Tero makes some good points but perhaps I have a different view on the Layer 1 
fragmentation introduced with 802.15.4k.

Firstly, the Layer 1 fragmentation was early on intended to be a general 
function that could be used with any PHY, that was why it was in the MAC 
section for 802.15.4k.  During the 802.15.4-2015 revision effort we saw that 
there were some PHYs that would require additional effort on the mechanism (I 
cannot remember the specifics).  Rather than delay the revision further we 
chose the expedient way to move on which was to put it into the LECIM section, 
since they were the PHYs (DSSS in particular)  that absolutely needed Layer 1 
fragmentation.  In the next revision I will propose that, once again, we 
revisit the concept of extending the Layer 1 fragmentation to more PHYs as well 
as change the moniker LECIM to LPWAN.  

One of the requirements that we placed on Layer 1 fragmentation was the ability 
to abort the fragmentation/reassembly process at any point in the process.  
This was achieved with minimal impact to the overhead by setting the 
acknowledgment’s fragmentation number field and the acknowledgment content set 
to zero (ie. no packets were received).

While Layer 1 fragmentation has benefits over Layer 2 fragmentation, it does 
have some disadvantages.  Firstly, since the MAC will not accept datagrams that 
are larger than the size dictated by “aMaxPhyPacketSize” (a PHY constant that 
dictates the largest frame that the PHY can transmit, currently either 127 or 
2047 octets), if this value is set to 127, a Layer 2 fragmentation method is 
required for datagrams larger than 127 octets.  The second that seems important 
for mesh networks is that a Layer 1 fragmentation is solely 
link-by-link/hop-by-hop, ie. the reassembly must occur at each receiver.  While 
this is no problem for star topologies, it is a concern for mesh or multi-hop 
networks.

Pat

Pat Kinney
Kinney Consulting LLC
IEEE 802.15 WG vice chair, SC chair
ISA100 co-chair, ISA100.20 chair
O: +1.847.960.3715
[email protected] <mailto:[email protected]>
On 6, Apr2017, at 7:34, Tero Kivinen <[email protected]> wrote:

Pascal Thubert (pthubert) writes:
> IEEE 802.5.4 understood and started addressing those issues in their
> protocols years ago, consider for instance how this is done at PHY
> layer in 802.15.4K LECIM - LP-WAN (cc'ing Pat). In the light of that
> art, the legacy 6LoWPAN technique is shamefully inadequate.

802.15.4k is really only a phy level fragmentation, and the main issue
there is that the max packet size in that PHY is extremely small
(less than 30 octets).

When designing the IEEE std 802.15.9 we checked 4k that and realized
that it is not something that can be used in general case, and thats
why the MAC level fragmentation in the 802.15.9 is using different
method. Fragmentation was needed in the 802.15.9 as KMP payloads do
not fit to the small frames of 802.15.4. This fragmentation method
(and the multiplexing) from the 802.15.9 will most likely be adopted
in the 802.15.12 later...

Btw, the 802.15.9 is now available using the get IEEE 802 program, and
you can download it from

http://standards.ieee.org/getieee802/download/802.15.9-2016.pdf

The fragmentation wire formats are described in the section 7, and
state matchines are described in section 9.

> My suggestion to the group is to take a holistic view of these
> issues, and redesign or fragment support. This is what my draft
> does.

The 802.15.9 is hop by hop fragmentation, so if you want to be able to
forward fragments, you need to specify your own way of doing that... 
-- 
[email protected]

_______________________________________________
6tisch mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6tisch

Reply via email to