RE: Requirements

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

> For the past 2.5 years I have been asking the question is
> there anything special about wireless that impacts application,
> transport, or IP protocols that is different than wired.
>
> Up til now, the only exception for wireless than wired that
> I have discovered is handoff.  I believe that we SHOULD
> view the wireless access network as just another Layer 2.


Dana,

I've been on a similar quest and generally agree with your findings. Most so-called wireless issues have to do more with disageements on optimization than they do with unique technology.

Wireless links may have: asymmetric bandwidth; asymmetric, time-variant error characteristics; high transmission or resource utilization costs. Sure, any or all of these can happen on a wired system as, but they are relatively uncommon in wired systems and very common in wireless systems. A protocol designed for wired probabilities / profiles can behave pathologically given wireless probabilities / profiles.

The canonical example is TCP, which was designed (rather, optimized, in the back-off algorithms) on the assumption that the most likely cause of packet loss is congestion. In any system - whether wired or wireless - where packet loss occurs frequently for other reasons, TCP behaves pathologically.

The emphasis need not be on the word "wireless"; rather, it should be on parameterizing transmission characteristics and making certain that protocols work properly for more than one set of parameters (or, alternatively, that protocols are developed for specific profiles, with assumptions stated explicitly).

Chris Burke
Motorola Internet Software and Content Group