“In addition to improved marketing of oneM2M, TP50 activities explored new requirements linking AI with IoT and, testing for interoperability”

 

In this interview, Roland Hechwartner of Deutsche Telekom (DT) and Chairman for oneM2M’s Technical Plenary (TP) describes the latest developments and lessons learned from oneM2M’s recent TP50. It comprised sessions of the technical plenary and all working groups, over the period from May 10 to June 3, 2021.

Q: What were the major developments to come out of this TP?

RH: Among the various activities during this TP, there are four topics I would like to cover. I will talk about how our Marketing Committee (MARCOMs) is raising the profile of oneM2M in the wider market and then report on the election of new representatives to leadership positions within oneM2M. I will then talk briefly about progress on oneM2M standardization roadmap and Release 4 and 5 developments. Finally, I would like to highlight a couple of new work items that reflect issues in the IoT market. On the one hand, there is a growing need to solve challenges associated with the integration of AI with IoT systems. There is also a need to ensure interoperability across different IoT applications and systems, with implications on testing regimes.

Q: Let us begin with the latest news on increasing the visibility of oneM2M in the IoT market. What is happening on that front?

RH: Over the past few months, there has been a lot of work involving participants in oneM2M’s MARCOMs committee to update the content and back-end systems for the oneM2M web site. The new website provides an improved overview of oneM2M news, specifications, guidelines, and events as well as information about deployments.

There is a lot of new content on the website for organizations interested in deploying oneM2M systems. We want to speak to business executives and product managers, so we have added an overview of the rationale for using oneM2M. There are separate sections for organizations that wish to build systems using the oneM2M specifications and for service providers that may simply wish to build solutions using (pre-built) components that conform to oneM2M standards.

In addition to the updated web site, our MARCOMs team is sharing more information about our activities and oneM2M standards via our LinkedIn and Twitter channels. Altogether, we are seeing greater interest in information about oneM2M. I encourage readers to join the oneM2M community by subscribing to one or more of these channels.

Q: And what is the status of new appointments to leadership positions in oneM2M?

RH: I would like to remind readers that oneM2M is a member organization that is open to all organizations that wish to participate. We operate an open process for electing representatives to leadership positions. This is one of several ways that oneM2M differs from sponsorship and fee-based models in many industry alliances. Our members nominate and vote for candidates who go on to serve 2-year terms. For this election cycle, Shane He from Nokia, was re-elected as Chair of the Requirements and Domain Model working group (WG), and Peter Niblett from IBM was elected as Chair of the System Design and Security WG. These are important roles in terms of coordinating member contributions, ensuring progress and adhering to oneM2M’s internal standards for quality.

Q: We have spoken on several occasions about oneM2M’s release cycle so what are the latest updates from this TP?

RH: The virtual format for TPs has affected the way we work, especially as we no longer experience the week-long TPs where members are fully immersed in WG activities. That factor, combined with a high quality of review feedback has affected our schedule. On Release 4, members agreed to allow one more meeting cycle in order to conclude all open change requests and finalize the technical specification work on the architectural and protocol aspects in particular. To achieve the new target, seven virtual meetings of the System Design and Security (SDS) working group are scheduled before the next virtual TP which is due to take place between August 30 and September 17. Consequently, the publication of oneM2M Release 4 is postponed to Q4 2021.

During this TP, we commenced discussions on the scope and timeline Release 5. Its scope will comprise functionalities like advanced semantic discovery, system enhancements to support data protection regulations, interworking with the Sensor Thing API, effective IoT communication to protect 3GPP networks. There is also some work on system enhancements to support Data License Management which will enable smart city and many other use cases. The freeze milestone for stage 1, which means the deadline for accepting new normative requirements for Release 5, has been set as Q1 2022.

While these are extended items of work, I would like to report progress on two deliverables that were ratified during TP50. One of these is the technical report (TR), TR-0046, which is a “Study on Public Warning Service Enablers”. The second is the corresponding technical specification (TS), TS-0037 on IoT Public Warning Service Enablement. Both were completed under the rapporteur-ship of TaeHyun KIM of SyncTechno Inc. in Korea.

Q: And finally, what new work items were agreed at TP50?

RH: The first work item deals with the convergence of Artificial Intelligence (AI) and Machine Learning (ML) with IoT. This is developing into an important topic as the business world adapts to the adoption of IoT technologies and the potential to capture a tremendous amount of data from multiple sources. There is a new challenge, however, in relation to the variety of data coming from countless numbers of IoT devices which makes it complex to collect, process, and analyse the data using AI/ML sub-systems.

Combining these two streams – IoT and AI/ML - benefits the common person and specialists alike. While IoT deals with devices using the internet, AI/ML makes the devices learn from their data and experience. So oneM2M needs to investigate what features are needed to support AI/ML capabilities in the oneM2M architecture and service layers.

These are some of the factors that led to the approval of a new work item (WI), targeting Release 5, on system enhancements to support AI capabilities. Several members supported the new WI including KETI, Hyundai Motors, Exacta GSS, SBS, Nokia, Hansung University, Orange, Convida Wireless and Deutsche Telekom.

The WI will initially focus on a Technical Report (TR) analysing existing AI, which includes ML, technologies that can be resourced into the oneM2M architecture. The TR will also investigate potential AI service use cases that use IoT data. The study on AI technologies and use cases will be further analysed to understand what features are and are not supported by the oneM2M system. Our intention is to analyse unsupported features for the purpose of generating potential requirements for future standardization.

Some of the ideas that came up in discussion include investigations into state-of-the-art AI technologies that use data from IoT systems. Other ideas involve the analysis of potential use cases and requirements to support AI services including different forms of data management, model-training, and model-application functions. The investigation will explore ways to support different parameter schemes relating to power consumption or future-cost priorities, for example. It will also consider support for AI services and distributed or federated ML on Edge/Fog nodes. This will provide insights into service requirements to assist in deploying trained machine learning models on field device nodes to enable distributed inferencing.

Fortunately, the group will be able to draw some prior work and parallel activities in this area. There are, for example, three relevant initiatives in ETSI. These are the specialist task force (STF) projects STF584 (Artificial Intelligence for IoT Systems), STF601 (Cross-domain usability of IoT devices for humans and machines) and work underway in the ETSI Industry Specification Group on Securing AI. The 3GPP standardization body’s Architecture Group, SA2, is also working on the “Study on enablers for network automation for 5G” as part of the 3GPP Release 17. Finally, the South Korean government recently funded the AIStar project which will address IoT data management and interoperability for AI/ML.

The timetable to finalize a Technical Report from this work item is expected to be TP56 (Q3 2022). This should then lead to the development of a technical specification and changes to existing TSs.

Q: The second of the new work items is about interoperability. What are the key issues there?

RH: Yes, this is a work item on interoperability and the new WI focuses on testing to accompany the Release 4 updates. The WI is an enhancement to the TS-0013 Interoperability Testing document and corresponds to the TS-0001 Functional Architecture and TS-0004 Service Layer Core Protocol documents for Release 4.

The outputs of this activity add new Test Descriptions which address the new resources, features and functions specified in the Release 4. The purpose of the interoperability testing is to prove end-to-end functionality between Application Entities and Common Service Entities (CSE) over the Mca and Mcc reference points. The testing is intended to address the access to resources on local CSE and/or remote CSE.

The organizations supporting this WI are Hansung University, Exacta GSS, C-DOT and Deutsche Telekom. Finalization is targeted for the end of 2022, depending on planning of the Interoperability Test Event targeting oneM2M Release4.