Re: [smartpowerdir] question on identification of sensors

[Henning Schulzrinne <[e-mail]>]

Generalized, how about UUIDs?
http://en.wikipedia.org/wiki/Universally_Unique_Identifier
http://tools.ietf.org/html/draft-kindel-uuid-uri-00

On Jul 28, 2010, at 2:32 PM, Jari Arkko wrote:

> I have a practical question for the directorate. The question relates to what I'm doing as a hobbyist in my own sensor networking, but I think its also a wider question that would have importance for others.
> 
> My sensors have identities. They may or may not have (stable) network layer addresses or DNS names. In some cases a sensor might get different addresses at different times (movement), not exist as an IP host but as a legacy device behind a gateway, or a single IP host might be able to make a number of measurements (temperature, location, humidity). Information from the sensors gets processed in various ways, including filtering by servers, archival storage, application layer routing for the interested consumers, etc. In my mind, all of these factors speak in favor of identifying sensors or "things" in application layer data. As part of some XML structure sent somewhere, in a log file, and so on.
> 
> My question relates to whether we need ways to represent "thing" identities in URNs. To make the question more concrete, I'm running a bunch of sensors that have either 64-bit 1-wire hardware identifiers or Ethernet MAC addresses. I would like to have an identifier format that I can stick to different places with a consistent syntax. However, there is no way to represent MAC addresses or other hardware in URNs at the moment, AFAIK. The question is, would adding such identifiers in the URN space make sense from an architectural perspective? If I google for some old discussions around supporting MAC addresses in URNs I find flame wars. So as an IP guy I plead some assistance from the group, hoping that you have more understanding of the application layer constructors and appropriate designs there.
> 
> Jari
> 
>