In this interview, we hear from Michelle Wetterwald about her activities in the IoT industry. She also discusses a current project to improve the cross-domain usability of IoT devices for humans and machines.
Q1: Would you begin by introducing yourself and your role in this project?
MW: I graduated in engineering and with a doctorate from two French Telecom Schools. I am the Networking and Mobile Systems expert at my own company, Netellany. This is located close to ETSI in the South of France. I work in close partnership with FBConsulting which is based in Luxembourg.
I now have forty years of experience in various positions, both in academia and the ICT industry. My main activities involve the standardization of Cooperative Intelligent Transport Systems (C-ITS) and IoT solutions, as well as emergency communications analysis. I am also a part time lecturer in engineering schools. I have authored or co-authored six patents on early WLAN systems and over 60 papers on advanced wireless networking mechanisms.
One of my current projects involves working as the leader of STF601. STF stands for Specialist Task Force. This standardization project is financed by ETSI. STF601 operates under the steering of the ETSI TC SmartM2M committee.
The objective of this project is to cover a missing key link in the IoT eco-system chain. This deals with the data usability of IoT devices involving humans and machines. The aim is to standardize the usability of data and services that IoT devices and platforms deliver. Usability is normally addressed through ergonomic approaches that enhance the user experience and improve accessibility to ICT equipment. Since other ETSI committees already work on these topics, our focus is on data usability across domains as the project title indicates.
Q2: What issues does the project address and how are these reflected in your scope of work?
MW: The usability of data and services that IoT devices and platforms deliver is an important and under-addressed issue. This is a conclusion from two reports - ETSI TR 103 376 and AIOTI Gaps report Release 2 – that analysed IoT standardization gaps.
In addition to ergonomic and equipment accessibility factors, there is also a need to consider usability from a data and service point of view. In other words, we are concerned about the usability of data and services that IoT devices and platforms deliver.
The STF601 project also complements other artificial-intelligence (AI) activities in ETSI TC SmartM2M. Standardizing the usability of data and services provided by IoT devices and platforms will have a strong impact on big data and AI technologies. This can happen in two ways. Firstly, data usability can directly benefit from AI algorithms. Secondly, data usability can be an enabler of AI systems. For example, data usability factors can improve knowledge presentation, in terms of how data are provided to AI systems. Data usability also helps with data management, such as the organization and visualization of IoT data for both machines and humans. In addition, improving configuration and management tasks in IoT devices and platforms increases reliability and therefore the usefulness of data feeding AI systems.
Q3: How have you approached the project?
MW: Our standardization project aims to deliverer two documents. The first is a technical report that collects use cases to highlight the importance of data usability from real life scenarios. We look across different industries and practices to assess how data might be compromised in these different use cases.
The thirty-three use cases we documented covers a wide span. It includes healthcare covering disabilities, the environment, elderly users, and the pandemic. There are industry and manufacturing use cases, including towards the safety management of manufacturing plants and construction site workers. We also looked at agriculture, farming, energy, and the transportation sectors. We covered public and emergency services, an important class of IoT systems. Other areas covered buildings, vending machines, retail supply chains, large events, smart lifts, and smart cities. The project team analysed the consequences and impact if data are not usable in each of these scenarios. This analysis led to the second deliverable which specifies measures to ensure data usability from a generic point of view across all these domains.
The STF601 project team consists of four experts, including myself, with skills in IoT standardization, IoT and eHealth research and, oneM2M standards. Our STF started in February 2021 and the project team expects to complete mid-July 2022. We split our activities into three main tasks. One covers project management and interactions with standardization groups other than TC SmartM2M and the other two are for the technical activities.
Q4: What are some of the project’s early findings and IoT industry implications?
MW: The technical specification contains both service and operational requirements that different elements in an IoT system need to fulfil. These elements include sensor/data sources, IoT platforms, AI/ML or monitoring functions, IoT system operators, and data users.
As an illustrative example, let us take a service requirement on IoT platforms to mitigate data heterogeneity. The requirement specifies that data from various sources shall be transformed and/or aggregated, as necessary, to feed an ML algorithm while enabling system scalability. On the sensor side, there is a need to know the data confidence levels to enable proper processing by data consumers, whether these are humans or algorithms.
These service requirements are complemented with operational requirements. One might be necessity to provide reliable timestamps and geolocation of the data. These kinds of meta data contribute to the correct interpretation of the core IoT data. Another requirement is for the system operator to arrange periodic maintenance of the system to verify that the overall infrastructure works properly and that all components are able to communicate with each other. This leads to an operational necessity to resolve all issues found during the verification process. While this might look like common sense, such procedures might be dropped for financial reasons with significant consequences, as highlighted by our study results.
Other operational requirements cover the scalability of deployed systems and the capability to accept inputs from all sorts of sensors. There is also a requirement for monitoring components (for example sensing devices) to have the capability to issue alerts in cases where new data are expected but not provided.
Q5: How does the project relate to oneM2M?
MW: One of the objectives of the project is to deliver its finding to oneM2M. As a starting point, we aligned our use case template with the structure used in oneM2M. One of our team members has been actively involved with a new oneM2M technical report (TR). This is TR-0068 on System Enhancements to Support AI capabilities which is in the process of being prepared for publication.
We have reported progress on our work to oneM2M’s Requirements and Domain Modelling (RDM) working group throughout the course of our project. In parallel, our team also received progress updates on the oneM2M TR. We contributed a relevant subset of our use cases to the oneM2M TR. The same approach continues with the set of requirements from our technical specification (TS) deliverable, which will be considered to feed the oneM2M TR.
Q6: How can interested readers find out more about the project?
MW: I would advise people to start with our project webpage on the ETSI portal. I mentioned our Technical Report (ETSI TR 103 778) which has been published and is ready for download
The Technical Specification (TS 103 779) will be published over the coming weeks. We have delivered presentations at workshops and conferences of the OntoCommons European project. We also plan to present our results at the ETSI IoT Week event, which is due to take place in the week of October 10-14, 2022, in Sophia Antipolis, France.
People interested in related information can download other standards from TC SmartM2M: the SAREF standards (TS 103 264 and the series of SAREF extensions in TS 103 410), as well as the TR 103 674 (Artificial Intelligence and the oneM2M architecture; Architecture AI4IoT).
