RIOT in comparison to other 6LoWPAN implementations

Development archive
1

Hi everyone,

a presentation[1] in yesterday's meeting of the IETF 6LO group, Hudson Ayers et al compared various 6LoWPAN implementations with the goal of understanding how interoperability barriers can be overcome.

Some comparisons between the implementations are an intermediate result, and I was happy to find that while RIOT does not win in the individual categories, it provides the second largest set of features, while having the second smallest code size footprint (thus likely giving the best "bang for the buck").

There will be recording of the presentation online at a later point in time; the full paper appears to be online at [2].

Seems to me you all did a pretty good job here :slight_smile:

chrysn

[1]: https://www.ietf.org/proceedings/108/slides/slides-108-6lo-design-considerations-for-low-power-internet-protocols-00 [2]: https://arxiv.org/abs/1806.10751

2

Hi,

very interesting set of slides (only had a cursory look over the paper so far, but will definitely have a closer look!)

Some comments as one of the main implementors:

  • They missed IPv6 tunneling and extension header support by a few days :wink: (paper was published Jan 21, [1] was merged Jan 31).
  • Not 100% sure what the question on slide 8 is about, but the quoted line is not an optimization actually, but to allow for UDP header compression even when UDP is not compiled in (e.g. on a thin forwarder). So while the other examples are all about saving memory within 6LoWPAN, our line is actually about providing a 6LoWPAN feature, despite the upper layer missing.

Best regards, Martine

[1] https://github.com/RIOT-OS/RIOT/pull/12879