IoT Naming System Said ‘Critical’ for Network of Connected Devices But Which One Unclear

The naming issue is complicated because three key topics must be distinguished, Santucci said. The first is the split between multiplicity and uniqueness of object identifiers. No one can say at this stage whether usage will drive developments toward a globally unique identification scheme or several distinct ID spaces, he said. The public policy implications are different for each scenario, he said. If a globally unique regime emerges, akin to Internet Protocol (IP) addresses, questions similar to the critical Internet resources debate will probably have to be addressed, he said. If the scenario of multiple identifier spaces prevails, interoperability among the different systems “would be a major challenge,” he said.

The second issue is the difference between the identification of an object and its network address, Santucci said. The former provides a unique handle for the object itself, but the latter might change depending on where the object is located, its logical membership in one or more networks or its current role, he said. The Electronic Product Code is one of the well-known object identification systems that can uniquely identify objects associated with a RFID tag, he said. Similarly, addressing programs could be different, he said. Objects now connected to the Internet use the global IP addressing system (IPv4 or 6), he said. Some other things may use private addresses rather than the global system, he said. Even where objects use a private network, they might still be connected to the Internet, which will often be used to bridge the various private networks, he said. In that way, a device should be able to “speak” different addressing schemes to operate as part of different networks, he said.

The third major issue relates to the approaches for finding information about a thing with a specific identification, Santucci said. There are two approaches. One is object resolution, a straightforward process based on prior knowledge that will yield at least one address that should have information about the device. The assumption is the existence of a naming system that lets the user or another device in the system find the object it’s looking for, he said.

The Internet generally uses a hierarchical naming system, the URL, but that’s not suitable for a highly mobile environment, Santucci said. Since URLs only disclose information about the network where something -- say, a computer -- is located, it’s very expensive to move between networks, where the “name of the computer” has to change when moving from one network to the next, he said. The second approach is object discovery, Santucci said. This is “more like ‘Googling’ for information” without before-hand knowledge, finding previously unknown sources of information, he said. Object discovery is a trivial task in relatively small networks of devices but certainly not in a network of millions of objects where using the same scheme would impose huge performance requirements on any network design, he said. With new objects and services constantly evolving and network topologies changing all the time, automated discovery mechanisms are needed, he said.

The IoT naming debate is part of the larger discussion on IoT governance. “There is no real progress with regard to a clear definition of IOT governance,” EU IoT Task Force on Governance Chairman Wolfgang Kleinwächter wrote in a June 19 interim report. One reason is the continued disagreement between those who see IoT governance as a special subject that needs unique rules and mechanisms, and those who view it as part of Internet governance as whole, he said.

There are several issues which have a governance dimension and need further clarification, the task force said in a February issue paper. One is how to guarantee the uniqueness of identifiers linked to objects, including the allocation and management of IPv6 addresses and of RFID tags, it said. Talks so far, however, have indicated that there’s no need for IoT governance aimed at ensuring that each object can communicate with any other object; or that there’s a single, “authoritative” or “legacy” root for managing an object naming and numbering system, it said.

On top of all that there are efforts to introduce a new approach which goes beyond the DNS and its existing governance structure, the June interim report said. Robert Kahn, founder and president of the Corporation for National Research Initiatives (CNRI), has proposed using the “Handle” system and establishing a digital object naming authority to work independently from existing Internet governance institutions or intergovernmental treaty organizations as a service provider for interested customers, the report said.

Can ‘Handle’ Handle the IoT?

The Handle system is concerned with management of information in digital form, Kahn and Patrice Lyons wrote in a paper, “Addressing in the Internet: Application to RFID Information,” submitted as a contribution to an upcoming American Bar Association book. CNRI developed a digital object architecture (DO architecture) and a protocol for accessing a DO, they said. A DO is a data structure that is machine- and platform-independent and has a unique, persistent identifier known as a “handle” or digital object identifier.

A resolution system maps identifiers into “state information” about the DO being identified, the paper said. The state information could be the number of locations where the DO may be accessed on the Internet, such as a list of IP addresses, or terms and conditions for access to and use of the DO, they said. Individual DOs are stored in Internet repositories and accessed by their unique identifiers, they said.

There are many IPv6 addresses which can be used for objects on the IoT, but many people feel IPv6 is less than optimal for everything, although the IoT was originally envisioned as one of its main applications, Kahn told us. Movement of things is another reason why IPv6 may be less than perfect, especially when routing issues and IPv6 are tightly linked, he said. Handle would allow the two to be separated and enable identification of information about a thing by its handle, Kahn said. Each handle could be resolved to all kinds of relevant state information about the thing, including any current IPv6 address it has, he said.

The Handle system is separate from the DNS, Kahn said. The ONS appears to work well today, but it’s limited because it’s based on the location-restricted DNS, he said. For interoperability across heterogeneous information systems, such as those under development in the healthcare and financial industries, where it may be useful to have connectivity between an identifier embedded in an RFID chip and other related data, “there is a need for a more flexible technical approach,” he said. The Handle system, together with the other components of the DO architecture, provides a dynamic capability for doing business on the Internet, he said.

Kahn has also proposed a digital object naming authority independent from existing Internet governance institutions for the IoT. Many organizations claim roles in Internet governance but most, if not all, would argue that there’s no such thing as a “Governor of the Internet, unless it’s the protocols and procedures that make it work,” Kahn said. CNRI now is the sole administrator of the root for the Handle system, but not everyone is happy with that administration sitting in a single organization, he said. So CNRI is setting up the digital object numbering authority foundation in which multiple parties will take over administration of the global Handle registry in the public interest, he said.

Several groups in the Internet Engineering Task Force have discussed Handle and its relationship to existing systems of identifiers, “but I don’t think there is consensus today,” Santucci said. That explains why Handle hasn’t been deeply investigated by the IoT Expert Group, a panel of international representatives from industry, academia, government and civil society advising the EC, so far, he said. But, he said: “I believe that it actually provides a relevant option if we assume that the IoT will not develop separately from today’s Internet -- why should we take the risk of encouraging market fragmentation? -- and if we take into consideration that it takes approximately 10 years from inception to production implementation -- so why should we wait for a new architecture and delay the IoT take up and adoption?"

In the near future, the IPv6 and Handle together might help address key requirements such as scalability, centralization or distribution of deployment models, versatility to handle all types of traffic, and interoperability among all available industry standard network links, Santucci said. It’s too early to know if there will be agreement on Handle, he said. The IoT governance group “believes there is a need for further discussion” by a broader range of stakeholders, Kleinwächter said in his June report.