Questions about multiple HSMM-MESH nodes

Peter Dalinis peter at dalinis.com
Tue Jul 20 12:16:14 PDT 2010


I will.  Looks very cool.

Works great as far as budget goes.

I am located in Greenwood with great LOS south and east.


On Tue, Jul 20, 2010 at 11:34 AM, Bill V WA7NWP <wa7nwp at gmail.com> wrote:

> Here's a good post on the HSMM-MESH project.   While some of their
> "amateur only" concepts are a bit limiting for what I'd eventually
> like to do, we can't help be admire their energy and efforts.  It's
> probably worth it to flash a node or two with HSMM-MESH and give it a
> go.   Anybody else want to try?
>
> 73,
> Bill - WA7NWP
>
> ---------- Forwarded message ----------
> From: Glenn Currie <kd5mfw_7 at yahoo.com>
> Date: Tue, Jul 20, 2010 at 11:07 AM
> Subject: Re: [ntms-hsmm] Re:Questions about multiple HSMM-MESH nodes
> To: ntms-hsmm at yahoogroups.com
>
>
>
>
> Thanks for all the info and interest.  Here is some info to consider.
> I hope it is helpful.
>
> Please get some of this gear and get out in the field.  It is fun and
> there is a HUGE amount to learn and share with others!
>
> I have found that mixed antennas, linking different nodes does work.
> I am also sure the design application notes provided, contain useful
> information.  Most of the things we are doing with the WRT54G are not
> what it was designed to do and nobody provides detailed information
> for our applications.
>
> Broadcom, who makes the main chip inside the WRT54G does not make
> detailed information available to users, just major vendors.  The
> vendors provide information for the products they design, and how they
> intend for them to be used.
>
> --
>
> We have HSMM-MESH(tm) firmware stable and have released it for free
> for other hams to experiment with and report their findings.  Some
> hams are already using HSMM-MESH(tm) in ways the developers had not
> expected or tested!  The system is wonderfully (confusingly) flexible.
>
> So we have a situation were the vendors supply some interesting
> products but very limited information.  We have little choice but to
> experiment and share findings.
>
> Distance, throughput, environmental conditions vary widely and we are
> not using the equipment in a living room or office for the most part,
> so we are on our own.  This is radio and the analog physics apply -
> and they are complex if you want to get deep into the details.  We
> cannot expect maximum throughput in all cases.  There are always trade
> offs and due to the complexity of the radio, not all the trade offs
> are immediately apparent, I assure you.
>
> Also, the HSMM-MESH(tm) developers main interest is quickly deployed
> emergency communications.  If we can get a link that is usable, it
> will beat a 1200 baud packet link in speed.  So we declare victory in
> such cases and move on.  We do not have definitive information for
> every possible use of the firmware and hardware.  That is one of the
> main reasons we released the firmware - to get help in discovering
> what will and will not work.
>
> No system can be optimized for all possible cases.  So if you trade
> off some throughput for a particular configuration, does that work
> well enough for your purposes?  If so declare victory.  If not, do
> some more field work.  Try placing multiple nodes in various locations
> and work towards an acceptable solution.  There are usually multiple
> ways any particular site can  be configured.  Your first idea might
> not be what ends up working best.  The radios can be set to do antenna
> diversity, and they do so many things to maintain a link it would fill
> a book to try to describe.  (See the book by Gast on 802.11 - it is a
> thick book for a good reason - lots and lots of details - tons of
> them).  We all value simple answers and rules of thumb, but these
> radios and antenna systems are complex by their nature.  You should
> not expect to know everything about how to use them without some
> effort.  Understand the basics then try some field work when the
> thick books make you sleepy.  You can be safe and not not bother other
> people easily enough.  But digital microwave broadband communications
> is not a simple topic.
>
> So what is going on with the radio is very complex, self adapting, not
> documented in detail by the maker, and would require a lab full of
> very expensive test gear to fully track and understand.  Few hams have
> all the test gear to fully instrument a WRT54G and even attempt to
> fully describe everything that it is doing in a particular
> application.  So we trade information and see what works and what does
> not work.
>
> Unmatched antennas on a mesh node do work, and I am confident that in
> most, possibly all cases it cuts throughput.   But if you do not have
> enough gain and directivity on a link to make the trip, your
> throughput is zero.  So I see no conflict in the information, just
> more pieces of a very  complex analog puzzle.  One should not expect
> maxim throughput in all cases in all places. All we can do is to share
> our experiences and figure out what will work for a particular
> situation.
>
> I have been working with the WRT54G and similar device for several
> years.  I have a pretty good feel for what works consistently around
> where I live in the Austin area.  I wish I had all the test gear to do
> definitive testing and measuring of everything.  That is not going to
> happen.   I really encourage you to do field work as I have found a
> lot of what works and does not work, is counter intuitive.  The theory
> is great, but without tons of technical information on the hardware,
> you really have to experiment.  You cannot sit at a desk and fully
> design and model a system then just go set it up and expect it to work
> every time.  We simply do not have enough information to feed into a
> model to do that.  You would need a detailed site survey, for one
> thing, and somebody has to go out in the field to do that work for any
> model to be of use.
>
> I am happy to share what I have learned in about 8 years of field
> work.  I cannot stress enough that some of how the radios consistently
> work, seems counter intuitive, at least to me.  So if I had not done
> the field work myself, and just did thought experiments, I would have
> been dead  wrong on a number of important issues.  So collect all the
> good info you can, work out a plan and give it a try.  Sometimes it
> just works, sometimes it does not.  I have found more links worked
> than I would have guessed when I first started.
>
> 73's
>
> -Glenn
>
> Examples:
>
> The 2.4 GHz band is full.  Yes it is and we still consistently, at
> will do 10 mile links across Austin.  Don't sit on the couch and think
> that it is not worth trying.
>
> Slap on a big RF amp before you even try a link, it has to make things
> better.
> Nope.  We do our 10 mile links with simply good antennas on stock
> WRT54G hardware.
> You MUST have good sites.  If you do not have Line of Sight, then an
> amp is not usually going to do any good unless you can burn a hole in
> obstructions to obtain a line of sight path.
>
> Not all locations will work.  There can be bad interference that
> causes problems.  This is true no mater how expensive your equipment
> or how much you would like the link to work.
>
> So the make your best guess, collect your gear and get out in the
> field and try stuff.
>
> --- On Tue, 7/20/10, Kipton Moravec <kip at kdream.com> wrote:
>
> From: Kipton Moravec <kip at kdream.com>
> Subject: Re: [ntms-hsmm] Re:Questions about multiple HSMM-MESH nodes
>
> To: ntms-hsmm at yahoogroups.com
> Date: Tuesday, July 20, 2010, 10:09 AM
>
>
>
> In practice the guys in Austin have not had any problems with using
>
> different antennas on the two ports of a WRT54G. I questioned them on
>
> this also, and they assure me that in practice it works.
>
> A lot of times they use a omni for local mesh connections and a
>
> directional (most often horizontal polarity) for long distance. I have
>
> not tested this but believe the group in Austin know what they are
>
> talking about. It does work and they have not seen any bad effects.
>
> Kip
>
> On Tue, 2010-07-20 at 14:05 +0000, Steve W3AHL wrote:
>
> > The dual antenna ports are intended only for spatial diversity of two
> similar antennas mounted 1-4 wavelengths apart (5-20" at 2.4 GHz)covering
> the same area. The diversity logic selects the antenna with the best signal
> quality (lowest group delay) based upon multipath interactions.
>
> >
>
> > Using antennas with different characteristics to cover different areas
> may seem to work in initial testing, but will result in unusable links when
> the traffic picks up a little.
>
> >
>
> > Refer to:
> http://www.cisco.com/en/US/tech/tk722/tk809/technologies_tech_note09186a008019f646.shtml
>
> >
>
> > The golf course example toward the end addresses why your proposal
> doesn't work.
>
> >
>
> > Steve, W3AHL
>
> >
>
> > --- In ntms-hsmm at yahoogroups.com, "Gerald Pelnar" <wd0fyf at ...> wrote:
>
> > >
>
> > > For nodes that didn't have a good signal between omni's, I planned to
> use 2
>
> > > antennas on a node. An omni for local coverage and a directional to get
> a
>
> > > better signal to another node. As far as I know there is no reason not
> to
>
> > > have 2 antennas on a node. With diversity RX, the radio should transmit
> any
>
> > > replies on the antenna it received the request on.
>
> > >
>
> > > Gerald Pelnar WD0FYF
>
> > >
>
> > >
>
> > > ----- Original Message -----
>
> > > From: "WILLIAM WALLACE" <bill at ...>
>
> > > To: <ntms-hsmm at yahoogroups.com>
>
> > > Sent: Monday, July 19, 2010 8:26 AM
>
> > > Subject: [ntms-hsmm] Re:Questions about multiple HSMM-MESH nodes
>
> > >
>
> > >
>
> > > >I have the exact same question. I guess as we try to plan a network
> for
>
> > > > this function we run into a situation where we have a need for a
> relay
>
> > > > point
>
> > > > that needs more than an omni antenna to tie our other locations
> together.
>
> > > >
>
> > > > Bill Wallace KC0TGY
> _______________________________________________
> Seatcp mailing list
> Seatcp at wetnet.net
> http://wetnet.net/mailman/listinfo/seatcp
>


More information about the Seatcp mailing list