Evolution or revolution

Document Sample
Evolution or revolution Powered By Docstoc
					March 2004                           doc.: IEEE 802.11-04/247r0




             Evolution or revolution?

                     Andrew Myles

                     18 March 2004




Submission               Page 1                     Andrew Myles
March 2004                                           doc.: IEEE 802.11-04/247r0

                             Health warning!



             This presentation contains material that may offend

     However, it is important to at least discuss the issues and
           questions raised in this presentation, now!

     With the appropriate thought we might even uncover new
            ways to ensure 802.11 has a bright future …




Submission                          Page 2                          Andrew Myles
   March 2004                                                    doc.: IEEE 802.11-04/247r0

        802.11 needs to make difficult choices to ensure
          its place as the “802.3 of the wireless world”

                                  The 802.11 standard has been wildly
Situation?
                                    successful in the market place


                              802.11 could be heading towards irrelevance,
Problem?
                                with other groups ready to take the reins


                     Continue              Evolve after fixing
 Options?                                                                “Revolution!”
                     like today             some problems


                802.11 is supplanted         802.11’s life is         802.11 becomes the
 Results?
                in the medium term             extended                802.3 of wireless


                Easy but only a short     Probably need to do        Will ensure long term
Best way?
                   term solution              this anyway                  leadership

   Submission                              Page 3                               Andrew Myles
March 2004                                                                doc.: IEEE 802.11-04/247r0

     The 802.11 standard has been wildly successful in
                     the market place

                           World-wide sales are forecast to rise 1500% from 2000 to 2004
                  40,000                       My Mum discovers WiFi?
                  35,000

                  30,000         Why is 802.11 successful?
                                 • Was it luck?
  Units (000's)




                  25,000
                                 • Was it simplicity?
                  20,000
                                 • Was it something else?
                  15,000

                  10,000

                   5,000

                      0
                               2000        2001         2002       2003           2004

Submission                                          Page 4                               Andrew Myles
March 2004                                        doc.: IEEE 802.11-04/247r0

 The 802.11 standard is heading towards irrelevance,
      with other groups ready to take the reins

                                                   There are probably
  802.11 takes too        It is not clear that
                                                 less than 10 people in
   long to produce          802.11 has an
                                                  the world who really
 poorly written, badly    architecture that is
                                                   “know” the 802.11
  structured & error     suitable for the next
                                                  document, and they
   prone standards             generation
                                                     often disagree!


   The situation is                                  Other groups &
   getting worse as      Document collapse          technologies are
  802.11 introduces         and 802.11’s            ready to replace
    more parallel        irrelevance is the          802.11, and are
 groups & addresses         natural result         already starting to
    niche markets                                        do so!



Submission                      Page 5                           Andrew Myles
March 2004                                                   doc.: IEEE 802.11-04/247r0

  The 802.11-1999 rev-2003 base standard is poorly
written, badly structured and contains too many errors

 • The base standard prose is poorly written
    – It is written by a bunch of engineers, and it shows!
 • The base standard is badly structured
    – You should not have to flip backwards and forwards multiple times to
      find an answer to a simple question
    – And yet one must, often finding 2 or 3 different answers!
 • The base standard contains 100’s, 1,000’s or 10,000’s of errors
    – Depending on how you count
    – Depending on whether or not you include the SDL




Submission                             Page 6                               Andrew Myles
March 2004                                                       doc.: IEEE 802.11-04/247r0

 The recent 802.11 amendments are in an even worse
  state than the 802.11-1999 rev 2003 base standard

 • The amendments are poorly written and reviewed
    – Amendments are subject to same issues as the base standard
    – It is very difficult to write or review a “delta”, particularly in parallel with
      other “deltas” despite the best efforts of the amendment editors and
      many WG members
 • The amendments are based on even more convoluted structures
    – The base standard’s structure was not designed with 11e/g/h/i/j/k/n/…
      in mind
 • The TG’s often make decisions not to touch parts of the base
   standard because it would “open too many cans of worms”, eg
    – e.g. TGh and TGj have avoided many changes to Clause 17
    – e.g.TGk is using 11h in a non-extensible manner, but it is the easy path

Submission                                Page 7                                Andrew Myles
March 2004                                           doc.: IEEE 802.11-04/247r0

       802.11 takes too long to develop amendments,
            especially MAC related amendments
802.11-1997
                                                             MAC
802.11-1999
                                                             PHY
   802.11a
                                                             MAC & PHY
   802.11b
   802.11d
   802.11e
   802.11F
   802.11g
   802.11h
    802.11i          5+ years is
    802.11j        unacceptable
    802.11k
   802.11n

              91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09


Submission                          Page 8                          Andrew Myles
March 2004                                                 doc.: IEEE 802.11-04/247r0

 There are multiple underlying causes for the current
  documentation and timescale problems in 802.11

 802.11 does its work “by                  802.11 uses engineers as
 committee”                                editors*
 • 802.11 tends to use a “design by        • The 802.11 document contains
   committee” process                        complex information
 • Often deliberations are based on        • There are ways of organising and
   poor quality submissions                  presenting complex information
                                             well
 • Even worse, we often waste
   meeting time doing “editing by          • Professional editors know how to
   committee”                                do this, most engineers don’t
 • Unfortunately, the complexities         • Other standards organisations
   and size of the documents we are          have recognised the value of
   writing do not lend themselves to         professional editors
   this style of work
                                           * including the TGh editor
Submission                            Page 9                              Andrew Myles
March 2004                                                           doc.: IEEE 802.11-04/247r0

 There are multiple underlying causes for the current
  documentation and timescale problems in 802.11

 802.11 has too many features                     802.11 has the wrong features
 • We keep throwing new features                  • Most WLAN devices are based
   into 802.11 “just in case”                       on a minimal 802.11 subset that
    – eg PCF in base standard                       solve a real problem
    – eg DLP in 802.11e                                – eg DCF

 • This only results in “protocol                 • They often then have a variety of
   wars”, complexity and delay                      proprietary “added value”
    – eg EDCA vs. HCCA in TGe                       features promoted by vendors
    – eg what is really required for radar             – eg SuperG by Atheros
      detection in TGh                                 – eg CCX by Cisco

 • We often forget that most                           – eg … by …
   “features” will never be used                  • Maybe we should focus on
    – WiFi really defines the “standard             standardising non-core features
      interoperable features”                       after they have proved useful?
Submission                                   Page 10                                Andrew Myles
March 2004                                                          doc.: IEEE 802.11-04/247r0

 There are multiple underlying causes for the current
  documentation and timescale problems in 802.11

 Many 802.11 groups work in                    The letter ballot process in
 parallel                                      802.11 is broken
 • Multiple streams just magnifies             • The size of many documents
   difficulties of writing and                   means they can’t be reviewed
   reviewing deltas to a single base             properly in the available time
   document
                                               • Many documents are not ready
 • There are not enough experts in
                                               • The number of letter ballots just
   some areas to spread around the
                                                 means many are ignored
   groups
                                                    – i.e. people vote “yes” even when it
    – eg the MLME and MIB are virtually
                                                      is quite clear the document is
      useless because very few TG’s
                                                      incomplete or incorrect
      have the skills to do a good job
                                               • Consequently many documents
                                                 reach 75% based on “boredom”
                                                 or “lethargy”, not merit
Submission                                Page 11                                  Andrew Myles
March 2004                                                                                          doc.: IEEE 802.11-04/247r0

   The issues facing 802.11 will only get worse as we
           create more parallel Task Groups
      TG’s with unapproved documents by IEEE-SA

                                                  12
                                                                                                          ???
                                                  11   As of March 2004, there has been very
                                                       little commercial success from the TG’s:          Voice
                                                  10
                                                       • 11b, 11g,                                       WNM
                                                  9
                                                       • 11a slowly                                     WAVE
                                                  8
                                                       • 11i through WPA
                                                                                                          FR
                                                  7
                                                                                            TGk          WEIN
                                                  6
                                                                           TGi              TGj          WPP
                                                  5
                                                                           TGh              TGi         MESH
                                                  4
                                                                           TGg              TGh          TGn
                                                  3
                                                                           TGF              TGg          TGk
                                                  2
                                                          TGb              TGe              TGF           TGj
                                                  1
                                                          TGa              TGd              TGe          TGe?
                                                  0
                                                         Jan 99         Jan 01             Jan 03       Jan 05
Submission                                                                       Page 12                           Andrew Myles
March 2004                                                     doc.: IEEE 802.11-04/247r0

 802.11 architecture may not even be suitable for the
                   next generation

 802.11 is “hamstrung” by                  The future should include a
 separate MAC & PHY                        combined MAC & PHY
 • 802.11 is based on a separate           • A more optimal architecture is
   MAC & PHY                                 based on a MAC & PHY combo
 • The MAC works with a generic                 – MAC is designed based on the
                                                  properties of the PHY & vice versa
   PHY and thus is unable to take
   advantage of the features of any        • Examples in the WMAN space
   particular PHY                            include systems by:
 • The architecture is thus not                 – Flarion
   optimised for a difficult radio              – Arraycomm
   environment                                  –…

                                           • It is only a matter of time before
                                             combined MAC/PHY designs
                                             appear in the WLAN space
Submission                            Page 13                                 Andrew Myles
March 2004                                                    doc.: IEEE 802.11-04/247r0

        Other bodies are either starting to take over
       802.11’s work or would like to supplant 802.11

 WiFi is dangerously close to               Other standards groups want to
 becoming a standards body                  move into the WLAN space
 • WiFi certified WPA before 802.11i        • 802.15 is starting to drift into
   ratification, but at least WPA was         traditional 802.11 territory
   a subset of an early draft
                                            • The proponents of 802.16 are
 • WiFi is probably going to certify          discussing it as a replacement in
   WME before 802.11e is ratified,            the local area for WLANS
   and there are currently areas
                                            • A new research effort is starting
   where WME and 802.11e might
                                              in Europe under the umbrella of
   be different
                                              the 6th Framework
 • WiFi is talking about MBSSID
   architectures that have never
   been reviewed or approved in
   802.11

Submission                             Page 14                               Andrew Myles
March 2004                                           doc.: IEEE 802.11-04/247r0

 We could be depressed but instead lets evaluate the
             options and the solutions


             Option 1             Option 2               Option 3


             Continue        Evolve after fixing
                                                      “Revolution!”
             like today       some problems


      802.11 is supplanted     802.11’s life is    802.11 becomes the
      in the medium term         extended           802.3 of wireless


     Easy but only a short   Probably need to do   Will ensure long term
        term solution            this anyway             leadership




Submission                         Page 15                          Andrew Myles
March 2004                                         doc.: IEEE 802.11-04/247r0

   For 802.11 to continue developing standards like it
         does today is a recipe for irrelevance

 • 802.11 could struggle onwards for a while doing the “same old,
   same old”
 • 802.11 processes and documentation would eventually grind to a
   halt, maybe around the time of the next revision
 • It is possible 802.11 could have defined the “perfect WLAN” by
   then and no further development is needed
 • A more likely scenario is that technology will continue developing
   and market demands will continue expanding
 • In this case, 802.11 will become the 802.5 and not the 802.3 of
   the wireless world



Submission                       Page 16                          Andrew Myles
March 2004                                                 doc.: IEEE 802.11-04/247r0

         Regardless of the long term, 802.11 needs to
          “band-aid” over some of today’s problems

 Quick hits                                • Encourage TG’s to undertake
                                             internal reviews before ballots
 • Put new work into “stand-alone”,
   well written documents, where                – eg TGk
   possible
                                           Harder ideas
 • Demand higher quality
   submissions and require                 • Arrange for professional technical
   submissions before meetings               editors to assist in writing our
                                             standards
 • Avoid inserting immature “slide-
   wear” into the standard                 • Change the rules so that people
                                             who are not interested do not
 • Regularly revise PARs, usually by         distort voting results
   reducing scope, to ensure work is
   useful and clearly defined              Impossible dreams
 • Discourage the creation of too          • Rewrite the current base
   many (?) parallel task groups             standard
Submission                            Page 17                             Andrew Myles
March 2004                                                    doc.: IEEE 802.11-04/247r0

 Revolution will ensure 802.11 becomes the “802.3 of
                   the WLAN world”

 Situation                               Problem
 • We need the “next generation          • 802.11 WG is terrible at
    – ie the 1000baseT of WLANs            determining exact requirements
                                              – Backward compatibility vs.
 • Assumption: 802.11n is only an               coexistence
   interim solution
                                              – Capacity vs. throughput
                                              – etc

                                         • 802.11 WG is terrible at making
                                           compromises
                                              – Any wireless solution will contain
                                                lots of compromises




Submission                          Page 18                                  Andrew Myles
 March 2004                                                       doc.: IEEE 802.11-04/247r0

  Revolution will ensure 802.11 becomes the “802.3 of
                    the WLAN world”


 Decide on very       Send out a CFP                Standardise
                                                                            Let the market
     broad            with a 2-3 year              one or more of
                                                                                decide
  requirements           deadline                   the winners

•Maybe only co-      •Accept that                 •Choose one or         •Its actually pretty
 existence rather     802.11 is not very           more of the            good!
 than backward        good at defining             winners, possibly
 compatibility        large complex                addressing
                      protocols                    different markets
•No need to be too
 exact – the         •Proposers make              •Standardise, with
 responders will      the compromises              WG defining final
 make the                                          5%
                     •Require 95%
 compromises
                      proposals, inc
                      “standardese” &
                      at least
                      prototypes
 Submission                             Page 19                                  Andrew Myles
March 2004                                      doc.: IEEE 802.11-04/247r0

                       What is the “revolution”?


    Process revolution!                       Technology revolution!
 • Industry does the work                     • 802.11 escapes/moves
   of making difficult          A family of     on from the current
   compromises                 wireless LAN     generation
                                standards
 • 802.11 standardises            making      • 802.11 defines the
   something that is            802.11 the      MAC & PHY together,
   mostly complete and is        “802.3 of      for better results
   known to work                wireless”
 • 802.11 does not
   attempt to second
   guess the market




Submission                        Page 20                      Andrew Myles

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:12/17/2011
language:
pages:20