback to the swarm. It supplies real-time environmental data that individual agents may not directly perceive, thereby enhancing the swarm’s overall performance and adaptability. This feedback mechanism enables agents to make informed decisions, optimize delivery routes, and respond effectively to changing conditions.
Complementing the IoT platform, the Edge infrastructure—exemplified by the MEC platform—processes data locally to deliver ultra-low-latency feedback. By hosting both IoT services and computational resources at the edge, the MEC platform ensures that swarm agents receive timely updates and actionable insights. This local processing capability is crucial for improving coordination among agents, optimizing task execution, and maintaining operational efficiency in dynamic environments.
Together, the integration of Swarm Computing, edge-based IoT platforms, and MEC infrastructure demonstrates a robust solution for complex delivery optimization challenges. The system’s ability to self-organize, adapt to obstacles, and continuously refine its operations highlights the transformative potential of combining decentralized intelligence with real-time edge computing.
![Figure 5.3.2-1: Data flow in a hybrid swarm system where robots share pheromone maps and report routes and obstacles to the MN-CSE, which uses MEC-provided real-time data to generate virtual pheromones and optimize swarm behavior.](media/usecase_swarm_robots.png)
**Figure 5.3.2-1: Data flow in a hybrid swarm system where robots share pheromone maps and report routes and obstacles to the MN-CSE, which uses MEC-provided real-time data to generate virtual pheromones and optimize swarm behavior.**
### 5.3.3 Smart Warehouse Automation
This use case describes how ETSI MEC and oneM2M platforms can work together to enable real-time warehouse automation, including coordination of autonomous guided vehicles (AGVs), environmental monitoring, and asset tracking. The system architecture reported in Figure 5.3.3-1 integrates a centralized oneM2M IN-CSE for managing warehouse infrastructure data and multiple MN-CSE instances deployed as MEC applications to provide low-latency control and decision-making.
The oneM2M IN-CSE collects and stores data from IoT devices such as temperature and humidity sensors, RFID tags, and AGV telemetry. When time-critical operations occur (e.g., routing AGVs, responding to a gas leak, or dynamic reallocation of assets), relevant tasks are offloaded to MEC-based MN-CSEs. The MEC platform processes this data locally, allowing immediate decisions such as rerouting AGVs, sending alerts to personnel, or controlling air filtration systems.
As warehouse operations span large areas, AGVs may move between MEC zones. The MEC platform supports service continuity by transferring control logic and task context across MEC nodes, ensuring uninterrupted automation and safety.
![Figure 5.3.3-1: High-level overview of the warehouse automation scenario.](media/usecase_smart_warehouse.png)
**Figure 5.3.3-1: High-level overview of the warehouse automation scenario.**
### 5.3.4 Industrial Digital Twins
This use case illustrates the integration of the oneM2M IoT platform with the ETSI MEC edge computing framework to support Industrial Digital Twins (IDTs) in smart manufacturing environments. The goal is to enable continuous monitoring, analysis, and optimization of industrial processes by deploying synchronized digital representations of physical assets across both cloud and edge infrastructures. This approach is fundamental to enabling real-time decision-making, predictive maintenance, and autonomous control in dynamic and distributed industrial settings.
The overall system architecture illustrated in Figure 5.3.4-1 relies on a cloud-based oneM2M IN-CSE (Infrastructure Node Common Service Entity), which functions as a centralized orchestrator and digital repository, and multiple MN-CSEs (Middle Node Common Service Entities) deployed at the edge on MEC nodes physically co-located with manufacturing equipment or microfactories. These edge-based MN-CSEs are responsible for processing time-sensitive data streams such as sensor readings from production lines, robotic cell statuses, energy usage metrics, and environmental conditions with ultra-low latency.
Real-time insight is achieved through the ability of MN-CSEs to execute localized, time-critical functions including anomaly detection, in-line quality control, and safety compliance checks. The system ensures seamless service continuity across the edge-cloud continuum by dynamically migrating services and synchronizing digital twin data as assets move across operational zones. This continuity is crucial for uninterrupted performance of digital twins as the physical counterparts transition between factory cells or sites. The solution is grounded in oneM2M’s standardized resource and data models, enabling interoperability across diverse devices and platforms.
Initially, the IN-CSE in the cloud hosts the master representations of industrial assets—such as CNC machines, collaborative robots (cobots), or Automated Guided Vehicles (AGVs)—maintaining records of their operational states, historical data, and environmental contexts. When local operations are initiated in a specific factory cell that is served by a MEC node, the IN-CSE delegates processing to the nearest MN-CSE deployed on the edge. This MN-CSE ingests live data streams including vibration patterns, temperature levels, system workload, and diagnostic logs, and performs localized analytics such as early fault detection, micro-adjustments to optimize yield, and predictive maintenance alerts. Additionally, the MN-CSE may interact with other applications hosted on the MEC platform, such as AI-based vision systems for quality inspection or real-time control systems for robotic motion, by leveraging MEC-provided APIs.
As mobile industrial assets like AGVs or modular production units relocate within or between factory environments, the system ensures that all associated digital twin data and computational contexts are seamlessly migrated to a new MN-CSE at the appropriate MEC node. This handover is coordinated by the IN-CSE and utilizes MEC’s Application Mobility and Platform Services to maintain uninterrupted functionality. Throughout this process, the oneM2M system keeps cloud and edge representations synchronized, enabling both high-level analytics—such as key performance indicators and production benchmarking—and immediate operational control. The benefits of this integrated approach are significant. Sub-second response times are achieved by executing control loops directly at the edge, reducing latency to a minimum. The architecture supports uninterrupted service delivery even during the movement of assets between operational zones, enhancing system reliability. Its modular and scalable nature allows deployment across a wide range of industrial environments. Finally, adherence to oneM2M standards and MEC orchestration capabilities guarantees compatibility and future-proof integration across heterogeneous platforms.
![Figure 5.3.4-1: Industrial Digital Twins supported and enabled for both IoT/IIoT interactions and edge-cloud computation by the integrated capabilities of oneM2M and ETSI MEC.](media/usecase_digital_twins.png)
**Figure 5.3.4-1: Industrial Digital Twins supported and enabled for both IoT/IIoT interactions and edge-cloud computation by the integrated capabilities of oneM2M and ETSI MEC.**
## 5.4 Maritime
### 5.4.1 Description
The maritime sector is undergoing significant transformation through the adoption of automation and advanced sensing technologies that enable vessels to navigate and perform tasks autonomously. These connected ships must maintain continuous communication with operation centers across the edge-cloud continuum to ensure effective monitoring and decision-making. This section presents a use case that integrates the oneM2M IoT platform with the ETSI MEC edge computing framework to deliver real-time assisted manoeuvring for unmanned vessels in seaport areas, supporting low-latency data processing, seamless service continuity, and mission-critical functionalities.
### 5.4.2 Assisted Manoeuvring for Autonomous Ship
Maritime Autonomous Surface Ships (MASS) technologies have been developing rapidly during last decade. Unmanned ships should be able to navigate without colliding with other vessels, considering the naval traffic and unexpected situations. These ships are usually equipped with advanced technologies like integrated automation systems based on AI-models and sensing capabilities which allow them to navigate, make decisions and perform tasks autonomously. They usually rely on the communication with the Remote Operation Centers (ROCs) serving as command hubs for monitoring, controlling and managing ships from the shore. ROCs serve as a pivotal hub for overseeing and controlling unmanned vessels remotely from onshore locations so that operators can utilize advanced sensor technologies and data telemetry to monitor vessel position, speed, course, environmental conditions, and other operational parameters in real-time.
The ROC is implemented following a cloud-based architecture, which is able to receive data coming from the on-board equipment (e.g., LIDARs, Radars, Cameras, IoT sensors, AIS, etc.), store and process it using AI-based models so that to provide early warnings and prevent risks in advance. The communication between the vessel and the ROC is performed exploiting existing low-latency communication technologies such as 5G NR and beyond. This scenario proposes a collaborative architecture between oneM2M-compliant platform and ETSI MEC edge computing framework to provide assisted manoeuvring services to the unmanned ships in real-time within the seaport waters.
In this architecture, schematically represented in Figure 5.4.2-1, edge-located MN-CSE instances are deployed as MEC applications on distributed edge nodes to process data coming from the vessel when it approaches the assigned berth or departs from it. On top of this, a cloud-based oneM2M IN-CSE instance is deployed as central hub for storing and managing data related to the vessel such as position, speed, course and environmental conditions. When the vessel enters the area covered by a MEC node, the cloud IN-CSE automatically offloads computational tasks (e.g., localization of other vessels) to the MN-CSE on the edge to guarantee low-latency processing capabilities for the generation of possible warnings. Using this approach, the MN-CSE can immediately analyse data coming from the vessel instead of relying on distant cloud servers.
As the vessel transits through different MASS zones, the proposed oneM2M/MEC architecture allows to seamlessly move tasks between edge nodes, ensuring service continuity and supporting mission-critical services for the unmanned vessels (e.g., assisted maneuvering, collision avoidance and situational awareness).
![Figure 5.4.2-1: Scenario for assisted manoeuvring exploiting oneM2M and ETSI MEC.](media/usecase_ship_maritime.png)
**Figure 5.4.2-1: Scenario for assisted manoeuvring exploiting oneM2M and ETSI MEC.**
## 5.5 Metaverse
### 5.5.1 Description
The metaverse refers to a persistent, shared, and immersive digital environment where users interact with each other, digital objects, and services through avatars or virtual representations. It encompasses a range of technologies—including Virtual Reality (VR), Augmented Reality (AR), 3D web platforms, and real-time data integration—that collectively enable seamless blending of digital and physical experiences. In the context of IoT and edge computing, the metaverse can extend traditional retail, education, entertainment, and social interaction by connecting virtual spaces with live data and intelligent services from the real world. For example, a virtual shopping mall in the metaverse can mirror the inventory and layout of a physical store, allowing users to browse, interact, and purchase items as if they were present on-site. Similarly, virtual classrooms can leverage real-time sensor data and edge analytics to create adaptive, engaging learning environments. This clause introduces a use case that integrates the oneM2M IoT platform with the ETSI MEC edge computing framework to enable a smart virtual shopping service. By combining live data from IoT-enabled physical stores with edge-hosted AI analytics, the system delivers personalized, low-latency interactions within the metaverse, ensuring a synchronized, context-aware user experience that connects virtual behaviour with real-world actions.
### 5.5.2 Smart Shopping with Edge-AI and Cloud IoT Integration
This use case describes a smart virtual shopping service within a metaverse environment, where users can experience immersive, interactive shopping that is tightly integrated with real-world data and edge intelligence (schematically depicted in Figure 5.5.2-1). A user enters a virtual shopping mall through a metaverse interface—either via VR/AR devices or standard 3D web platforms. Inside the virtual store, the layout, available products, and store conditions reflect those of an actual physical store in real time. This synchronization is made possible through IoT devices deployed in the physical store, which are directly connected to a cloud-based oneM2M platform (IN-CSE). These devices stream data such as shelf inventory levels, product locations, environmental conditions (e.g., temperature), and user presence to the clould IoT platform.
As the user browses virtual shelves or picks up digital items, these actions are interpreted by an application on the edge MEC platform running near the physical store. The MEC application subscribes to relevant IoT data from the IN-CSE and uses AI models to analyze both real-time device states and user behavior. For example, if a user shows interest in a particular product—such as hovering over or virtually picking up a package—the MEC App considers stock levels, previous interaction patterns, and related product information to generate real-time recommendations ('People who looked at this also bought...”) or alerts (“Only 2 left in stock!”). This feedback is sent instantly back to the metaverse interface, enhancing engagement and encouraging informed purchasing decisions. When the user confirms a purchase in the virtual store, the system triggers real-world processes: the item is marked as reserved or sold in the cloud platform (IN-CSE), and notifications are sent to logistics systems, shelf displays, or even store staff.
If needed, IoT actuators in the real store—such as digital signage or voice assistants—can respond with contextual actions (e.g., displaying pickup instructions). The entire experience feels seamless to the user: every virtual interaction is grounded in live, physical-world data, while AI-driven recommendations create a personalized, context-aware shopping journey. The user benefits from smart, immersive shopping, while the system ensures real-time linkage between the digital and physical layers, powered by oneM2M and MEC.
![Figure 5.5.2-1: Smart Shopping with Edge-AI and Cloud IoT Integration.](media/usecase_smart_shopping.png)
**Figure 5.5.2-1: Smart Shopping with Edge-AI and Cloud IoT Integration.**
## 5.6 Future Home
### 5.6.1 Description
The Smart Home domain is rapidly evolving toward immersive, responsive, and personalized services powered by real-time data and intelligent automation. This clause presents a use case that integrates the oneM2M IoT platform with the ETSI MEC framework to enable advanced home experiences such as real-time media interactions, remote education, elderly care, and intelligent automation.
### 5.6.2 User Premises Edge and oneM2M Integration
This use case showcases the integration of ETSI MEC and oneM2M platforms to support immersive, responsive, and personalized Smart Home services. It addresses real-time media processing (e.g., holoportation), education delivery, elderly care (aging in place), and home automation. The architecture features a centralized oneM2M IN-CSE deployed on MEC node that manages smart devices across the multiple homes, and localized MN-CSEs deployed on non-MEC nodes (e.g., within an ISP gateway or smart home hub) that perform latency-sensitive tasks such as video processing, sensor fusion, and context-aware automation. If MEC supports CPEs, then MEC systems can discover them and MEP enables MEC applications to offload real time data processing tasks to this non-MEC systems.
In this extension repoterd in Figure 5.6.2-1, the non-MEC node is realized as a Customer Premises Edge device (have support of both computational as well as Multi-access network communication capabilities) deployed locally at the user's home (e.g., home gateway or dedicated edge server). This edge device hosts MEC applications or MN-CSE components and serves as the key enabler for local, low-latency processing of Smart Home functions, reducing dependency on centralized cloud services.
![Figure 5.6.2-1: Future Home user premises scenario with integrated Edge and IoT functionalities.](media/use_case_future_home.png)
**Figure 5.6.2-1: Future Home user premises scenario with integrated Edge and IoT functionalities.**
# 6 MEC-oneM2M Architectural & Use Case Mapping
## 6.1 Introduction
This clause provides an analysis of the main reference architectures and characteristics of MEC and oneM2M , highlighting their core functionalities and how they complement each other in the context of Edge-IoT deployments. For each of the identified use cases, the clause maps the relevant architectural elements and functionalities from both, illustrating how they contribute to the scenario. It also identifies potential new features or enhancements that could further support the use case. Where overlapping functionalities exist across use cases, these are acknowledged and will be further analyzed in the following clause 7, which focuses on deriving structured application requirements and identifying gaps for future standardization efforts.
## 6.2 MEC Frameworks
## 6.2 MEC Frameworks
ETSI ISG MEC (European Telecommunications Standards Institute Industry Specification Group for Multi-access Edge Computing) is the home of technical standards for edge computing. The work of the MEC initiative aims to unite the telco and IT-cloud worlds, providing IT and cloud-computing capabilities within the RAN (Radio Access Network). The MEC ISG specifies the elements that are required to enable applications to be hosted in a multi-vendor multi-access edge computing environment.
ETSI MEC framework (schematically illustrated in Figure 6.2-1) offers application developers and content providers, cloud-computing capabilities at the edge of the network, in an environment characterized by ultra-low latency and high bandwidth together with real-time access to radio network information that can be leveraged by applications.
![Figure 6.2-1. Multi-access edge system reference architecture.](media/mec_arch_6_1.png)
**Figure 6.2-1: Multi-access edge system reference architecture**
Figure 6.2-1 depicts the generic multi-access edge system reference architecture. The MEC framework defined by ETSI GS MEC 003 [i.1] includes a standardized architecture with the following key functional elements:
- MEC Orchestrator: manages resources and coordinates MEC operations.
- MEC Host: includes the MEC platform, MEC applications, and the underlying virtualization infrastructure.
- MEC Platform (MEP): provides collection of essential functionalities required to run MEC applications on a particular Virtualisation infrastructure and enable them to provide and consume MEC services. The MEC platform can also provide services.
- MEC Applications: instantiated and run on the Virtualisation infrastructure based on configuration or based on requests validated by the MEC management.
- API GW for Clients: controls access to MEC applications by client applications
- MEP Manager: handles the lifecycle management of the MEC platform.
- Virtualisation infrastructure manager (VIM): Mainly allocating, managing and releasing Virtualised (compute, storage and networking) resources of the Virtualisation infrastructure
MEC also offers another reference architecture variant for MEC in NFV, MEC federation and Support for Security Monitoring and Management. The architecture also defines the following multiple reference points across the various provided reference architecture variants:
- Mp: connects MEC applications with the MEC platform.
- Mm: used for management operations.
- Mx: links to external systems.
- Mff: enables federation with other MEC or non-MEC systems.
To support applications running at the edge, ETSI has developed a set of standardized service APIs. This local services environment is a flexible and extendable framework, as new services can be introduced by following the API guidelines in ETSI GS MEC 009, when creating new service APIs. These APIs allow applications to access useful network and context information through the Mp1 interface. MEC applications (API producer) can also register new services to MEP and consume/subscribe to existing ones through this interface. Transition from MEC Phase 1 to MEC Phase 4 (period 2024-2026) provided MEC service APIs include:
- Application lifecycle, rules and requirements management (ETSI GS MEC-010-2)
- Edge Platform Application Enablement (ETSI GS MEC-011)
- Location API (ETSI GS MEC-013)
- Radio Network Information API (ETSI GS MEC-012)
- Traffic Management API (ETSI GS MEC-015)
- Device Application Interface (ETSI GS MEC-016)
- Application Mobility Service API (ETSI GS MEC-021)
- WLAN Information API (ETSI GS MEC-028)
- Fixed Access Information API (ETSI GS MEC-029)
- V2X Information API (ETSI GS MEC-030)
- IoT API (ETSI GS MEC-033)
- Federation Enablement API (ETSI GS MEC-040)
- QoS Measurement API (ETSI GS MEC-045)
- Sensor Sharing API (ETSI GS MEC-046)
All APIs are described using the OpenAPI Specification (OAS), ensuring consistency and ease of integration across platforms.
## 6.3 oneM2M Components
### 6.3.1 Introduction to oneM2M
oneM2M is a global standards initiative that develops specifications for a horizontal platform (or Service Layer) used for the exchange and sharing of data among Internet of Things applications. Individual IoT solutions using oneM2M can share data and resources through common service layer functions such as resource discovery and semantic interoperability. As a result, application developers, solution providers and data suppliers can share data between applications that span multiple verticals and reduce their dependence on single-vendor solutions.
oneM2M provides a layer of abstraction - similar to an operating system - that shields developers from the details of network communication protocols, interfaces to IoT device and device management. It provides a framework for interworking with different technologies, for example 3GPP and Modbus.
It is also designed as a distributed software layer, making it particularly suitable for use in Edge computing deployments, mitigating the burdens on data centers and core networks and improving communication latency by acquiring, processing, and storing data at the edge of network.
### 6.3.2 oneM2M Architecture
The oneM2M service layer is a general-purpose standard that applies to all industry verticals. It brings together all components in the IoT solution stack and specifies a distributed software/middleware layer, sitting between applications and underlying communication networking HW/SW.
oneM2M service layer can be integrated into devices, gateways, and servers. Figure 6.3.2-1 shows a simplified oneM2M architecture:
![Figure 6.3.2-1: oneM2M architecture.](media/oneM2M_architecture.png)
**Figure 6.3.2-1: oneM2M architecture**
This architecture is composed of the following entities:
- _Common Services Entity (CSE)_: Provides the set of "service functions" that are common to multiple IoT domains.
- _Application Entity (AE)_: Represents application logic for the end-to-end IoT solutions. This includes device firmware, gateway logic or backend applications.
- _Interworking Proxy application Entity (IPE)_: A specialized AE that is used to interface between oneM2M and an external system.
- _Network Services Entity_: Provides networking services to CSEs that are more than just data transport.
- _Node_: Logical equivalent of a physical (or possibly virtualized) server or device.
oneM2M defines several different kinds of node, which are named based on their use by oneM2M.
- _Infrastructure Node (IN)_: A Cloud provider / company server. It hosts a CSE that is referred to as an IN-CSE and may host AEs. There is exactly one IN in the Infrastructure Domain of each oneM2M Service Provider.
- _Application Service Node (ASN)_: A Node that contains at least one AE co-located with a CSE (an ASN-CSE).
- _Middle Node (MN)_: A Node that contains one CSE (an MN-CSE) and zero or more AEs. It is typically found in an Edge Server or Gateway box.
- _Application Dedicated Node (ADN)_:A Node that contains at least one AE but does not contain a CSE.
In addition, the oneM2M architecture defines one other type of node, the so-called “Non-oneM2M Node” (NoDN), not shown in Figure 6.3.2-1. This is a node that does not contain native oneM2M Entities (neither AEs nor CSEs). Typically, such nodes would host some non-oneM2M IoT implementations or legacy technology which can be connected to the oneM2M system via interworking proxies or agents.
The principal _Reference Points_ defined by oneM2M are:
- _Mca_: The interface used by an AE to use the common services provided by the CSE.
- _Mcc_: Communication flows across the Mcc reference point enable a CSE to use the services supported by another CSE.
- _Mcc'_: Communication flows between CSEs that belong to different oneM2M service providers.
- _Mcn_: Communication flows between a Common Services Entity (CSE) and the Network Services Entity (NSE) allowing the CSE to use the network services (other than transport and connectivity) provided by the NSE.
### 6.3.3 oneM2M Architecture Common Service Layer Functions
From a functional perspective, oneM2M has defined fourteen common service functions (CSFs) as shown in Figure 6.3.3-1 below. These relate to network connectivity, device security, transport protocols, content serialization, IoT device services and management and IoT semantic ontologies. A given CSE does not have to implement every CSF and CSE's can communicate with each other giving an AE or CSE the ability to access oneM2M resources that are hosted on a remote CSE.
![Figure 6.3.3-1: Common Service Functions.](media/oneM2M_CSFs.png)
**Figure 6.3.3-1: Common Service Functions**
These oneM2M services are defined so that application developers can focus on application-specific functionality (e.g., turning a switch on or off), while relying on abstractions provided by oneM2M to mask the underlying technology-specific details, thus allowing bindings to different communications stacks and protocols such as HTTP, CoAP and MQTT. For example, a simple switch might use a fixed or Wi-Fi network, a CoAP or HTTP transport. It might use a JSON or XML serialization, an Open Connectivity Foundation (OCF) or thread service enablement, or an ontology based on Smart Appliances REFerence (SAREF) or W3C’s Thing Description.
Finer-grained capabilities underpin each service function as illustrated in the case of device management, security and application and service management functions. In keeping with a philosophy of leveraging existing standards rather than re-inventing them, oneM2M complements existing and proven security technologies to address IoT security challenges. It provides a common set of security capabilities to secure IoT devices and data; and to prevent or mitigate attacks. This is made possible by an abstracted set of security-related APIs designed to simplify security for devices and applications. oneM2M is a constantly evolving standard with a strategic roadmap designed to address new IoT requirements. This is made possible by the common service layer concept, which was conceived to accommodate new service functions.
### 6.3.4 Mapping between MEC and oneM2M
MEC applications can both consume services available in the MEC system and also produce services, which are made available by the MEC platform to other applications.
In the context of Internet of Things (IoT) and Machine-to-Machine (M2M) communications, we can consider the following mapping of oneM2M entities with ETSI MEC entities:
- A common service function (e.g., Network Service Function) or CSE in the oneM2M architecture can be represented as a MEC Service and/or as a service-producing MEC App instance. This service would be exposed by the MEC platform to be connected to (authorized) consumer Application Entities (AEs).
- Similarly, an AE in the oneM2M architecture can be implemented as a _MEC Application_ in an ETSI MEC system.
Architectural interworking between ETSI MEC and oneM2M is made possible by seeing the CSE and AE functional elements of oneM2M as particular instances of MEC services and applications from the point of ETSI MEC system and by providing MEC-specific IPE'a. More specific hooks can be envisaged to effectively integrate these two systems and gain more benefits from tighter integration (for example, specific mechanisms for application onboarding and instantiation, or additional MEC service APIs).
### 6.3.5 Virtualization of oneM2M Common Service Layer functions
Network Virtualization allows IT service providers to develop an end-to-end digital service based on network as a service (NaaS). Network Function Virtualization (NFV) targeted the oneM2M common service layer functions where the virtualization technologies are used to replace hardware servers hosting a common set of IoT functions with VNFs that can run as software on virtual machine. On top of appropriate virtualization infrastructure, VNFs of oneM2M service functions can be deployed anywhere in the network on-demand. NFV enables to slice an oneM2M common service layer in the form of multiple virtual IoT CSFs at the network edge. To achieve this, virtualization technologies allow the transformation of IoT CSFs from the common service layer to virtual IoT CSFs like software images. These virtual IoT CSFs include resources and service functions that have attributes specifically designed to meet the needs for IoT vertical markets such as smart building, industrial IoT, smart city, and, subsequently, create a network-as-a-service model for oneM2M edge computing.
## 6.4 Use Cases & Frameworks Mapping
This section provides a synthesis of the previously identified use cases in section 5 and the architectural features of the MEC and oneM2M frameworks. It maps each use case to the relevant framework components and functionalities, showcasing how MEC and oneM2M support specific operational needs. The mapping highlights the alignment between the frameworks and the identified scenarios, while also pointing out areas where enhancements or additional integration mechanisms may be required to better support cross-domain use cases.
### 6.4.1 Deployment Options
The selection of the most appropriate deployment option for each use case is guided by insights from the ETSI white paper[i.2], where 4 deployment options are mentioned for interworking of oneM2M and MEC technologies.
#### 6.4.1.1 Option A: deploy the oneM2M as a cloud, MEC as an edge
This option captures the MEC and oneM2M standard as they exist at the start of the effort to integrate both standards with only changes to support MEC devices and oneM2M devices (see Figure 6.4.1.1-1). This can be done by developing applications that use the oneM2M Mca reference point and the MEC Mp1 reference point.
- An oneM2M IN-CSE is deployed in the cloud and an IN-AE represents the application that is connected to the IN-CSE.
- The edge gateway connects IoT devices to the MEC framework to provide services to the devices.
- The MEC gateway is connected to the IN-CSE through an MEC IPE. The IN-CSE represents the MEC platform and devices that are registered to the MEC platform using the MEC IPE.
- devices may use MEC services and APIs (Mp1) with support from the MEC IPE hosted as an MEC APP.
- devices may use non-MEC services, e.g. UPC-UA devices, using a 3rd party IPE hosted as an MEC APP.
- devices may use oneM2M services through a Mca proxy. This Mca proxy can be hosted as an MEC APP. These oneM2M devices can also take advantage of exposed MEC services.
![Figure 6.4.1.1-1. Deployment Option A.](media/OptionA.png)
**Figure 6.4.1.1-1. Deployment Option A.**
Although some of the benefits of edge computing can be achieved through network and processing, the advantages of 100% edge computing are not fully utilized because the cloud remains the final endpoint where data is stored and managed.
- A standard document providing interoperability and interworking between the two platforms, oneM2M and ETSI MEC, is required.
#### 6.4.1.2 Option B: oneM2M and MEC as an edge with the different physical node
This option illustrated in Figure 6.4.1.2-1 enhances the performance of the edge deployment by introducing additional services through the integration of a oneM2M MN-CSE. An oneM2M IPE is added that interworks MEC services and devices. This can be realized using an exemplar setup consisting of one platform with an MEC framework and a second platform hosting the MN-CSE.
- An oneM2M IN-CSE is deployed in the cloud and an IN-AE represents the application that is connected to the IN-CSE.
- The edge consists of the MEC platform that connects devices using MEC apis.
- The edge also includes an oneM2M MN-CSE that offers services to oneM2M devices and other types of devices.
- The MEC IPE is deployed with the oneM2M platform or the MEC platform.
- Additional services can be exposed by oneM2M AEs as such interworking with other devices, swarm computing and federated learning.
- These additional oneM2M AEs can be deployed on either the oneM2M platform or the MEC platform.
![6.4.1.2-1. Deployment Option B.](media/OptionB.png)
**6.4.1.2-1. Deployment Option B.**
This deployment uses oneM2M IoT service providers and ETSI MEC entities that may be different, this scenario can still be used in the early edge computing market by using the same custom MEC IPE that was defined in option A. Additionally, new capabilities from oneM2M can be delpoyed to support compute intensive services.
- A standard document providing interoperability and interworking between the two platforms, oneM2M and ETSI MEC, is required.
#### 6.4.1.3 Option C: oneM2M and MEC in the same physical edge node
This option focuses on integrating the services of oneM2M and ETSI MEC as reported in Figure 6.4.1.3-1. This achieves better performance by combining the oneM2M MN-CSE and the MEC framework. The MEC IPE is no longer needed because the oneM2M and MEC framework are integrated.
- An oneM2M IN-CSE is deployed in the cloud and an IN-AE represents the application that is connected to the IN-CSE.
- The edge consists of the MEC platform that connects devices using MEC apis and the oneM2M MN-CSE to support a variety to devices.
- The MEC IPE is replaced with new APIs that take advantage of new standards work from ETSI MEC and oneM2M.
![6.4.1.3-1. Deployment Option C.](media/OptionC.png)
**6.4.1.3-1. Deployment Option C.**
In this scenario, oneM2M and MEC platforms are installed and operated on the same physical edge node. This can significantly improve services by eliminating unnecessary data movement and information exchange. It also supports customizing the system to take advantage of high compute platform capabilities that may be needed for federated learning or swarm computing.
- A standard document providing interoperability and interworking between the two platforms, oneM2M and ETSI MEC, is required.
#### 6.4.1.4 Option D: oneM2M and MEC are tightly coupled in the same edge node
This option attempts to simplify the services by offered by ETSI MEC and oneM2M by fully integrating the services as schematically illustrated in Figure 6.4.1.4-1.
- An oneM2M IN-CSE is deployed in the cloud and an IN-AE represents the application that is connected to the IN-CSE.
- The edge consists of the MEC platform that connects devices using MEC apis and the oneM2M MN-CSE to support a variety to devices.
- MEC exposes APIs that integrate the oneM2M services.
![6.4.1.4-1. Deployment Option D.](media/OptionD.png)
**6.4.1.4-1. Deployment Option D.**
In this scenario oneM2M integrates to MEC and offers services that support directly supporting providing data source, processing, and multi-access networking.
- A standard document providing interoperability and interworking between the two platforms, oneM2M and ETSI MEC, is required.
### 6.4.2 Autonomous Vehicle with Continuous Edge Computing
#### 6.4.2.1 Use Case Driving Deployment
For the “Autonomous Vehicle with Continuous Edge Computing” use case, achieving ultra-low latency, context awareness, and localized processing is essential to handle data from nearby sources and enable real-time analytics. To meet these requirements, certain tasks can be offloaded to oneM2M Edge Instances (MN-CSEs).
As autonomous vehicles move across MEC zones, orchestration and synchronization across multiple oneM2M platforms is required. For this, a centralized IN-CSE deployed in the cloud is suitable to maintain global context, coordinate edge instances, and ensure seamless service continuity.
In Table 6.4.2.1-1, we consider all the relevant operational requirement for this use case.
##### Table 6.4.2.1-1 – Operational Requirements and Platform Support for Autonomous Vehicle with Continuous Edge Computing
| Operational Requirement | Support in MEC | Support in oneM2M |
| ----------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------- |
| Up and running IoT platform | Not Required | Supported: oneM2M platform can be deployed at cloud or any edge node |
| Registration of vehicles, IoT devices, and applications | Not Required | Supported: oneM2M features provide Commom Service Functions (CSFs) to register both devices and AEs (Application Entities) |
| Data collection from IoT Devices (Vehicle sensors, Road sensors, LIDAR) | Not Required | Supported via oneM2M’s Common Service Layer using Mca interfaces for data exchange |
| Instantiate of MN-CSE on Edge Node | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiation of AE on Edge Node | AE can be instantiated as MEC Application on MEC Host | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
### 6.4.3 Vulnerable Road Users
#### 6.4.3.1 Use Case Driving Deployment
The “Vulnerable Road Users” (VRU) use case shares deployment characteristics with the “Autonomous Vehicle with Continuous Edge Computing” scenario, where achieving ultra-low latency, localized processing, and context awareness is critical. This is necessary to effectively process data from nearby roadside units, sensors mounted on traffic signals, and other urban infrastructure to enable real-time analytics and immediate response. To support this, specific processing tasks can be offloaded to oneM2M Edge Instances (MN-CSEs) operating close to the data sources. As vehicles and VRUs dynamically move across different geographic zones, coordination across multiple MEC nodes becomes essential. Therefore, a centralized IN-CSE deployed in the cloud is ideal for maintaining global context, orchestrating and synchronizing edge instances, and ensuring uninterrupted service continuity. Meanwhile, MN-CSEs should be deployed close to the edge to provide low-latency decision support
In Table 6.4.3.1-1, we consider all the relevant operational requirement for this use case.
##### Table 6.4.3.1-1 – Operational Requirements and Platform Support for Vulnerable Road Users
| Operational Requirement | Support in MEC | Support in oneM2M |
| ----------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
| Up and running IoT platform | Not Required | Supported: oneM2M platform can be deployed at cloud or any edge node |
| Registration of vehicles, IoT devices, and applications | Not Required | Supported: oneM2M features provide CSFs to register both devices and AEs (Application Entities) |
| Data collection from IoT Devices (Vehicle sensors, Road sensors, LIDAR) | Not Required | Supported via oneM2M’s Common Service Layer using Mca interfaces for data exchange |
| Instantiate of MN-CSE on Edge Node | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiation of AE on Edge Node | AE can be instantiated as MEC Application on MEC Host | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
### 6.4.4 Swarm-based Autonomous Ant Delivery Optimization
#### 6.4.4.1 Use Case Driving Deployment
The “Swarm-based Autonomous Ant Delivery Optimization” use case enables optimized coordination of autonomous delivery robots using distributed intelligence and real-time data exchange. The oneM2M IN-CSE is deployed in the cloud to manage global swarm state, while MN-CSEs are instantiated at the edge to handle local swarm behavior.
Swarm applications (e.g., route optimization, obstacle detection) are deployed on edge nodes and interact with MN-CSEs for low-latency data processing. Robots continuously report environmental data, pheromone map updates, and coordination feedback to the platform.
As robots move across delivery zones, the system ensures seamless session handover and synchronization of swarm intelligence using MN-CSE coordination. This setup supports dynamic optimization, real-time feedback, and resilient autonomous operations as schematically reported in Table 6.4.4.1-1.
##### Table 6.4.4.1-1 – Operational Requirements and Platform Support for Swarm-based Autonomous Ant Delivery Optimization
| Operational Requirement | Support in MEC | Support in oneM2M |
| ---------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- |
| Up and running IoT platform | Not directly required | Supported: oneM2M IN-CSE can be deployed at cloud |
| Registration of devices (Swarm Robots) | Not required | Supported: oneM2M enables registration of sensors, actuators, controllers as either IN-AEs or ADN-AE via standardized resource structures and CSFs |
| Registration of applications | Not required | Supported: oneM2M enables registration of Applications as AEs via standardized resource structures and CSFs |
| Real-time data ingestion from AEs (pheromone map updates, route feedback and obstacle reports) | Not required | Supported via oneM2M’s CSF and Mca interfaces |
| Instantiate of MN-CSE on Edge Node | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiate of AE (e.g., Swarm optimization) on Edge Node | AE can be instantiated as MEC Application on MEC Host | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
| Low-latency control for Hybrid Swarm Robots (feedback for Swarm optimization) | Supported: MEC Apps handle time-sensitive operations with minimal delay | Supported: MN-CSE enables fast coordination through subscriptions, notifications, and data routing |
| Offloading low latency tasks of IoT Platform for real time data processing | Not required. But MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: Tasks can be offloaded to MN-CSE instances using Mcc interface mechanisms. |
| Service continuity during Hybrid Swarm Robot zone transitions | Supported via MEC013 Location API (tracks UE movement) and MEC040 (supports MEC Federation and cross-MEP orchestration) | Supported: oneM2M handles session handover and task migration to a new edge (MN-CSE) instance coordinated by IN-CSE |
### 6.4.5 Smart Warehouse Automation
#### 6.4.5.1 Use case Driven Deployment
The architecture integrates a centralized oneM2M IN-CSE, responsible for managing infrastructure-level data (sensors, tags, operational policies), and multiple MN-CSEs deployed at the MEC edge, enabling real-time, latency-sensitive decision-making close to the action. For time-critical events such as rerouting AGVs, responding to hazards, or optimizing resource allocation, processing tasks are offloaded to MEC-hosted MN-CSEs. These instances provide ultra-low-latency data processing
In Table 6.4.5.1-1, we consider all the relevant operational requirement for this use case.
##### Table 6.4.5.1-1 – Operational Requirements and Platform Support for Smart Warehouse Automation
| Operational Requirement | Support in MEC | Support in oneM2M |
| -------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------- |
| Warehouse IoT platform deployment | Not directly required | Supported: oneM2M IN-CSE can be deployed at cloud or control center |
| Registration of sensors, AGVs, and applications | Not required | Supported: oneM2M features provide CSFs to register both devices and AEs (Application Entities) |
| Collection of environmental and telemetry data | Not required | Supported via oneM2M’s Common Service Layer and Mca interfaces |
| Instantiate of MN-CSE on Edge Node | MEC provides the infrastructure to host oneM2M as a service producing MEC application (using Mp1 interface or as a new MEC service inside the MEC Platform. MEC IoT API (MEC033) enables registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiate of AE (e.g., AGV controller, sensor manager) on Edge Node | AE can be instantiated as MEC Application on MEC Host | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
| Low-latency control for AGVs and asset tracking | Supported: MEC apps handle time-sensitive operations with minimal delay | Supported: MN-CSE enables fast coordination through subscriptions, notifications, and data routing |
| Offloading low latency tasks of IoT Platform for real time data processing | Not Required | Supported: Tasks can be offloaded to MN-CSE instances using Mcc interface mechanisms. |
| Service continuity during AGV zone transitions | Supported via MEC013 Location API (tracks UE movement) and MEC040 (supports MEC Federation and cross-MEP orchestration) | Supported: oneM2M handles session handover and task migration to a new edge (MN-CSE) instance coordinated by IN-CSE |
### 6.4.6 Industrial Digital Twins
#### 6.4.6.1 Use case Driven Deployment
The “Industrial Digital Twin” use case requires tight integration between oneM2M and MEC to support low-latency analytics, real-time control, and dynamic migration of digital twin services across production zones. The IN-CSE is deployed in the cloud to manage global digital twin state and analytics, while MN-CSEs are instantiated on MEC nodes located near industrial assets.
MEC hosts edge applications (e.g., quality inspection, anomaly detection) that interact with MN-CSEs. As mobile assets like AGVs move, the MEC, MN-CSE and IN-CSE coordinate handovers and synchronize twin states across MN-CSEs, ensuring seamless operation and minimal downtime.
In Table 6.4.6.1-1, we will consider all the relevant operational requirement for this use case.
##### Table 6.4.6.1-1 – Operational Requirements and Platform Support for Industrial Digital Twins
| Operational Requirement | Support in MEC | Support in oneM2M |
| ----------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| IoT platform deployment | Not directly required | Supported: oneM2M IN-CSE can be deployed at cloud or control center |
| Registration of Industrial devices (Digital Twin (DT)) | Not required | Supported: oneM2M enables registration of industrial sensors, actuators, controllers as either IN-AEs or ADN-AE via standardized resource structures and CSFs using \<flexContainer> data |
| Registration of Industrial applications | Not required | Supported: oneM2M enables registration of industrial |
| Applications as AEs via standardized resource structures and CSFs | | |
| Real-time data ingestion from factory floor (sensors, robots, AGVs) | Not required | Supported via oneM2M’s CSF and Mca interfaces |
| Instantiate of MN-CSE on Edge Node | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiate of AE (e.g., AGV controller, sensor manager, analytics & controller) on Edge Node (Edge Digital Twin) | AE can be instantiated as MEC Application on MEC Host | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
| Low-latency control for AGVs and asset tracking | Supported: MEC Apps handle time-sensitive operations with minimal delay | Supported: MN-CSE enables fast coordination through subscriptions, notifications, and data routing |
| Offloading low latency tasks of IoT Platform for real time data processing | Not required. But MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: Tasks can be offloaded to MN-CSE instances using Mcc interface mechanisms. |
| Service continuity during AGV zone transitions | Supported via MEC013 Location API (tracks UE movement) and MEC040 (supports MEC Federation and cross-MEP orchestration) | Supported: oneM2M handles session handover and task migration to a new edge (MN-CSE) instance coordinated by IN-CSE |
| Synchronization of Digital Twin state between edge and cloud | Not required | oneM2M ensures data consistency and context synchronization between MN-CSE (edge) and IN-CSE (cloud) whereas DT state plays vital role |
### 6.4.7 Assisted Manoeuvring for Autonomous Ships
#### 6.4.7.1 Use case Driven Deployment
The “Assisted Manoeuvring for Autonomous Ships” use case supports autonomous ship operations through coordinated deployment of oneM2M and MEC components. A cloud-hosted IN-CSE collects and processes global vessel data (e.g., location, speed, sensor feeds), while edge-hosted MN-CSEs (ROC) are deployed near ports to assist with real-time decision-making.
When a ship enters port waters, MEC hosts localized MEC/Edge applications (e.g., assisted maneuvering, collision avoidance) that interact with MN-CSEs for low-latency analytics. As the vessel moves, tasks are seamlessly offloaded across MN-CSEs, with the IN-CSE ensuring synchronization and service continuity.
This deployment enables responsive control, enhanced safety, and uninterrupted autonomous operation of unmanned ships near coastal and harbor zones. In the Table 6.4.7.1-1, we consider all the relevant operational requirement for this use case.
##### Table 6.4.7.1-1 – Operational Requirements and Platform Support for Assisted Manoeuvring for Autonomous Ships
| Operational Requirement | Support in MEC | Support in oneM2M |
| ------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------- |
| IoT platform deployment | Not directly required | Supported: oneM2M IN-CSE can be deployed at cloud |
| Registration of devices on the Ships | Not required | Supported: oneM2M enables registration of sensors, actuators, controllers as either IN-AEs or MN-AE or ADN-AE via standardized resource structures and CSFs |
| Registration of applications | Not required | Supported: oneM2M enables registration of Applications as AEs via standardized resource structures and CSFs |
| Real-time data ingestion from AEs (vessel localization, its speed, course and other environment variables) | Not required | Supported via oneM2M’s CSF and Mca interfaces |
| Instantiate of MN-CSE on Edge Node plays role ROC with AI capabilities | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiate of AE (e.g., generate warnings in real-time and optimize its transmission to the vessel) on Edge Node | AE can be instantiated as MEC Application on MEC Host | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
| Low-latency control for unmanned Vessels (i.e. assisted maneuvering, collision avoidance and situational awareness) | Supported: MEC Apps handle time-sensitive operations with minimal delay | Supported: MN-CSE enables fast coordination through subscriptions, notifications, and data routing |
| Offloading low latency tasks of IoT Platform for real time data processing | Not required. But MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: Tasks can be offloaded to MN-CSE instances using Mcc interface mechanisms. |
| Service continuity during vessels transitions | Supported via MEC013 Location API (tracks UE movement) and MEC040 (supports MEC Federation and cross-MEP orchestration) | Supported: oneM2M handles session handover and task migration to a new edge (MN-CSE) instance coordinated by IN-CSE |
### 6.4.8 Smart Metaverse Shopping with Edge-AI and Cloud-IoT Integration
#### 6.4.8.1 Use case Driven Deployment
The “Smart Metaverse Shopping” use case requires the oneM2M IN-CSE deployment in the cloud to manage IoT devices and to store data from the physical retail environment. MEC hosts an AI-powered application at the network edge, close to the store, enabling real-time interpretation of user interactions within the metaverse.
The MEC App subscribes to live data from the IN-CSE and processes it to deliver immediate, personalized feedback in the virtual shopping environment (e.g., recommendations, stock alerts). Upon purchase, the system updates the IN-CSE and triggers real-world actions such as product reservation or notifications to store staff and actuators.
Optionally, an MN-CSE may be instantiated on the MEC platform to further reduce latency and improve service continuity. This hybrid deployment ensures real-time, synchronized experiences across the digital and physical retail layers.
In the Table 6.4.8.1-1, we consider all the relevant operational requirement for this use case.
##### Table 6.4.8.1-1 – Operational Requirements and Platform Support for Smart Metaverse Shopping
| Operational Requirement | Support in MEC | Support in oneM2M |
| ---------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| IoT platform deployment | Not directly | Supported: oneM2M IN-CSE can be deployed at cloud |
| Registration of IoT devices deployed in the physical store | Not required | Supported: oneM2M enables registration of industrial sensors, actuators, controllers as either IN-AEs or ADN-AE via standardized resource structures and CSFs |
| Registration of IoT applications | Not required | Supported: oneM2M enables registration of industrial Applications as AEs via standardized resource structures and CSFs |
| Real-time data ingestion from Stores (shelf inventory levels, product locations, environmental conditions) | Not required | Supported via oneM2M’s CSF and Mca interfaces |
| Instantiate of MN-CSE on Edge Node with AI capabilities (If required) | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiate of AE (e.g., sensor manager, analytics & controller) on Edge Node | AE can be instantiated as MEC Application on MEC Host. It can use AI models for analysing real time data and providing feedback. Virtual Store App instantiate as an MEC application/Edge Application on MEC Host. | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
| Low-latency control for real time data processing | Supported: MEC Apps handle time-sensitive operations with minimal delay | Supported: MN-CSE enables fast coordination through subscriptions, notifications, and data routing |
### 6.4.9 Future Homes
#### 6.4.9.1 Use case Driven Deployment
The “Future Homes” use case delivers immersive and personalized smart home services through ETSI MEC and oneM2M integration. A centralized CSE (IN-CSE or MN-CSE) runs on a Cloud or MEC node to manage devices across homes, while local CSEs (MN-CSE/ASN-CSE) are deployed on non-MEC nodes such as home gateways or CPEs.
Latency-critical tasks like video analytics and sensor fusion are offloaded to Customer Premises Edge devices hosting MEC apps or CSEs, with MEC platforms leveraging the MEP interface for discovery and task delegation.
In the Table 6.4.9.1-1, we consider all the relevant operational requirement for this use case.
##### Table 6.4.9.1-1 – Operational Requirements and Platform Support for Future Homes
| Operational Requirement | Support in MEC | Support in oneM2M |
| ---------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------- |
| IoT platform deployment | Not directly | Supported: oneM2M IN-CSE can be deployed at cloud or as MN-CSE elsewhere |
| Registration of devices at the home premises (lights, thermostats, cameras, wearables) | Not required | Supported: oneM2M enables registration of sensors, actuators, controllers as either IN-AEs or MN-AE or ADN-AE via standardized resource structures and CSFs |
| Registration of applications | Not required | Supported: oneM2M enables registration of Applications as AEs via standardized resource structures and CSFs |
| Real-time data ingestion from AEs (location data, Health alerts or anomalies, room temperature, energy consumption, object movements) | Not required | Supported via oneM2M’s CSF and Mca interfaces |
| Instantiate of MN-CSE/ASN-CSE on Edge Node (Customer Premises Edge) | MEC provides the infrastructure to host oneM2M as a service producing MEC application using Mp1 interface. Can provide support to integrate a new MEC IoT service (MEC 0xx to be defined later) inside the MEC Platform which should be more coupled with oneM2M standards. Whereas MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms. Not Supported in MEC: If MEC supports CPEs, then MEC systems can discover them, and MEP enables MEC applications to offload real time data processing tasks to this non-MEC systems | Supported: oneM2M platform needs to include a CSF to integrate with the MEC platform |
| Instantiate of AE (e.g., Energy Optimization, Real-time Safety Alerts, Contextual Scene Adaptation, Manage home appliances) on Edge Node | AE can be instantiated as MEC Application on MEC Host or as Edge application on CPEs | Supported: oneM2M platform needs to include a CSF to instantiate an AE as MEC application |
| Offloading low latency tasks of IoT Platform for real time data processing | Not required. But MEC IoT API (MEC033) enables minimal registration and discovery of IoT platforms | Supported: Tasks can be offloaded to MN-CSE instances using Mcc interface mechanisms. |
| Service continuity during devices movements. | Supported via MEC013 Location API (tracks UE movement), MEC021 (Application Mobility service) and MEC040 (supports MEC Federation and cross-MEP orchestration) | Supported: oneM2M handles session handover and task migration to a new edge (MN-CSE/ASN-CSE) instance coordinated by IN-CSE/MN-CSE |
# 7 Proposed Recommendations for Federation and Orchestration Mechanisms
## 7.1 Introduction
The main focus of integration between **ETSI MEC** and **oneM2M** is to enable efficient data processing, low-latency communication, and enhanced service delivery in edge computing environments. The major capability of this integration includes **federation of the ETSI MEC platforms** and **interaction with oneM2M instances**, allowing for distributed computing and data management across multiple edge nodes. Both ETSI MEC and oneM2M provide some basic APIs and protocols in order to facilitate such a federation. However, orchestration mechanisms are still required to ensure seamless operation across multiple MEC/oneM2M instances. This clause describes a preliminary set of functional and technical recommendations for enabling a framework for the federation and orchestration of MEC/oneM2M instances. These recommended processes, which not only complement many use cases in clause 5 but also unlock additional value by introducing capabilities that extend and enhance their intended scope, including:
- **Handover**: refers to the process of transferring control of a device or service from MEC/oneM2M instance to another, ensuring seamless connectivity and service continuity.
- **Swarm Computing**: refers to the coordination of multiple MEC/oneM2M instances to perform distributed computing tasks, leveraging the capabilities of edge devices and networks.
- **Federated Learning**: refers to a machine learning approach where multiple MEC/oneM2M instances collaboratively train and share models across multiple decentralized nodes without sharing row data, enhancing privacy and reducing latency.
The federation and orchestration framework will explore the existing APIs for discovery and resource management principles defined in both standards, while proposing extensions to them with federation and synchronization capabilities. Before orchestrating handover, swarm computing, or federated learning mechanisms, MEC/oneM2M instances must be able to discover, register, and organize themselves into collaborative groups:
- **Instances Registration**: MEC/oneM2M nodes (e.g., IN-CSE, MN-CSE, MEC Platform, MEC Host, etc.) should be able to register to a federation registry, that may include its identity, capabilities, and available services.
- **Instances Discovery**: MEC/oneM2M nodes **should** be able to advertise its capabilities through standard discovery mechanisms (e.g., oneM2M Discovery resource, MEC location, etc.). **This might be achieved by publishing metadata about the MEC/oneM2M node’s processing power, storage capabilities, network connectivity, and supported APIs. For example, using MEC IoT API could allow MEC applications to discover oneM2M platforms and route device traffic accordingly.**
- **Instances Federation**: once discovered, MEC/oneM2M instances can be logically grouped into a federation group. Such a group might be a collection of MEC/oneM2M instances that are allowed to intercommunicate and cooperate in executing distributed tasks. Federation groups could be defined based on geographic proximity, application domain or QoS requirements.
- **Roles Assignment**: based on their capabilities, the MEC/oneM2M nodes could be selected for specific roles such as handover anchor, swarm computing worker or federated learning participant. For the instance, MEC/oneM2M nodes with high processing power and storage capacity could be selected for swarm computing and federated learning training tasks, while nodes with good connectivity could be used for handover, service continuity and low-latency communications.
Therefore, the orchestration framework should:
- Implement resource replication mechanisms to mirror or move device/application registration between MN-CSE instances **or IN-CSE or across MEC hosts including the reference criteria for doing that**.
- Ensure a handover-safe mechanism for migrating subscriptions (first establish a new subscription and then terminate the old one).
- Establish synchronization mechanisms between MN-CSE and IN-CSE to ensure that the recent device states are propagated upward and historical data are migrated asynchronous, when necessary.
- **Identify and discover MEC/oneM2M instances from logical federation group.**
- Match available resources (e.g., storage, compute, connectivity, etc.) with service requirements (e.g., handover, swarm computing or federated learning).
- Assign tasks to federation nodes.
- Keep registrations, subscriptions, and data synchronized across MN-CSEs/IN-CSEs and MEC nodes.
- Continuously monitor nodes performances and workload by reassigning tasks or triggering handover when needed.
The orchestration framework **should** create a **federated, service-aware, and resource-optimized environment** where ETSI MEC and oneM2M instances jointly deliver advanced distributed computing and IoT data management, by addressing handover continuity, swarm scalability, and federated learning privacy.
## 7.2 Handover
The Handover mechanism refers to the process of transferring control of a device or service from one MEC/oneM2M instance to another, ensuring seamless connectivity and service continuity. This process illustrated in Figure 7.2.1 is critical when devices move between different edge nodes or when services must be migrated to maintain low latency, reliability, and high availability.
![Fig 7.2.1](media/handover.png)
**Fig 7.2.1 IoT device handover**
Depending on system design and device capability, handover control can follow one of two approaches:
- Device-controlled handover: The device itself detects conditions that trigger a handover (e.g., deteriorating radio quality, mobility, policy changes). It initiates discovery, re-registration, and session updates with the target MEC/oneM2M instance. This approach offers autonomy but assumes the device has sufficient resources to evaluate conditions, make decisions, and handle signaling.
- Network-controlled handover: The MEC/oneM2M infrastructure (e.g., MN-CSEs or gateways) monitors device connectivity and system load, then initiates or enforces a handover. The device simply follows the instructions (e.g., re-register to a new MN-CSE) without needing complex logic or extensive signaling capability. This approach could be suitable for constrained IoT devices with limited processing or decision-making capacity.
In practice, the chosen strategy should balance device capability, network intelligence, and the requirements of the application domain. For resource-constrained IoT devices, offloading decision-making to the MEC/oneM2M layer may reduce device complexity and power usage, while device-controlled handovers may provide faster reaction times and greater autonomy in heterogeneous deployments.
To design an effective handover mechanism, the following high-level considerations should be addressed:
- **Identity & Security**: The device must maintain a consistent identity across MEC/oneM2M instances, with fast and secure re-authentication (e.g., token-based methods). Security credentials and keys should be managed automatically to ensure a smooth transition.
- **Session & State Continuity**: Handover must preserve session continuity and data consistency. This includes mechanisms for state replication (buffers, acknowledgments, sequence numbers), subscription reinstatement, downlink catch-up, and, where possible, context transfer of QoS or profiles between gateways.
- **Performance & Latency**: Interruptions must be minimized with well-defined handover budgets (e.g., ≤200 ms for critical telemetry). Local buffering, ordering policies, duplicate suppression, and time synchronization help maintain low latency and reliable delivery.
- **Network Awareness & Adaptation**: The mechanism should leverage network conditions such as bandwidth, signal strength, and congestion when selecting target gateways. Neighbor awareness, proactive discovery, and policy-based triggers (RSSI/SNR thresholds, cost/latency trade-offs) are key to ensuring optimal handover decisions. Admission control policies must also be considered, allowing gateways to accept or defer connections gracefully.
- **Monitoring & Reliability**: System reliability depends on the ability to measure and manage handovers. KPIs such as attach/authentication time, packet loss, and latency deltas should be tracked. Tracing, alarms for frequent or failed handovers, and robust retry/backoff logic ensure stability and resilience.
- **Device Constraints & Efficiency**: IoT devices, often resource- or power-constrained, require lightweight solutions. Gateway-local registrations can reduce complexity, while power-aware scanning and cached neighbor reports minimize RF cost. Feature negotiation and versioned message contracts further improve compatibility and efficiency across heterogeneous deployments.
These considerations will be further analyzed in oneM2M and ETSI MEC.
## 7.3 Swarm Computing
**Swarm Computing** refers to the coordination of multiple MEC/oneM2M instances to perform distributed computing tasks, leveraging the capabilities of edge devices and networks. In this paradigm, individual MEC/oneM2M nodes act like members of a swarm, each contributing processing power, storage, connectivity, or sensing capabilities to achieve collective goal. The system operates in a decentralized and adaptive manner, where tasks can be dynamically partitioned, distributed, and recombined across nodes depending on resource availability, network conditions, and application requirements. This enables resilient, scalable, and low-latency processing, as task are executed closer to the data sources and devices while ensuring cooperative load balancing, fault tolerance, and energy efficiency.
When mapped onto MEC/oneM2M interworking, distributed instances could be implemented as Application Entities (AEs) hosted on Middle Nodes (MNs) or Application Dedicated Nodes (ADNs). In this scenario, local coordination within a swarm could be supported by nearby Common Service Entities (CSEs). In ETSI MEC terms, these can be MEC Services or MEC Applications which provide contextual information aggregated from other swarm nodes.
In this clause, the involved entities are defined as follows:
- **Swarm Agent (Local/Edge)**: is a software entity running on a device, MEC host, or oneM2M Application Entity that performs local computation and participates in cooperative swarm behavior. It executes subtasks of a global distributed task, shares local state, sensor data, or intermediate results with peers. The local Swarm Agent may run on constrained devices or ADNs, focusing on lightweight processing and sensing, while the Edge Swarm Agent may run on MEC hosts or MNs, handling more complex subtasks and acting as a bridge to cloud/federated nodes.
- **Swarm Node:** is any participating entity (e.g., device, MEC host, AE, MN, or ADN) that contributes computing, storage, or sensing resources to the swarm. It executes assigned subtasks and exchanges state updates with other nodes via oneM2M group communication or MEC APIs.
- **Swarm Collector Node:** is a designated node (e.g., MN-CSE, MEC host, or IN-CSE) that aggregates results from multiple swarm nodes and produces a unified output. It collects partial results from distributed nodes, performs aggregation (e.g., fusion of sensor data), and provides the final output back to the orchestrator, applications, or end-users.
- **Swarm Orchestrator:** is the central coordination entity of the swarm that manages task distribution, resource allocation, synchronization, resilience, and policy enforcement. It splits global tasks into subtasks, assigns subtasks to swarm nodes based on capacity, connectivity, and energy, detects node failures and reassigns tasks, and ensures QoS, latency, energy, and fault-tolerance goals. It may operate centrally (e.g., cloud IN-CSE) or decentrally (e.g., at MEC/MN-CSE) depending on deployment.
Compatibly with the deployment options described in Clause 6.4.1, the swarm computing could be implemented considering the following options:
- **Option1:** CSE at the edge and MEC host close to the devices so that swarm computing agents interact directly via local MN-CSEs/MEC Applications with low-latency (see Figure 7.3.1)
![Fig 7.3.1](media/swarm_computing_option1.png)
**Fig 7.3.1 Swarm Computing Implementation - Option 1**
- **Option 2**: Part of swarm intelligence runs on constrained devices, while heavier computation is offloaded to MEC nodes for aggregation and route optimization (see Figure 7.3.2).
![Fig 7.3.2](media/swarm_computing_option2.png)
**Fig 7.3.2 Swarm Computing Implementation - Option 2**
- **Option 3**: high-level pattern recognition in an IN-CSE hosted in cloud, while real-time behaviors remain at the edge (see Figure 7.3.3)
![Fig 7.3.3](media/swarm_computing_option3.png)
**Fig 7.3.3 Swarm Computing Implementation - Option 3**
In order to allow multiple MEC/oneM2M nodes to collaboratively perform distributed tasks, the orchestration mechanisms should take into account the following aspects:
- **Task Decomposition**: a global computation task is divided into smaller subtasks. The orchestrator distributes these subtasks to participating MEC/oneM2M nodes based on their declared processing and storage capacity, network connectivity, or CPU/GPU availability.
- **Synchronization**: appropriate synchronization mechanisms (e.g., publish/subscribe over oneM2M and MEC APIs) should ensure swarm nodes operate on consistent views of data. To maintain consistent swarm knowledge across federated nodes, a distributed state synchronization is required.
- **Task offloading**: the orchestration mechanism should support task offloading policies where swarm members dynamically delegate high-complexity processing to nearby MEC/oneM2M nodes.
- **Resilience**: in case of node failures, the orchestrator dynamically reallocates unfinished subtasks to other swarm nodes.
- **Coordination and Aggregation**: results from swarm nodes are aggregated at a designed MEC host, IN-CSE or MN-CSE which acts as a collector node. The oneM2M’s group communication primitives could be used to coordinate actions across swarm members, while MEC APIs could be used to optimize routing and QoS for tasks exchanges.
The swarm computing orchestrator is the coordinator, load balancer, and fault manager of swarm computing. It does not always execute tasks itself, instead it manages how, where, and when the tasks are executed, ensuring that the swarm MEC/oneM2M nodes behave like a single resilient and adaptive system as reported in the sequence diagram of Figure 7.3.4.
![Fig 7.3.4](media/diagram_swarm_computing_orchestration.png)
**Fig 7.3.4 Swarm Computing - Orchestration Lifecycle**
## 7.4 Federated Learning
Federated Learning is a distributed machine learning approach in which model training is performed locally on multiple nodes and only model’s parameters or weights, not row data, are exchanged with aggregators. Instead of sending sensitive data to a central location, each node trains a local model using its own data and then shares only the model updates with a central location, which aggregates these updates to improve the global model. This enhances privacy, reduces bandwidth usage, and allows model adaptation close to data sources and IoT devices.
In the MEC/oneM2M interworking context, local training can be executed within Application Entities (AEs) on Middle Nodes (MN-CSE) or even on constrained edge devices, while aggregation roles can be hosted on Infrastructure Nodes CSEs (IN-CSE) in the cloud environment, or on ETSI MEC applications acting as intermediate servers at the edge. ETSI MEC services could provide capabilities such as Location API, Radio Network Information API, or IoT API to enrich local training with contextual features.
In this clause, the involved entities are defined as follows:
- **FL Client:** a participating device or node (e.g., IoT sensor, MEC node or application entity) that trains a machine learning model locally using its own private data. It holds raw data (never shared outside the device), performs local model training based on the global model received from the server/aggregator, and sends only model updates/gradients (not raw data) back to the FL Server or Aggregator.
- **FL Server:** coordinates the training process across all clients by distributing the initial global model, collecting updates, and managing the learning rounds. It selects which clients participate in each training round, sends the current global model parameters to selected clients, receives updated parameters or gradients from clients after local training, and passes these updates to the FL Aggregator for combination.
- **FL Aggregator:** is responsible for combining model updates from multiple clients into a new, improved global model. It performs model aggregation (e.g., weighted averaging of client models), handles data heterogeneity by balancing contributions from clients (e.g., clients with more data may weigh more), and ensures robustness against noisy updates, stragglers, or malicious clients (e.g., secure aggregation techniques).
- **FL Orchestrator:** manages the overall lifecycle and policies of the federated learning process, ensuring scalability and resilience across distributed clients and servers. It decides which clients participate in each round (e.g., based on resource availability, network conditions, or data diversity), optimizes how training workloads are distributed, especially in heterogeneous edge/cloud environments, handles privacy rules, security requirements, energy constraints, and fault tolerance, and orchestrates interactions across multiple FL Servers/Aggregators.
Compatibly with the deployment options described in Clause 6.4.1, the federated learning could be implemented considering the following options:
- **Option 1**: federated learning clients run entirely on MN-CSEs co-located with MEC hosts, while the aggregation process occurs within a (edge) MEC application (see Figure 7.4.1).
![Fig 7.4.1](media/federated_learning_option1.png)
**Fig 7.4.1 Federated Learning Orchestration - Option 1**
- **Option 2**: local MN-CSE/MEC nodes act as intermediate aggregators before sending updates to a central (in-cloud) IN-CSE federated learning server (see Figure 7.4.2).
![Fig 7.4.2](media/federated_learning_option2.png)
**Fig 7.4.2 Federated Learning Orchestration - Option 2**
- **Option 3**: most training happens at the edge, but global model updates and advanced analytics are performed in the cloud IN-CSE (see Figure 7.4.3).
![Fig 7.4.3](media/federated_learning_option3.png)
**Fig 7.4.3 Federated Learning Orchestration - Option 3**
Therefore, the orchestration mechanism should be able to perform the following steps:
- **Federation Group Selection**: based on the federation registry, MEC/oneM2M nodes or instances with sufficient compute and local training capabilities are selected as participants to the federated learning process.
- **Model Distribution**: a global training model initialized at IN-CSE or MEC Orchestrator is distributed among selected participants.
- **Local Training**: each selected node trains the model locally using its own data. The training parameters (e.g., weights, gradients, etc.) are stored locally and not exposed outside the node.
- **Aggregation and Synchronization**: model’s updates are periodically transmitted to the aggregation node which might be either an IN-CSE or a MEC host. The aggregator combines updates using secure aggregation protocols and redistributes the improved global model. This allows to maintain consistency between IN-CSE/MEC host, MEC/oneM2M nodes and MEC orchestrator (aggregator).
The federated learning orchestrator ensures nodes with low computational capabilities or high latency are assigned smaller training loads, while the MEC Location APIs help prioritize nodes with better connectivity to reduce synchronization delays as schematically illustrated in Figure 7.4.4.
![Fig 7.4.4](media/diagram_federated_learning_orchestration.png)
**Fig 7.4.4 Federated Learning - Orchestration Lifecycle**
# Annex A: <br>Title of annex
<br />
# Annex B: <br>Title of annex
# Annex: Bibliography
## B.1 First clause of the annex
## B.1.1 First subdivided clause of the annex
<br />
# Annex: Bibliography
# Annex : Change history
<br />
# Annex : Change history
| Date | Version | Information about changes |
|------------|---------|---------------------------|
| --------------- | ------- | ----------------------------------------- |
| &lt;Month year> | <#> | &lt;Changes made are listed in this cell> |
| | | |
| | | |
......@@ -239,17 +957,13 @@ For the purposes of the present document, the [following] abbreviations [given i
<br />
# History
+-------------------------------------------------+
| Document History |
+:==============+:==============+:================+
# Document history
| Version | Date | Milestone |
+---------------+---------------+-----------------+
|<Month year> | <#> |<Changes made> |
+---------------+---------------+-----------------+
| | | |
+---------------+---------------+-----------------+
| | | |
+---------------+---------------+-----------------+
| ------- | ----------- | ---------------------- |
| V0.0.1 | April 2025 | Early Draft available |
| V4.0.2 | June 2025 | Early Draft Update |
| V4.1.0 | July 2025 | Stable Draft available |
| V4.1.1 | August 2025 | Stable Draft Update |
media/deployment_architecture.excalidraw 0 → 100644
View file @ b5fb1b5d
This diff is collapsed.
media/diagram_federated_learning_orchestration.puml 0 → 100644
View file @ b5fb1b5d
@startuml
hide footbox
participant Orchestrator
skinparam participant {
BackgroundColor #ADD8E6
}
participant Registry as "Federation Registry"
participant Orchestrator as "FL Orchestrator (MEC/IN-CSE)"
participant Node as "MEC/oneM2M Nodes (FL Clients)"
participant Aggregator as "MEC Host/IN-CSE (FL Aggregator)"
Orchestrator -> Registry : Query available nodes & capabilities
Registry --> Orchestrator : Return eligible nodes
Orchestrator -> Node : Select participants\n(Federation Group Selection)
Orchestrator -> Node : Distribute initial global model
loop Local Training
Node -> Node : Train local model on private data
Node -> Aggregator : Send model updates (weights/gradients)
end
Aggregator -> Aggregator : Aggregate model updates securely
Aggregator -> Orchestrator : Send aggregated global model
Orchestrator -> Node : Redistribute improved model
alt Node with limited resources
Orchestrator -> Node : Assign smaller training load
end
alt High connectivity nodes
Orchestrator -> Node : Prioritize for synchronization
end
@enduml
media/diagram_swarm_computing_orchestration.puml 0 → 100644
View file @ b5fb1b5d
@startuml
hide footbox
skinparam participant {
BackgroundColor white
}
participant Orchestrator
skinparam participant {
BackgroundColor #ADD8E6
}
participant "MEC Host/MN-CSE (Swarm Node A)" as NodeA
participant "MEC Host/MN-CSE (Swarm Node B)" as NodeB
participant "MEC Host/MN-CSE (Swarm Node C)" as NodeC
participant "MEC Node" as MEC
participant "MN-CSE/MEC Host/IN-CSE (Collector Node)" as Collector
participant "MEC/oneM2M APIs" as Bus
== Task Decomposition ==
Orchestrator -> Orchestrator : Divide Global Task
Orchestrator -> NodeA : Assign Subtask A
Orchestrator -> NodeB : Assign Subtask B
== Synchronization ==
NodeA -> Bus : Publish State Update
Bus -> NodeB : Deliver Update
Bus -> NodeC : Deliver Update
NodeB -> Bus : Acknowledge Sync
NodeC -> Bus : Acknowledge Sync
== Task Offloading ==
NodeB -> Orchestrator : Request Offloading (Heavy Task)
Orchestrator -> MEC : Forward Heavy Task
MEC -> Orchestrator : Processed Result
Orchestrator -> NodeB : Return Output
== Resilience ==
Orchestrator -> NodeA : Monitor Progress
NodeA -> Orchestrator : Failure Notification
Orchestrator -> NodeC : Reassign Subtask
NodeC -> Orchestrator : Processed Result
== Coordination & Aggregation ==
NodeA -> Collector : Partial Result
NodeB -> Collector : Partial Result
NodeC -> Collector : Partial Result
Collector -> Orchestrator : Aggregated Results
@enduml
\ No newline at end of file
media/federated_learning_option1.puml 0 → 100644
View file @ b5fb1b5d
@startuml
hide footbox
participant Orchestrator
skinparam participant {
BackgroundColor #ADD8E6
}
participant Orchestrator as "FL Orchestrator"
participant Node as "MN-CSE (FL Clients)"
participant Aggregator as "MEC Application (FL Aggregator)"
Orchestrator -> Node : Distribute initial global model
loop Local Training
Node -> Node : Train model locally on private data
Node -> Aggregator : Send model updates
end
Aggregator -> Aggregator : Aggregate updates securely
Aggregator -> Orchestrator : Return aggregated global model
Orchestrator -> Node : Redistribute improved model
@enduml
\ No newline at end of file
media/federated_learning_option2.puml 0 → 100644
View file @ b5fb1b5d
@startuml
hide footbox
participant Orchestrator
skinparam participant {
BackgroundColor #ADD8E6
}
participant Orchestrator as "FL Orchestrator (Cloud IN-CSE)"
participant Node as "MN-CSE Nodes (FL Clients)"
participant EdgeAgg as "MEC Host (FL Aggregator)"
participant CloudAgg as "IN-CSE (FL Aggregator-Global)"
Orchestrator -> Node : Distribute initial global model
loop Local Training
Node -> Node : Train model locally on private data
Node -> EdgeAgg : Send model updates
end
EdgeAgg -> EdgeAgg : Aggregate updates locally
EdgeAgg -> CloudAgg : Forward aggregated updates
CloudAgg -> CloudAgg : Aggregate global model
CloudAgg -> Orchestrator : Send updated global model
Orchestrator -> Node : Redistribute improved model
@enduml
media/federated_learning_option3.puml 0 → 100644
View file @ b5fb1b5d
@startuml
participant Orchestrator
skinparam participant {
BackgroundColor #ADD8E6
}
hide footbox
participant Orchestrator as "FL Orchestrator (Cloud IN-CSE)"
participant Node as "MN-CSE Nodes (FL Clients)"
participant CloudAgg as "IN-CSE (FL Aggregator-Global)"
Orchestrator -> Node : Distribute initial global model
loop Local Training
Node -> Node : Train model locally on private data
Node -> CloudAgg : Send model updates
end
CloudAgg -> CloudAgg : Aggregate updates globally
CloudAgg -> Orchestrator : Return improved global model
Orchestrator -> Node : Redistribute updated model
@enduml
\ No newline at end of file
media/handover.puml 0 → 100644
View file @ b5fb1b5d
@startuml
hide footbox
' More examples https://www.planttext.com/
'Here is a gallery of the available themes: : https://the-lum.github.io/puml-themes-gallery/
' use 'plain' theme for oneM2M standards submissions and
skinparam BackgroundColor transparent
!theme plain
' Call Flow Starts Here
actor "device" as AE
participant "IN-CSE" as IN
participant "MN-CSE/MEC1" as MN1 #lightblue
participant "MN-CSE/MEC2" as MN2 #lightblue
participant "MN-CSE/MEC3" as MN3 #lightblue
'autonumber 1
== device handover ==
' IPE registers and creates the Washing Machine Model
AE -> MN1 : device sends 'data'
group Monitor system metrics
note over AE,MN3 : monitor network conditions
note over AE,MN3 : discover neighbors
note over AE,MN3 : monitor neighbor conditions
end group
group Coordinate handover
note over AE,MN3 : Prepare / trigger handover
note over AE,MN3 : device moves to MN2
note over AE,MN3 : State / context transfer
note over AE,MN3 : establish QOS
end group
AE -> MN2 : device sends 'data'
group Monitor system metrics
note over AE,MN3 : continue to monitor system metrics
end group
group Coordinate handover
note over AE,MN3 : handover to MN3
end group
AE -> MN3 : device sends 'data'
@enduml
\ No newline at end of file