Last Call Review of draft-ietf-6lo-btle-13
review-ietf-6lo-btle-13-genart-lc-davies-2015-06-22-00

I am the assigned Gen-ART reviewer for this draft. For background on
Gen-ART, please see the FAQ at

<

http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.

Please resolve these comments along with any other Last Call comments
you may receive.

Document: draft-ietf-6lo-btle-13.txt
Reviewer: Elwyn Davies
Review Date: 2015/06/02
IETF LC End Date: 2015/06/09
IESG Telechat date: (if known) -



Summary: Almost ready.  A couple of minor issues - I think it is 


important to bring out the multi-link subnet model as a separate section 


and there are some details about peripheral to peripheral communication 


that seem to have conflicting statements.  Otherwise, there are a 


largish number of missing definite and indefinite articles and a few 


clarifications that I have suggested.




Major issues:
None.

Minor issues:


s2.2:  As I read this section, it seems to imply (but doesn't make it 


totally clear) that a peripheral would only connect to a single central 


over IPv6 at a time.  Is this true?  Mesh networking being out of scope 


would appear to be another issue that would not necessarily preclude a 


peripheral from having multiple interfaces to different centrals - 


provided it was capable of handling the multiple interfaces - but would 


not need to act as a router.  OK, this might not be something that would 


be an everyday event in the IoT but it would be good to make it clear 


what the specification implies in this area.






s3.2, para 5:  [This is just fractionally more than editorial]  The 


selection of the multilink subnet model is important and deserves a 


(sub-sub-)section to itself.  I suggest moving the second and subsequent 


sentences of para 5 of s3.2 to a new s3.2.1 and renumbering the 


remaining sub-sub-sections, thus:



OLD:
   Bluetooth LE connections used to build a star topology are point-to-
   point in nature, as Bluetooth broadcast features are not used for
   IPv6 over Bluetooth LE (except for discovery of nodes supporting
   IPSS).  As the IPv6 over Bluetooth LE is intended for constrained
   nodes, and for Internet of Things use cases and environments,
   multilink model's benefits are considered to overweight multilink
   model's drawbacks described in RFC 4903 [RFC4903].  Hence a multilink
   model has been chosen, as further illustrated in Section 3.3.
   Because of this, link-local multicast communications can happen only
   within a single Bluetooth LE connection, and thus 6LN-to-6LN
   communications using link-local addresses are not possible. 6LNs
   connected to the same 6LBR has to communicate with each other by
   using the shared prefix used on the subnet.  The 6LBR ensures address
   collisions do not occur (see Section 3.2.2) and forwards packets sent
   by one 6LN to another.

   After the peripheral and central have connected at the Bluetooth LE
   level, the link can be considered up and IPv6 address configuration
   and transmission can begin.

3.2.1.  Stateless address autoconfiguration
...
NEW:
   Bluetooth LE connections used to build a star topology are point-to-
   point in nature, as Bluetooth broadcast features are not used for
   IPv6 over Bluetooth LE (except for discovery of nodes supporting
   IPSS).
   After the peripheral and central have connected at the Bluetooth LE
   level, the link can be considered up and IPv6 address configuration
   and transmission can begin.

3.2.1. IPv6 Subnet Model

   In the Bluetooth LE piconet model (see Section 2.2) peripherals each
   have a separate link to the central and the central acts as an IPv6
   router rather than a link layer switch.  As discussed in [RFC4903],
   conventional usage of IPv6 anticipates IPv6 subnets spanning a single
   link at the link layer.  As IPv6 over Bluetooth LE is intended for
   constrained nodes, and for Internet of Things use cases and
   environments, the complexity of implementing a separate subnet on
   each peripheral-central link and routing between the subnets appears
   to be excessive.  In the Bluetooth LE case, the benefits of treating
   the collection of point-to-point links between a central and its
   connected peripherals as a single multilink subnet rather than a
   multiplicity of separate subnets are considered to outweigh the
   multilink model's drawbacks as described in RFC 4903 [RFC4903].

   Hence a multilink model has been chosen, as further illustrated in
   Section 3.3  Because of this, link-local multicast communications
   can happen only within a single Bluetooth LE connection, and thus
   6LN-to-6LN communications using link-local addresses are not
   possible. 6LNs connected to the same 6LBR have to communicate with
   each other by using the shared prefix used on the subnet.  The 6LBR
   ensures address collisions do not occur (see Section 3.2.3) and
   forwards packets sent by one 6LN to another.

3.2.2.  Stateless address autoconfiguration
....
END



s3.2, para 5: Additional query:  As written and in the nEW version just 


above, the text seems to imply that you can't do peripheral to 


peripheral comms between peripherals connected to the same central using 


link local addresses ("thus 6LN-to-6LN communications using link-local 


addresses are not possible").  This appears to conflict with various 


statements later in the draft in s3.2.3, paras 4 and 5 which seem to 


imply that comms using link local addresses is entirely possible.  I may 


have misunderstood this but would be grateful for explanation.





Nits/editorial comments:
========================
General: Prefer 'octet' to 'byte'.



General: Consistent use of 'link layer' or 'link-layer' - both are used 


in a somewhat aribitrary fashion.






Abstract: s/is standardized since the revision/has been standardized 


since revision/




Abstract:  The 6LoWPAN acronym needs to be expanded.



s1: To be consistent with the abstract it would be best if the 


connection (equivalence) between Bluetooth Smart and Bluetooth LE was 


reiterated in the first sentence and noted that Bluetooth LE is used in 


the rest of the document.






s1, para 1: To avoid jargon s/coin cell/very low capacity (e.g., "coin 


cell")/






s1, para 2: Suggest s/ideal protocol/ideal protocol for communication 


with such devices/  to make it clear what it is ideal for.






s1, para 3: 802.15.4 needs a reference.  Should this be to the updated 


2011 version ( RFC 4944 refers to the 2003 standard)?


IEEE Computer Society, "IEEE Std. 802.15.4-2011 IEEE Standard for Local 


and metropolitan area networks--



Part 15.4: Low-Rate Wireless Personal Area Networks (LR-WPANs)", June 2011.



s1, para 3:  The general connection to 6LoWPAN mentioned in the last 


sentence of the Abstract ought to be repeated at the start of para 3;



OLD:
   RFCs 4944, 6282, and 6775 [RFC4944][RFC6282][RFC6775] specify the
   transmission of IPv6 over IEEE 802.15.4.
NEW:


   This document describes how IPv6 is transported over Bluetooth Smart 


(also known as


   Bluetooth LE) low power connections using IPv6 over Low power 


Wireless Personal Area


   Networks (6LoWPAN) techniques.  RFCs 4944, 6282, and 6775 


[RFC4944][RFC6282][RFC6775]


   developed for 6LoWPAN specify the transmission of IPv6 over IEEE 


802.15.4.



END



s1.1, para 2: Please include the expansions of the acronyms 6LN, 6LR and 


6LBR here.




s2, para 1:



    Bluetooth LE is designed for transferring small amounts of data
    infrequently at modest data rates at a very low cost per bit.



Presumably this is energy cost?   Maybe better:
s/at a very low cost/with a very small energy expenditure/

s2, para 2: s/published/published the/,
                    s/includes/includes the/,
                    s/link-layer/a link-layer/



s2, para 2: To be clear 'newer' (presumably) applies to both BT 4.1 and 


IPSP 1.0:


s/newer/more recent versions of either specification to provide 


necessary capabilities/




s2, para 3:
OLD:
which will include Bluetooth 4.1 chipsets
NEW:
that incorporate chipsets implementing Bluetooth 4.1 or later
END

s2, para 3:  The standard cannot guarantee presences of Bluetooth LE:
s/will also be included/is also expected to be included/

s2.1, para 1:



The device internal Host Controller Interface (HCI)


I don't think the qualifier 'device internal' really helps here and 


could be usefully omitted.



The nature of the interface is explained in the following text.



s2.2, para 1: 'but since Bluetooth 4.1' is a bit ambiguous.  After some 


thought, I assume it is supposed to mean that the capability to connect 


to multiple centrals at once started only with BT 4.1.  How about:



OLD:
but since Bluetooth 4.1 can also connect to multiple centrals.
NEW:


but with versions of Bluetooth from 4.1 onwards it can also connect to 


multiple centrals at the same time.



END

s2.2, para 1: s/i.e. a Bluetooth/i.e., a Bluetooth/

s2.3, para 1:  Hope the device isn't running on AC!!
OLD:
This typically happens at every power cycle of a device.
NEW:
New addresses are typically generated each time a device is powered on.
END

s2.4, title: s/packets sizes/packet sizes/

s2.4: s/Optimal/The optimal/,
         s/Default MTU/The default MTU/,
         s/exclusing L2CAP/excluding the L2CAP/,
         s/protocol data/a protocol data/,
         s/link layer fragmentation/a link layer fragmentation/,
         s/provides MTU/provides an MTU/


         s/single link-layer MTU value/an equal link layer MTU for the 


two directions/




s3, para 1: s/due Bluetooth LE's/due to Bluetooth LE's/

s3, para 2: s/both star and mesh topology/both star and mesh topologies/,
                   s/inter- peripheral/inter-peripheral/ [Space removed]

s3, para 4: s/Bluetooth SIG/the Bluetooth SIG/



s3, last sentence:  This duplicates the statement made in s2 (subject to 


the mod proposed above).  It could be omitted.  However, if repeating it 


is thought worthwhile, it might be better right at the start of s3.




s3.1: s/IPv6 stack/the IPv6 stack/,
         s/GATT stack/the GATT stack/,
        s/Bluetooth LE/the Bluetooth LE/,
        s/GATT/The GATT/
        s/Internet/The Internet/



s3.2, para 1: s/The concept of/The distinct concepts of the/, 


s/needs/need/, s/the relationship/their relationship/




s3.2, para 2: s/6LoWPAN/the 6LoWPAN/,
              s/use of 1280 byte/use of a 1280 byte [octet] MTU/

s3.2.1, para 1: s/underlying/the underlying/,
                s/guidance/the guidance/,
                s/formed from 48-bit/from the 48-bit/,
                s/a bit from Bluetooth/a bit from the Bluetooth/,
                s/no bit in IID/no bit in the IID/

s3.2.1, para 2: s/appended with/prepended with the/,
                s/After Bluetooth/After a Bluetooth/


                s/existing L2CAP channel/existing L2CAP channels/ [or 


"an L2CAP channel"]




s3.2.1, para 3: s/responsible of/responsible for/,
                s/and of/and for/,
                s/complexity/the complexity/,
                s/at link establishment/in the link establishment/,
                s/number/the number/

s3.2.1, para 4: s/6LN sends/the 6LN sends/

s3.2.1, para 5: s/alternatice/alternative/,
                s/6LN generates/that a 6LN generates/,
                s/with 6LBR/with the 6LBR/



s3.2.1, para 6:  Referring to the new section on the subnet model 


proposed in the 'Minor Issues' section above would be helpful instead of 


s2.2.






s3.2.1, para 6:  s/Prefix Information Option [RFC4861]/Prefix 


Information Option in Neighbor Discovery messages [RFC4861] (see Section 


3.2.2)/




s3.2.3, para 3: s/of EUI-64/of an EUI-64/

s3.2.3, para 4: Suggest a little clarification of this para:
OLD:
   To enable efficient header compression, the 6LBR MUST include 6LoWPAN
   Context Option (6CO) [RFC6775] for all prefixes the 6LBR advertises
   in Router Advertisements for use in stateless address
   autoconfiguration.
NEW:
   To enable efficient header compression, when the 6LBR sends a
   Router Advertisement it MUST include a 6LoWPAN Context Option (6CO)
   [RFC6775] matching each address prefix advertised via a Prefix
   Information Option (PIO) [RFC4861] for use in stateless address
   autoconfiguration.
END



s3.2.4, para 5: s/related to compression context/related to the 


compression context/ (2 places)




s3.2.4, para 5:
OLD:


6LBR has sent Neighbor Advertisement with ARO having status field set to 


success.



NEW:


6LBR has sent a Neighbor Advertisement with an ARO having its status 


field set to success.



END

s3.2.4, para 5: s/address is 6LBR's/address is the 6LBR's/,


                s/for the used destination prefix./for the destination 


prefix used./




s2.2.4, para 6: s/packets to 6LN/packets to a 6LN/,
                s/on 6LBR's/on the 6LBR's/,
                s/based on 6LN's/based on the 6LN's/,
                s/then 6LN/then the 6LN/,
                s/of IID match/of the IID match/

s3.2.3.1, para 1: s/full or part of IID/part or all of the IID/,
                  s/depending which/depending on which/

s3.2.3.2, last para: s/secondly/most recently/