RE: What is the purpose of the discussion? Answer

[Burke Chris-CCB007 <Chris.Burke@motorola.com>]

James Kempf asked about the point of my laundry analogy. Many industries are being changed by internet technologies, and each industry has its own legacy issues and values. I'm pretty close to the wireless industry and have been for some time, and so when thinking about an issue that may be "too close to home" for maximum objectivity, it helps me to consider a parallel construct from an entirely different industry. Hence the laundry example.

If someone were to submit a draft about internet issues related to laundry systems, my first reaction would be, "Just use the technologies that are out there - we don't need a special 'laundry internet.' If you have some small number of specific internet technology issues, lets focus on those; the rest is prior art (out of scope) or a private, huge legacy systems integration issue (out of scope)."

And so, this I think will be the reaction of many outside of the wireless industry, to the draft under discussion. I don't find the reaction off-base; the laundry example helped me, so I shared it.

Paul Reynolds noted that the draft is not intended to describe a framework or architecture. I think we're into a semantic argument - the draft is peppered with "the wireless internet framework needs to..." and that reads to me like (1) description; (2) of a framework. For me any set of requirements is an abstract description of the solution; that's probably the source of my interpretation of the text.

James Kempf also offered QoS and location-based services as wireless-specific requirements.

Certainly QoS by overprovisioning and statistical multiplexing isn't an ideal solution when bandwidth is scarce. The current IETF solution is sub-optimal, regardless of whether the cause of the scarcity is "wirelessness"; for example, it is sub-optimal over 56k telephone links as well. Circuits may turn out to be a pretty good QoS solution at narrowband edges; work is needed in the area of narrowband QoS but the work is not wireless-specific.

Location-based services are also not wireless-specific. Any commercial internet, whether wired or wireless, benefits from being able to offer location-based services. A common application layer interface to location information is desirable. I think a good starting point is to define a data format for location information (XML DTD or other preference). This can be wrapped in a variety of method invocation technologies while concealing the underlying implementation.

These two points might be a start of our "laundry list" (forgive the reference) of issues that the IETF (or W3C) could address, in non-wireless-specific ways, that would help enable all-IP wireless networks.

1) Narrowband QoS
2) XML DTD for Location Information
3) (put more stuff here)
...

Chris