Dear Mister Jara,

Thank you for your reply.


> But now with the WPAN such as IEEE 802.15.4 and Bluetooth Low Energy, where
> we have a constrained frame size, 127 and 27 bytes respectively, we are not
> only interested in embed the IP stack and reduce the code and memory
> requirements, we are also interested in reducing the header size, in order
> to reach the higher payload capability.

I perfectly understand your goals. But you made one mistake in your
reasoning : you assume that there is a future for IEEE 802.15.4.

Not only, IEEE 802.15.4 doesn't provide the openness needed to be
supported by strong communities like the Free Software Foundation [0],
but also the industry is clearly moving towards Wifi that has the main
advantage to be available now :

http://www.slideshare.net/mkovatsc/connecting-things-to-the-web-using-programmable-lowpower-wifi-modules

So, on my side, I assume that IEEE 802.15.4 is dead. And I doesn't
seem to be the only one :)

Best Regards;

[0] http://en.qi-hardware.com/wiki/Ben_WPAN

--
Guillaume FORTAINE
[email protected]
DevOpSpace
http://www.devopspace.com
+33(0)631.092.519


On Fri, Jan 27, 2012 at 1:38 AM, Antonio Jara <[email protected]> wrote:
> Dear Guillaume,
>
> On Fri, Jan 27, 2012 at 12:41 AM, Guillaume Fortaine
> <[email protected]> wrote:
>>
>> Hello,
>>
>>
>> > I am not agree with the sentence of "However, 6LoWPAN is not the way to
>> > go,
>> > we need plain IPv6", and this is not coming from the [1] reference such
>> > as
>> > indicated in previous emails. And remark that I am agree with Prof.
>> > Bormann
>> > that 6LoWPAN is in some way plain IPv6 sinde for the end nodes/clients
>> > this
>> > looks as IPv6 and they don't need to make any differente, this
>> > considerations are only required in a local level between the 6LoWPAN
>> > device
>> > and the 6LoWPAN Border Router.
>>
>> I totally assume my sentence. Especially, I am asking again, where is
>> the business need for a new stack when you can embed a full IPv6 stack
>> ? IPv6 was thought from the beginning for the Internet of Things, so
>> why should we reinvent a (squared) wheel ?
>>
> I have already mentioned to you, one issue is to be able to embed a full
> IPv6 stack, such as uIP, lwIP and Flexible IP are reaching.
>
> Please, notice that lwIP is coming even before than IEEE 802.15.4 was
> defined. lwIP was defined for embedded systems in general, for example over
> Ethernet, where it has not so frame size constrains.
>
> But now with the WPAN such as IEEE 802.15.4 and Bluetooth Low Energy, where
> we have a constrained frame size, 127 and 27 bytes respectively, we are not
> only interested in embed the IP stack and reduce the code and memory
> requirements, we are also interested in reducing the header size, in order
> to reach the higher payload capability.
>
> That is the main reason because it is defined 6LoWPAN, and that is also the
> main difference between lwIP and 6LoWPAN.
>
> Glowbal IP is just a new generation of optimizations to continue reducing
> header overload, assuming that each time new IoT technologies such as
> Bluetooth Low Energy are focused on reduce frame size, in order to reach
> high lifetime, reduce power consumption etc...
>
> Remark, it is not only memory issues, it is also performance isues regarding
> payload capability, overload from IP headers and lifetime from the devices.
>
>>
>>
>> >> I would prefer to discuss these things from a technical angle, not by
>> >> pointing to content-free marketing sites/slides.
>>
>> Let's discuss from a technical angle. Especially, I would greatly
>> appreciate the profile of an embedded engineer working for a Telecoms
>> Service Provider, if you wouldn't mind.
>>
>> By the way, I really appreciated DASH7 approach by making a market
>> study and focusing on very low cost devices (2 USD). According to my
>> knowledge, the highest density Cortex-M0 device is the Samsung S3FN41F
>> (32K SRAM/256K Flash) [0]. Why should we bother with the obsolete
>> MSP430 Family that is, moreover, higher priced ? There is absolutely
>> no economic reason to support this.
>>
>> Moreover, you are all claiming the necessity  of 6LoWPAN. However, I
>> didn't you see your names in the Linux Kernel 6LoWPAN stack [1].
>> Especially, did you know that the only supported USB device on Linux
>> for 802.15.4 was the ATUSB [2] ? Again, I didn't see your names on the
>> mailing-list.
>>
>> How could you claim that it is essential or even working if you don't
>> actively participate in its implementation with an end-to-end approach
>> ? If it is not under Linux, under which OS do you plan to use a
>> "6LoWPAN Border Router" ? Why a network service provider should bother
>> with this kind of devices ? Did you think about the associated
>> increase of CAPEX/OPEX costs for networks at a global scale ? How
>> could we justify a such investment ?
>>
>> 6LoWPAN is dead :
>>
>>
>> http://daruino.blogspot.com/
>>
>> "The blog of the Daruino open source hardware project - Breadboard to
>> production, evolve with Daruino!"
>>
>> Best Regards,
>>
>> [0]
>> http://www.samsung.com/global/business/semiconductor/productInfo.do?fmly_id=804&partnum=S3FN41F&xFmly_id=802
>> [1]
>> http://linux-zigbee.git.sourceforge.net/git/gitweb.cgi?p=linux-zigbee/kernel;a=shortlog;h=refs/heads/6lowpan
>> [2] http://en.qi-hardware.com/wiki/Ben_WPAN
>>
>> --
>> Guillaume FORTAINE
>> [email protected]
>> DevOpSpace
>> http://www.devopspace.com
>> +33(0)631.092.519
>>
>>
>> On Thu, Jan 26, 2012 at 11:56 PM, Antonio Jara <[email protected]> wrote:
>> > Dear All,
>> >
>> > Since, I have been involved in this discussion from my slides in the IoT
>> > Forum regarding IPv6 and IoT.
>> >
>> > I would like remark that I am agree with the mention from Joe Touch
>> > about
>> > that we need to make a difference between overload of the headers and
>> > also
>> > the lightweight of the stack, i.e. number of KB required in ROM and RAM
>> > space.
>> >
>> > It is the main issue, we need to mark the difference between lwIP, uIP,
>> > Flexible IP which are approaches focused on reduce the number of KBs for
>> > ROM
>> > and RAM, with other approaches such as 6LoWPAN which is mainly focused
>> > on
>> > reduce the number of bytes in the overload i.e. in the headers through
>> > their
>> > header compression.
>> >
>> > Both are reducing and optimizing IP but in two different ways.
>> >
>> > Few weeks ago, I already had an exchange of emails with Gullaime
>> > Fortaine,
>> > where he was interested about our Glowbal IP approach. Which is
>> > following
>> > the idea of 6LoWPAN about reduction of header size, at the same way that
>> > RFC4944 was optimized with RFC6282 through the contexts for the global
>> > IP
>> > addressing, we consider that it needs to be yet reduced for new
>> > technologies
>> > such as Bluetooth Low Energy.
>> >
>> > I would like also remark that the my sentence is that " 6LoWPAN is
>> > really
>> > useful, but this presents limitations for  global
>> > communications (RFC6282)" from [1].
>> >
>> > I am not agree with the sentence of "However, 6LoWPAN is not the way to
>> > go,
>> > we need plain IPv6", and this is not coming from the [1] reference such
>> > as
>> > indicated in previous emails. And remark that I am agree with Prof.
>> > Bormann
>> > that 6LoWPAN is in some way plain IPv6 sinde for the end nodes/clients
>> > this
>> > looks as IPv6 and they don't need to make any differente, this
>> > considerations are only required in a local level between the 6LoWPAN
>> > device
>> > and the 6LoWPAN Border Router.
>> >
>> > [1]
>> >
>> > http://www.iot-forum.eu/events/launch-event/presentations/wg-technology/slides%20IoT%20Forum%20upload%20Web_Jara.pdf
>> >
>> > And I explain that mention, since such as presented in the slide 7 from
>> > [1],
>> > this continues presenting a high overload for global IP addressing, e.g.
>> > 2001::1, instead of the usual link local FE80::1.
>> >
>> > Then, not only for IEEE 802.15.4 where we have a frame size of 127
>> > bytes,
>> > also for technologies such as Bluetooth Low Energy, where we have
>> > 27bytes,
>> > we define other approaches, the mentioned Glowbal IP, which you can find
>> > more information in:
>> >
>> > http://www.esiot.com/MIS_Glowbal_IP_Jara.pdf
>> >
>> > Finally, regarding these comments, I enclose a brief reflexion about the
>> > evolution of lwIP, 6LoWPAN, Glowbal IP and their reasons...
>> >
>> > ---------
>> >
>> > First, we need to consider the dates from each approach.
>> >
>> > LwIP is from 2001, it was a solution to support an implementation of the
>> > TCP/IP protocol stack in embedded systems with highly constrained
>> > capabilities (from 8 to 32 bits CPUs). The main goal of LwIP stack is to
>> > reduce memory usage and code size.
>> >
>> > LwIP stack was not defined for any specific physical technology. It can
>> > be
>> > from Ethernet (usually) to WiFi. Remark, that in 2001 was also being
>> > defined
>> > IEEE 802.15.4, and it was not extended until 2003 with the standard
>> > version.
>> > For example, if you review the lwIP document
>> > (http://www.sics.se/~adam/lwip/doc/lwip.pdf) it is not defined any
>> > reference
>> > to IEEE 802.15.4.
>> >
>> > After this... Then it was defined the LoWPANs, i.e. these constrained
>> > and
>> > embedded systems with Wireless Networking with capabilities for
>> > communications, and then it was extended to these networks the use of
>> > lwIP,
>> > but not only lwIP, it was also defined by SICS and even lower size
>> > solution
>> > with uIP.
>> >
>> > But then... this is defined a new requirement with IEEE 802.15.4 and the
>> > LoWPANs, it is not only the code/memory size, it is also the overhead of
>> > IP
>> > for technologies with a frame size of 127bytes (IEEE 802.15.4 case),
>> > which
>> > is even lower after of MAC layer headers. Leaving it to a few bytes.
>> > Then,
>> > it is feasible to have more than 60bytes on headers?, it is feasible to
>> > consider the current IP and UDP headers for this so frame size
>> > constrained
>> > technologies.
>> >
>> > The answer was obviously not, and in 2007 borns with this idea the first
>> > RFC
>> > about these requirements with the RFC4919, and after its consolidation
>> > with
>> > the RFC4944, it is defined 6LoWPAN!
>> >
>> > Then SICS people, implements a lightweight version also following some
>> > design issues and ideas from lwIP/uIP and define SICSLoWPAN, in order to
>> > also continue keeping the purposed of low size/memory in conjunction
>> > with
>> > this new goal of low overhead.
>> >
>> > 6LoWPAN was ok for reducing overload, but it was found several problems
>> > for
>> > the cases with global addressing, since the addressing compression was
>> > too
>> > focused on stateless-configuration address from the MAC EUI-64, in order
>> > to
>> > elide it.
>> >
>> > Then, it was starting to think how to solve more generic cases for
>> > multicast, global addressing etc..., and it was defined the idea of
>> > contexts
>> > instead of the prefix addresses, which idea was consolidated in the last
>> > RFC
>> > for 6LoWPAN, RFC6282 in 2011. But, this continues without a suitable
>> > compression, such as it is presented in the first section from the
>> > document
>> > of Glowbal IP, but ok it is useful.
>> >
>> > The problem in 2011, it is that we continue with our ideas to introduce
>> > Internet everywhere, and other technology also very present in the WPANs
>> > world is Bluetooth, and they lunch Bluetooth 4.0, with the capabilities
>> > for
>> > a version of Bluetooth Low Energy in order to be competitive with
>> > 6LoWPAN/IEEE 802.15.4/ZigBee.
>> >
>> > But the miracles are not so simple, and how they got it?, they define a
>> > too
>> > constrained communication medium with a limitation of frame size to 27
>> > bytes
>> > for the PDU!!!!!
>> >
>> > So, from the highly constrain of 127 bytes in IEEE802.15.4, now we have
>> > limitations to 27 bytes, and they are trying to move 6LoWPAN to Bluetooh
>> > Low
>> > Energy.
>> >
>> > Obviously, the current  6LoWPAN approach for Bluetooth Low Energy is
>> > focused
>> > on remove everything, limit it to local communications, remove even
>> > ports
>> > (defining a fixed port)... then when you analize this kind of
>> > approaches,
>> > and remember the Internet of Things goals, it is obvious that we are
>> > losing
>> > the link with the original motivation.
>> >
>> > For that reason, in 2011/2012 is defined Glowbal IP, which is a new
>> > approach
>> > to reduce the overloads to 5bytes, and define a solution suitable for
>> > LoWPANs such as IEEE 802.15.4-based, and also to satisfy the
>> > requirements
>> > from new technologies such as Bluetooth Low Energy
>> >
>> > Best regards,
>> > Antonio J. Jara
>> >
>> >
>> > On Thu, Jan 26, 2012 at 11:12 PM, JP Norair <[email protected]> wrote:
>> >>
>> >> Marketing is in some ways a dirty business, but it other ways it is
>> >> completely necessary.  I do not intend to cheapen other peoples' work
>> >> and
>> >> research, but if I were not provocative this discussion would never
>> >> have
>> >> happened.  When we began thinking about DASH7 in 2007, the first thing
>> >> we
>> >> started doing was market research.  We did market research for three
>> >> years.
>> >>  As a result, we are not using 802.15.4 + 6LowPAN as a foundation for
>> >> DASH7.
>> >>
>> >> The 6LowPAN stack is a fine set of technologies, but there is no focus.
>> >>  For US$2 embedded devices, focus matters.  In our market research, it
>> >> became very clear that only three elements of the IP-stack mattered [to
>> >> our
>> >> market].  They are: UDP, TCP, IPsec.  Also, the market did not care at
>> >> all
>> >> about gateway vs. router.  Practically -- and especially with 6LowPAN
>> >> edge
>> >> routers -- most "routers" today actually operate at layer 4.  So we
>> >> found no
>> >> reason to re-implement IP at layer 3.  Instead, we developed a
>> >> completely
>> >> different approach to layer 3 routing, we made sure IPsec could be
>> >> integrated, and we made sure UDP and TCP-based applications could be
>> >> used.
>> >>  Surely, this is not how everyone thinks -- some users will want
>> >> special
>> >> functionality from IP, and for those users 6LoWPAN is the clear answer.
>> >>  But
>> >> for our market, all that matters are UDP, TCP, IPsec, so DASH7 is the
>> >> clear
>> >> answer.
>> >>
>> >> Best regards,
>> >> JP Norair
>> >>
>> >>
>> >> On 26 Jan 2012, at 09:49, Carsten Bormann wrote:
>> >>
>> >> he whole point of running the 6LoWPAN WG for the last half-decade was
>> >> to
>> >> exactly make IPv6 available for constrained node/networks. That may not
>> >> take
>> >> away those constraints, and if you want to sell something else, it may
>> >> make
>> >> sense to proclaim it silly.  Here, specifically, the guy is selling a
>> >> radio
>> >> that is different from 802.15.4, so he's trying to malign 802.15.4,
>> >> striking
>> >> 6LoWPAN in passing.
>> >>
>> >> I would prefer to discuss these things from a technical angle, not by
>> >> pointing to content-free marketing sites/slides.
>> >>
>> >> Grüße, Carsten
>> >>
>> >>
>> >
>
>
_______________________________________________
Lwip mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lwip

Reply via email to