[Berlin-wireless] [Fwd: [fsc04news] WiMax Facts. [Was: Volunteers Use Mesh, WiMax, Wi-Fi in Katrina-Hit Regions]]

[cven]

-------- Original Message --------
Subject: [fsc04news] WiMax Facts. [Was: Volunteers Use Mesh, WiMax, 
Wi-Fi in Katrina-Hit Regions]
Date: Tue, 11 Oct 2005 14:42:02 -0500
From: Sascha Meinrath <sascha at ucimc.org>
Reply-To: fsc04news at freifunk.net
To: fsc04news at freifunk.net
References: <20051011100000.13002.18763.Mailman at taberna.letras.de>

Hi Ken (et al.),

> Date: Tue, 11 Oct 2005 02:53:49 -0400
> From: Ken DiPietro <ken at new-isp.net>
> To: fsc04news at freifunk.net
> Subject: Re: [fsc04news] Volunteers Use Mesh, WiMax, Wi-Fi in Katrina-Hit
>  Regions
> Reply-To: fsc04news at freifunk.net
> 
> Gentlemen,
> 
> I am going to respectfully disagree and try to clear up a few points 
> with this response. This is an attempt to start an educated discussion 
> (which I know both of you are extremely capable of) and I believe this 
> will be a good opportunity for everyone here to watch, participate and 
> learn, including myself.
> 
> With respect to Sascha's comments, I have brought them forward and 
> edited them, hopefully without altering the original context.
> 
> Sascha Meinrath wrote:
> 
> "Yes, the standard is open (at least
> generally speaking), but often the
> networking equipment is proprietary
> -- meaning that devices from one vendor may
> not work with other vendors' systems."
> 
> This is not true, every single WiMAX device must interoperate with every 
> other (in the same frequency, of course) to carry the WiMAX 
> certification label. However, there will be a different set of features 
> that may be manufacturer specific. This is exactly the same as the WiFi 
> standard and as an example I would point out many of the "turbo modes" 
> that several different WiFi manufacturers introduced that would only 
> work within one manufacturer's product line.

With all due respect, what you are stating is simply untrue.  Every 
WiMax device
does _not_ interoperate, one has base stations and client devices (as one
example) -- and the hierarchical structure is required by the network
architecture.  I'm reluctant to get too into this debate, but it's a 
disservice
to propagate incorrect information to folks.

I have been in regular contact with engineers working on 802.16 
standardization
and have read through numerous drafts of various 802.16 documentation 
and follow
this area as part of my professional job as a policy analyst.  I have 
attached
one of the recent drafts ("Part 16: Air Interface for Fixed Broadband 
Wireless
Access Systems Amendment for Improved Coexistence Mechanisms for 
License-Exempt
Operation") so that we can have a primary source to base our discussion on.

> Sascha Meinrath also wrote:
> 
> "Unfortunately, it is unclear exactly what
> WiMax interoperability will look like, or
> how robust a WiMax system actually is
> when used by multiple competing carriers."
> 
> This is also not my understanding. There is a very strict standard that 
> mandates that all approved equipment will interoperate at the link 
> level. There have already been informal compliance tests that show this 
> does work as promised. Full WiMAX certification is in the works now and 
> preliminary results should be available in the next 90 to 120 days for 
> peer review.

There are currently numerous proposals for what this standard will look 
like.
Likewise different facets of the WiMax standard have been under peer 
review for
years.  To further confuse matters, there are numerous WiMax standards 
that are
now obsolete (i.e., 802.16-2001, 802.16a-2003, 802.16c-2002, 
802.16.2-2001), and
it is still unclear what the final standard will look like or when full
standardization will actually be completed.

> Sascha Meinrath then added:
> 
> "The technology itself is very much based
> upon an assumption of an ogopolistic market
> -- it's enshrined in the standard itself
> (for example the time division protocol used
> to "carve up" a single frequency when multiple
> WiMax systems co-exist in the same geographical
> region);
> 
> I'm sorry, I completely missed your point here. Would you please elaborate?

Check out Section 2: "Interference detection and prevention ? general
architecture 2.1 Operational Principles and Policies" (warning, this 
gets very
geeky).  As you can see in Figure 2, the main idea is that access to the
frequency is divided equitably among the available networks.  The sub-frame
structure is a bit more complicated in this, but, in essence, if you have 6
networks, they all get equal time on the network.

> Sascha Meinrath further stated:
> 
> "and the WiMax standard also looks like it will
> _require_ a central radius-type server (e.g., in
> order to coordinate among multiple WiMax systems) "
> 
> The central radius server is for authentication, to keep non-paying 
> customers out. There is additionally a mechanism to coordinate 
> basestations for increased efficiency but this has nothing to do with 
> the radius server. As I understand it, the radius server is optional. 
> I'm curious, how do you propose to keep unauthorized members off of your 
> network. This will certainly become a problem in the larger cities if 
> anyone can purchase a WiFi client and connect.

Again, what you are stating is simply incorrect.  For example, if you 
look at
the initialization process for new WiMax base stations (see Figure 12 on 
page 24
of the IEEE draft I've attached), you will see that the WiMax base stations
themselves have to "phone home" to "Register to the RADIUS server and 
get the IP
address of CIS". They then have to "Query the LE DB through CIS in order 
to get
information of neighbor BSs via Coexistence Protocol (CP) based on its 
location
information" and then "Use BSIDs to learn the IP addresses of neighbors 
from the
RADIUS server."  It is not the case that "this has nothing to do with 
the radius
server" and it is not the case that the radius server is used solely for 
client
access -- this is a draft proposal for radius servers and central 
databases that
are requirements of the WiMax architecture.  If you would like to see this
graphically, check out Figure 14 on page 29 -- you will see that a WiMax 
system
contains a shared database at the base stations, a radius server that
coordinates BSID and IP mapping for the base stations, a "coexistence
identification server", and a regional database -- this is a very 
hierarchical
system architecture.

> At the end of that last statement Sascha said:
> 
> "-- so all ad-hoc networking is out."
> 
> And this is a problem, why? We are discussing a massive network of over 
> a thousand locations, most of which will have a single computer 
> connected to the network. I fail to understand how an AdHoc network 
> would even be useful in this kind of deployment.

I like having the option to do ad-hoc networking -- I'm not saying it's 
optimal
for every situation, but I don't want someone taking away that choice 
either.
AdHoc and hierarchical, WiFi and WiMax are all different tools in the 
community
wireless arsenal -- we should be fighting to maximize our tool chest. 
And yes,
it's quite possible to build a metro-scale ad-hoc network (CUWiN is getting
closer to this technology -- and the FrieFunk/OLSR folks are getting ever
closer) -- we may still be a few years development away from a turn-key
solution, but we'll see it soon (perhaps around the same time WiMax
standardization actually takes place ;).  But seriously, look for 
interesting
developments within the next few months.

> Let me ask you several direct questions.
> 
> How many concurrent VoIP session can you run at a time on your network? 
> As this network will predominately be carrying voice (providing 
> immediate benefit to computer illiterate people) I believe this is a 
> critical point.

Unknown -- depends on whether they're intranet or Internet-connected 
sessions,
how many hops they each are, etc.  We haven't done any systematic 
testing in
this regard -- but I would agree with you, this is a critical facet of any
data-communications network.

> How can you monitor your network for failure, traffic and situations 
> like virus infections?

Yes -- minimally today (basically it requires active monitoring by a 
human); but
with integration of technologies like WiFiDog and automatic bandwidth 
shaping,
we could see major breakthroughs within the next year.  Mostly, this is a
problem of gaining funding to develop/integrate these technologies, not 
because
the tech isn't possible.

> Can you name one MESH network that has over a thousand clients 
> connected? If this solution is cheaper and a better technology why isn't 
> this technology being used in larger networks anywhere in the world? 
> Could it be that scalability is a serious problem? The only large scale 
> attempts at deploying a MESH network of any size have been done using 
> Tropos gear and my impression is that these have been less than a 
> rousing success.

Tropos has multiple mesh networks with 1000+ clients.  I was just 
speaking to
the Tropos CTO yesterday and they're very excited to be doing the Philly 
project
(which should have tens of thousands of concurrent clients at any one 
time).
The technology is being used in multiple different locations worldwide 
-- e.g.,
check out Nortel's Taipei announcement:
http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20050927005387&newsLang=en
and Ferris State University has already rolled out a mesh for its 12,000 
students.

> How many hops can you provide traffic across before speed and latency 
> becomes a serious problem? How many hops can you provide traffic across 
> if all of the intermediate hops are busy servicing local clients?

This is a very complicated issue -- the answer depends on node density,
congestion levels, etc.  We haven't yet done any systematic testing on 
network
capacity -- but mainly that's due to the fact that the bottleneck has 
continued
to be our T-1 connection, not intra-network throughput.

> Please explain to me why "hierarchical (command-and-control) 
> infrastructure development" is a problem. It functions in every single 
> large scale network I am aware of. You state that there are many ways 
> that "WiMax may foster path-dependence." Please elaborate on this, I am 
> fascinated. How can an open standard create such a situation. I am under 
> the impression that the scale of economics will drive the cost of 
> components down to next to nothing. How can this not be good for everyone?

I wasn't so much saying it's a problem as WiMax is highly biased towards
hierarchical command-and-control infrastructures.  Thus, WiMax fosters
path-dependence by eliminating certain technological options from the 
decision
field.  I would recommend reading the attached draft -- it should help 
clarify
why this is so.

Hope this is helpful,

--Sascha

P.S.

> And regarding prices for WiMax equipment will be low... I doubt it because 
> AFAIK you need a license for WiMax therefore it is already only 
> interesting for big companies and therefore the price will be high. \

WiMax was originally a 10-60GHz technology; but in January 2003 an 
amendment was
introduced (and subsequently passed) that lowered the allowable 
frequency down
to 2GHz -- allowing WiMax use in several unlicensed bands.

-- 
Sascha Meinrath
Policy Analyst    *  Project Coordinator  *  President
Free Press       *** CU Wireless Network *** Acorn Active Media
www.freepress.net *  www.cuwireless.net   *  www.acornactivemedia.com

-------------- nächster Teil --------------
Ein Dateianhang mit Binärdaten wurde abgetrennt...
Dateiname   : 80216h-05_017.pdf
Dateityp    : application/pdf
Dateigröße  : 730627 bytes
Beschreibung: nicht verfügbar
URL         : http://lists.olsrexperiment.de/cgi-bin/mailman/private/berlin/attachments/20051014/42c2e37b/attachment.pdf 
-------------- nächster Teil --------------
_______________________________________________
Berlin mailing list
Berlin at olsrexperiment.de
https://olsrexperiment.de/cgi-bin/mailman/listinfo/berlin