First foray into the wireless world, couple of questions...

Throwing out some thoughts/questions, welcoming any comments cause
I'm a major nub on this subject...

I've got an E1505 coming with an Intel 3945 a/b/g card. At least I think
so... in one place Dell says it is a/b/g and in another it just says a/g. So
I'm trying to pick out a wireless router. In addition to the notebook I'll
have a couple of desktops and a Canon MP780 multifunction printer.
Based on what I've read I gather that that, as well as most multifunction
printers, isn't designed to work when hanging off a [wireless] print server
and if you are lucky you might get just basic printing to work. That sound
about right? I don't print often and when I do use the Canon I'm just as
likely to be scanning or faxing, and seeing as how I don't need any storage
on the network I'm thinking that there is no real benefit to getting a wireless
router with USB port and print server functionality. Perhaps I should just
make due with what I have... switching USB cable or printing through a
desktop... and then somewhere down the road purchase a networked
multifunction printer(?).

Anyhoo, so this wireless router (lets say WR) will be in a back bedroom
hanging off my cable modem, and need to support notebook use in other
rooms on just that floor. The straight distance between the WR and the
notebook would be no more than 35'. However, there would be wood
doors and drywall/wood walls in beween the WR and notebook. Worst
case direct path of the signal would travel diagonally through a door and
like 4 to 6 walls. I'm kinda wondering if all that intervening wood/drywall
is gonna be a problem. Repositioning the WR is possible but that would
move the switched ethernet ports away from my desktops.

I've started looking at WRs and one of the ones I like is the WRT54GL.
I think I'd have to switch firmware to get IPv6 support which I might need
down the road, but upgradability is the nice feature of that WR. One thing
that has surprised me is that that router and all the others I've looked at so
far lack a wireless cutoff switch. If you won't be using the wireless network
for awhile it would be a good idea to disable it, right? If you are using a
WR with switched ethernet ports you can't simple kill the power without
taking out the latter, right?

If SSID broadcast is disabled, MAC addresses other than those used by
your notebook(s) are blocked, said notebook(s) are off, and we assume
for a moment that no one has sniffed your MAC addresses and is trying
to access your network, would the WR transmit anything? Is that as
close as you can get to disabling the wireless short of changing a config
setting?

Wow, this got long. A thank you if you made it this far and especially if
you can share any info or tips or whatever.

On Wed, 8 Nov 2006 08:11:06 -0500, "David" <(E-Mail Removed)> wrote in
<(E-Mail Removed)>:

>Throwing out some thoughts/questions, welcoming any comments cause
>I'm a major nub on this subject...
>
>I've got an E1505 coming with an Intel 3945 a/b/g card. At least I think
>so... in one place Dell says it is a/b/g and in another it just says a/g.

Likely typo -- essentially all 'g' products are 'b' as well.

>So
>I'm trying to pick out a wireless router. In addition to the notebook I'll
>have a couple of desktops and a Canon MP780 multifunction printer.
>Based on what I've read I gather that that, as well as most multifunction
>printers, isn't designed to work when hanging off a [wireless] print server
>and if you are lucky you might get just basic printing to work. That sound
>about right?

Yes, although some print servers have better support than others.

>I don't print often and when I do use the Canon I'm just as
>likely to be scanning or faxing, and seeing as how I don't need any storage
>on the network I'm thinking that there is no real benefit to getting a wireless
>router with USB port and print server functionality. Perhaps I should just
>make due with what I have... switching USB cable or printing through a
>desktop... and then somewhere down the road purchase a networked
>multifunction printer(?).

Your choice.

>Anyhoo, so this wireless router (lets say WR) will be in a back bedroom
>hanging off my cable modem, and need to support notebook use in other
>rooms on just that floor. The straight distance between the WR and the
>notebook would be no more than 35'. However, there would be wood
>doors and drywall/wood walls in beween the WR and notebook. Worst
>case direct path of the signal would travel diagonally through a door and
>like 4 to 6 walls. I'm kinda wondering if all that intervening wood/drywall
>is gonna be a problem. Repositioning the WR is possible but that would
>move the switched ethernet ports away from my desktops.

That many walls are likely to be a problem. Consider wired networking
instead: Ethernet, powerline, phoneline, or coax.

>I've started looking at WRs and one of the ones I like is the WRT54GL.
>I think I'd have to switch firmware to get IPv6 support which I might need
>down the road, but upgradability is the nice feature of that WR. One thing
>that has surprised me is that that router and all the others I've looked at so
>far lack a wireless cutoff switch. If you won't be using the wireless network
>for awhile it would be a good idea to disable it, right?

Not really. You'll have ample security with WPA and a strong
passphrase.

>If you are using a
>WR with switched ethernet ports you can't simple kill the power without
>taking out the latter, right?

Right.

>If SSID broadcast is disabled, MAC addresses other than those used by
>your notebook(s) are blocked,

Bad ideas -- won't improve security, and likely to cause problems.

>said notebook(s) are off, and we assume
>for a moment that no one has sniffed your MAC addresses and is trying
>to access your network, would the WR transmit anything? Is that as
>close as you can get to disabling the wireless short of changing a config
>setting?

Just leave it on. Helps to discourage neighbors from setting up on your
channel.

--
Best regards, FAQ for Wireless Internet: <http://Wireless.wikia.com>
John Navas FAQ for Wi-Fi: <http://wireless.wikia.com/wiki/Wi-Fi>
Wi-Fi How To: <http://wireless.wikia.com/wiki/Wi-Fi_HowTo>
Fixes to Wi-Fi Problems: <http://wireless.wikia.com/wiki/Wi-Fi_Fixes>

"David" <(E-Mail Removed)> hath wroth:

>Throwing out some thoughts/questions, welcoming any comments cause
>I'm a major nub on this subject...

Thoughts should be recycled, not thrown out. There may come a day,
probably after global warming, when the world runs out of thoughts.
Conserve now or risk rationing your thoughts in the future.

>I've got an E1505 coming with an Intel 3945 a/b/g card. At least I think
>so... in one place Dell says it is a/b/g and in another it just says a/g.

Look on the bottom of the Dell Inspiron E1505. It should have the
service code. Inscribe the service code into the Dell Support web
page and it will tell you exactly what's inside your specific machine.
It's the same information that was supplied on the manafest that came
with the computer. Hopefully, you didn't throw it out with the
thoughts and questions.

Incidentally, Dell has a new wireless support center at:
| http://www.dell.com/content/topics/g...=us&l=en&s=gen

>So
>I'm trying to pick out a wireless router.

For wireless hardware reviews, see:
| http://www.practicallynetworked.com
| http://www.tomsnetworking.com/wireless_tn/index.html

>In addition to the notebook I'll
>have a couple of desktops and a Canon MP780 multifunction printer.
>Based on what I've read I gather that that, as well as most multifunction
>printers, isn't designed to work when hanging off a [wireless] print server
>and if you are lucky you might get just basic printing to work. That sound
>about right?

Unfortunately true. Multi-malfunction printer/fax/scanner/etc
generally do not work unless the drivers and the print server are
supported by the printer manufacturer.

I think that DLink added a few more supported printers to their
DPR-1260 list. It has the Canon MP730 but not the MP780. That won't
work:
http://www.dlink.com/products/resour...rid=1848&sec=0

>I don't print often and when I do use the Canon I'm just as
>likely to be scanning or faxing, and seeing as how I don't need any storage
>on the network I'm thinking that there is no real benefit to getting a wireless
>router with USB port and print server functionality.

Methinks you should figure out what you're going to be doing before
you spend the money.

>Perhaps I should just
>make due with what I have... switching USB cable or printing through a
>desktop... and then somewhere down the road purchase a networked
>multifunction printer(?).

Or purchase a multifunction printer with either built in wireless
conenctivity, or a built in network print server. USB print servers
are in my opinion problematic.

>Anyhoo, so this wireless router (lets say WR) will be in a back bedroom
>hanging off my cable modem, and need to support notebook use in other
>rooms on just that floor. The straight distance between the WR and the
>notebook would be no more than 35'. However, there would be wood
>doors and drywall/wood walls in beween the WR and notebook. Worst
>case direct path of the signal would travel diagonally through a door and
>like 4 to 6 walls. I'm kinda wondering if all that intervening wood/drywall
>is gonna be a problem. Repositioning the WR is possible but that would
>move the switched ethernet ports away from my desktops.

Yes, it MIGHT be a problem depending on the construction. Wood and
drywall will act as a partial obstruction. These are usually not a
problem. It's the foil backed insulation in the wall that's the
killer. My usual rule of thumb is 1 wall is no problem. 2 walls are
a potential problem. 3 walls will be unreliable or not work. I think
4-6 walls will not work.

Since it appears that your wireless router will be at one end of the
house, and the users at the other, some manner of reflector behind the
wireless router might be useful.
http://www.freeantennas.com
Oh-oh. The site seems to be down. Also try different positions for
the cable modem and wireless router. Keep the antenna high and away
from the tangle of wires. Try relocating into hallways and open
areas. Maybe move it to the middle of the house.

>I've started looking at WRs and one of the ones I like is the WRT54GL.

Incidentally, that's the same as WRT54G v4. Avoid v5, v6, and v7 if
possible. I suggest you install some alternative firmware such as
OpenWRT or DD-WRT. Many more features. DD-WRT emulator:
| http://www.informatione.gmxhome.de/D...nal/index.html

>I think I'd have to switch firmware to get IPv6 support which I might need
>down the road, but upgradability is the nice feature of that WR.

IPv6 is supported on the alternative firmware versions.
| http://www.dd-wrt.com/wiki/index.php/IPv6

>One thing
>that has surprised me is that that router and all the others I've looked at so
>far lack a wireless cutoff switch.

None of the cheapo routers that I know of have a real switch. At
best, they have it in software.
| http://www.linksysdata.com/ui/WRT54G...0/Wireless.htm
Looks like the WRT54GL doesn't even have it in software.

DD-WRT doesn't either. However, you can adjust the xmit power down to
almost zero. See "xmit power" setting at:
| http://www.informatione.gmxhome.de/D..._Advanced.html

>If you won't be using the wireless network
>for awhile it would be a good idea to disable it, right?

Yep. I tell people that pulling the plug is the ultimate security.
Nobody seems to do it for security purposes. The only ones that do
are some of my coffee shop and bar hot spot customers, that don't want
the wireless active during peak business hours, or at night when
nobody is around.

>If you are using a
>WR with switched ethernet ports you can't simple kill the power without
>taking out the latter, right?

Nope. If you have seperate boxes for the router and wireless access
point, you can do that. The ethernet switch goes with the router. Of
course, you could turn off the whole thing.

>If SSID broadcast is disabled, MAC addresses other than those used by
>your notebook(s) are blocked, said notebook(s) are off, and we assume
>for a moment that no one has sniffed your MAC addresses and is trying
>to access your network, would the WR transmit anything?

Please do not bother with SSID broadcasting, MAC address filters, and
other security band-aids. You're basic protection is the WPA or WPA2
encryption. If that works and your WPA pass phrase is secure, you're
done with security. If not, all the band-aids suggested will create a
few obstacles but not impediments to intrusion. See the FAQ at:
http://wireless.wikia.com/wiki/Wi-Fi#Wi-Fi_Security

However, if you disable broadcasting, and fill the WPA pass phrase
with garbage, the wireless will be effectively unuseable which is
probably what you want.

>Is that as
>close as you can get to disabling the wireless short of changing a config
>setting?

Yep.

>Wow, this got long. A thank you if you made it this far and especially if
>you can share any info or tips or whatever.

http://en.wikipedia.org/wiki/Wrt54g

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

David wrote:
>
> Anyhoo, so this wireless router (lets say WR) will be in a back bedroom
> hanging off my cable modem, and need to support notebook use in other
> rooms on just that floor. The straight distance between the WR and the
> notebook would be no more than 35'. However, there would be wood
> doors and drywall/wood walls in beween the WR and notebook. Worst
> case direct path of the signal would travel diagonally through a door and
> like 4 to 6 walls. I'm kinda wondering if all that intervening
> wood/drywall
> is gonna be a problem. Repositioning the WR is possible but that would
> move the switched ethernet ports away from my desktops.

I have a similar number of walls and have no problems. However, the
layout of the walls and their relation to hallways and rooms may make
your situation worse. In my worst case I have 5 walls and a door
between the router and the client machine if you draw a straight line.
Most likely there isn't much signal going down that path. Instead I
probably get most of the signal from reflections going down the hall
after making a right turn in the kitchen. Using an 802.11b router I
still get 4Mbps transfer rate despite the walls and 40' distance. As
always though, your mileage may vary.


> I've started looking at WRs and one of the ones I like is the WRT54GL.
> I think I'd have to switch firmware to get IPv6 support which I might need
> down the road, but upgradability is the nice feature of that WR.

Also look into the Buffalo WHR-G54S. They are cheaper than the WRT54GL,
but have essentially the same hardware ($39 at newegg.com). Load up the
DD-WRT firmware and you'll have your IPv6 support if you need it.

> One thing
> that has surprised me is that that router and all the others I've looked
> at so
> far lack a wireless cutoff switch. If you won't be using the wireless
> network
> for awhile it would be a good idea to disable it, right? If you are
> using a
> WR with switched ethernet ports you can't simple kill the power without
> taking out the latter, right?
>
> If SSID broadcast is disabled, MAC addresses other than those used by
> your notebook(s) are blocked, said notebook(s) are off, and we assume
> for a moment that no one has sniffed your MAC addresses and is trying
> to access your network, would the WR transmit anything? Is that as
> close as you can get to disabling the wireless short of changing a config
> setting?

I would suggest just setting up good security and forget about turning
off the wireless (MAC filtering and SSID hiding don't count as good
security -- WPA with a good pass phrase does). If you really want to do
this, get a simple wired router and use the wireless router as an plain
access point. Then you can kill the wireless router and the wired
router will still be handling your wired clients.

>
> Wow, this got long. A thank you if you made it this far and especially if
> you can share any info or tips or whatever.
>

On Wed, 08 Nov 2006 10:37:35 -0700, Bryant Smith
<(E-Mail Removed)> wrote:

>I have a similar number of walls and have no problems. However, the
>layout of the walls and their relation to hallways and rooms may make
>your situation worse. In my worst case I have 5 walls and a door
>between the router and the client machine if you draw a straight line.

Many years ago, I was running some tests on the number of walls that
can be penetrated by a clients products. The signal would rapidly
decrease linearly with the addition of wall until about 3 walls. After
that, it would remain almost constant. I eventually found 8 walls (in
a large office building) and the signal was still there. I did some
crude direction finding and found that it was going out a window,
bouncing off the office building next door, and coming back through
another window. When I did the same test diagonally across the office
building floor, the signal did the expected linear fade into oblivion.

I once had to figure out how to distribute Wi-Fi in a hospital that
was terminally paranoid about RF, antennas, drilling holes, wiring,
and such. I had fairly good success using the HVAC ducting as a
waveguide. It worked fairly well until the maintenance group
discovered what I was doing and ended he experiment. I've often
suspected that miraculous indoor propagation is actually waveguide
effects through the ducting.
--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558 (E-Mail Removed)
# http://802.11junk.com (E-Mail Removed)
# http://www.LearnByDestroying.com AE6KS

"David" <(E-Mail Removed)> wrote in message news:(E-Mail Removed). ..

-snip-

Thanks for the earlier replies. I read them all and decided to give the
WRT54GL a try, reporting back on what I found. I'm pleased to say
that the E1505/Intel 3945 and WRT54GL/stock firmware combo was
up and running quickly. I think I have all the security bases covered
and then some, sanity check welcome...

- Changed default wireless router IP & DHCP starting address
- Max length, strong SSID, went ahead and disabled broadcasts since it
was easy
- WPA2 Personal, TKIP+AES, max length strong WPA shared key
- Went ahead and enabled MAC filtering
- AP Isolation on
- Blocking anon net requests, filtering multicast and idents
- Max length, strong admin username & password
- Admin access via HTTPS, remote management off, wireless access on
for convenience at least for now
- No holes opened for local network servers

The Intel wireless software reports 4 or 5 bars out of 5 for signal strength
regardless of where I am and so far I haven't had any connection problems
that I'm aware of. Speed tests via wireless are coming in around 7Mbps
no matter where I am. Which is about half of what I get through the router
via hardwired LAN. The wireless performance sounds low to me but I
need to do some more research into that.

On Wed, 15 Nov 2006 00:21:08 -0500, "David" <(E-Mail Removed)> wrote in
<(E-Mail Removed)>:

>"David" <(E-Mail Removed)> wrote in message news:(E-Mail Removed). ..
>
>-snip-
>
>Thanks for the earlier replies. I read them all and decided to give the
>WRT54GL a try, reporting back on what I found. I'm pleased to say
>that the E1505/Intel 3945 and WRT54GL/stock firmware combo was
>up and running quickly. I think I have all the security bases covered
>and then some, sanity check welcome...
>
>- Changed default wireless router IP & DHCP starting address

No need or point.

>- Max length, strong SSID, went ahead and disabled broadcasts since it
> was easy

Bad idea. Turn SSID back on. Just make it unique. Length is
irrelevant. There's no (zip, zilch, nada) "strength" in the SSID.

>- WPA2 Personal, TKIP+AES, max length strong WPA shared key

Good. Enterprise would be better.

>- Went ahead and enabled MAC filtering

Bad idea. Won't do any real good, likely to cause problems.

>- AP Isolation on

Good.

>- Blocking anon net requests, filtering multicast and idents

OK.

>- Max length, strong admin username & password

Good.

>- Admin access via HTTPS, remote management off, wireless access on
> for convenience at least for now

OK.

>- No holes opened for local network servers

Good.

>The Intel wireless software reports 4 or 5 bars out of 5 for signal strength
>regardless of where I am and so far I haven't had any connection problems
>that I'm aware of.

The bars are only truly meaningful when the network is under load.

>Speed tests via wireless are coming in around 7Mbps
>no matter where I am. Which is about half of what I get through the router
>via hardwired LAN. The wireless performance sounds low to me but I
>need to do some more research into that.

That is low. With good signal you should be getting about 22 Mbps
wireless to wired, or about 11 Mbps wireless to wireless.

--
Best regards, FAQ for Wireless Internet: <http://Wireless.wikia.com>
John Navas FAQ for Wi-Fi: <http://wireless.wikia.com/wiki/Wi-Fi>
Wi-Fi How To: <http://wireless.wikia.com/wiki/Wi-Fi_HowTo>
Fixes to Wi-Fi Problems: <http://wireless.wikia.com/wiki/Wi-Fi_Fixes>

"John Navas" <(E-Mail Removed)> wrote in message news:(E-Mail Removed)...
> On Wed, 15 Nov 2006 00:21:08 -0500, "David" <(E-Mail Removed)> wrote in
> <(E-Mail Removed)>:

>>- Max length, strong SSID, went ahead and disabled broadcasts since it
>> was easy
>
> Bad idea. Turn SSID back on. Just make it unique. Length is
> irrelevant. There's no (zip, zilch, nada) "strength" in the SSID.

>>- Went ahead and enabled MAC filtering
>
> Bad idea. Won't do any real good, likely to cause problems.

I've tried to read up on these and AFAICT both present low hurdles.
I'm not relying on them, and the way I see it a hurdle is a hurdle and
I'll take it even if it is low. The Intel software remembers the SSID
for me and I only have one MAC address to whitelist so I'm not seeing
an issue WRT inconvenience. Taking this into consideration, why do
you say they are a bad idea?

>>Speed tests via wireless are coming in around 7Mbps
>>no matter where I am. Which is about half of what I get through the router
>>via hardwired LAN. The wireless performance sounds low to me but I
>>need to do some more research into that.
>
> That is low. With good signal you should be getting about 22 Mbps
> wireless to wired, or about 11 Mbps wireless to wireless.

Well after netstumbling for any problems and updating software and
tweaking everything I could find I still get substantially lower test results
over wireless vs hardwire on the Speak Easy Speed Tests. As a sanity
check I used iperf to check desktop<->notebook bandwidth and it
reports a little over 25Mbps. Maybe after some sleep I'll finish getting
to the bottom of this <yawn>.

On Wed, 15 Nov 2006 04:21:20 -0500, "David" <(E-Mail Removed)> wrote in
<(E-Mail Removed)>:

>"John Navas" <(E-Mail Removed)> wrote in message news:(E-Mail Removed)...
>> On Wed, 15 Nov 2006 00:21:08 -0500, "David" <(E-Mail Removed)> wrote in
>> <(E-Mail Removed)>:
>
>>>- Max length, strong SSID, went ahead and disabled broadcasts since it
>>> was easy
>>
>> Bad idea. Turn SSID back on. Just make it unique. Length is
>> irrelevant. There's no (zip, zilch, nada) "strength" in the SSID.
>
>>>- Went ahead and enabled MAC filtering
>>
>> Bad idea. Won't do any real good, likely to cause problems.
>
>I've tried to read up on these and AFAICT both present low hurdles.
>I'm not relying on them, and the way I see it a hurdle is a hurdle and
>I'll take it even if it is low. The Intel software remembers the SSID
>for me and I only have one MAC address to whitelist so I'm not seeing
>an issue WRT inconvenience. Taking this into consideration, why do
>you say they are a bad idea?

<http://wireless.wikia.com/wiki/Wi-Fi#Wi-Fi_Security_Myths>

THE SIX DUMBEST WAYS TO SECURE A WIRELESS LAN
<http://blogs.zdnet.com/Ou/index.php?p=43>
(Wireless LAN security hall of shame)

MAC filtering: This is like handing a security guard a pad of paper
with a list of names. Then when someone comes up to the door and
wants entry, the security guard looks at the person’s name tag and
compares it to his list of names and determines whether to open the
door or not. Do you see a problem here? All someone needs to do is
watch an authorized person go in and forge a name tag with that
person’s name. The comparison to a wireless LAN here is that the
name tag is the MAC address. The MAC address is just a 12 digit long
HEX number that can be viewed in clear text with a sniffer. A
sniffer to a hacker is like a hammer to a carpenter except the
sniffer is free. Once the MAC address is seen in the clear, it takes
about 10 seconds to cut-paste a legitimate MAC address in to the
wireless Ethernet adapter settings and the whole scheme is defeated.
MAC filtering is absolutely worthless since it is one of the easiest
schemes to attack. The shocking thing is that so many large
organizations still waste the time to implement these things. The
bottom line is, MAC filtering takes the most effort to manage with
zero ROI (return on investment) in terms of security gain.

The downside of MAC filtering is that it often results in mysterious
problems that waste lots of time to troubleshoot and fix. With no real
upside, and a significant potential downside, it just doesn't make
sense. Think cost:benefit ratio.

SSID hiding: There is no such thing as "SSID hiding". You’re only
hiding SSID beaconing on the Access Point. There are 4 other
mechanisms that also broadcast the SSID over the 2.4 or 5 GHz
spectrum. The 4 mechanisms are; probe requests, probe responses,
association requests, and re-association requests. Essentially,
youre talking about hiding 1 of 5 SSID broadcast mechanisms. Nothing
is hidden and all youve achieved is cause problems for Wi-Fi roaming
when a client jumps from AP to AP. Hidden SSIDs also makes wireless
LANs less user friendly. You dont need to take my word for it. Just
ask Robert Moskowitz who is the Senior Technical Director of ICSA
Labs in his white paper Debunking the myth of SSID hiding.

The downsides of SSID hiding are that it (a) makes it more likely that a
neighbor will set up on the same channel as you, resulting in
interference that can make your Wi-Fi problematic, and (b) can cause
mysterious dropouts with products and/or drivers that don't handle it
well. Again, with no real upside, and a significant potential downside,
it just doesn't make sense. Cost:benefit ratio.

>>>Speed tests via wireless are coming in around 7Mbps
>>>no matter where I am. Which is about half of what I get through the router
>>>via hardwired LAN. The wireless performance sounds low to me but I
>>>need to do some more research into that.
>>
>> That is low. With good signal you should be getting about 22 Mbps
>> wireless to wired, or about 11 Mbps wireless to wireless.
>
>Well after netstumbling for any problems and updating software and
>tweaking everything I could find I still get substantially lower test results
>over wireless vs hardwire on the Speak Easy Speed Tests. As a sanity
>check I used iperf to check desktop<->notebook bandwidth and it
>reports a little over 25Mbps. Maybe after some sleep I'll finish getting
>to the bottom of this <yawn>.

A likely cause of your problem is RF interference. (Check your error
rate.) Perhaps a neighbor is using the same channel and hiding the
SSID, so you haven't noticed.

--
Best regards, FAQ for Wireless Internet: <http://Wireless.wikia.com>
John Navas FAQ for Wi-Fi: <http://wireless.wikia.com/wiki/Wi-Fi>
Wi-Fi How To: <http://wireless.wikia.com/wiki/Wi-Fi_HowTo>
Fixes to Wi-Fi Problems: <http://wireless.wikia.com/wiki/Wi-Fi_Fixes>

"John Navas" <(E-Mail Removed)> wrote in message news:(E-Mail Removed)...

> A likely cause of your problem is RF interference. (Check your error
> rate.) Perhaps a neighbor is using the same channel and hiding the
> SSID, so you haven't noticed.

Both the Intel software and Netstumbler report the presence of my own
wireless router when it is using SSID hiding so I wonder if the latter is a
possibility. According to NetStumbler all the other devices around here
(one or two of which are on my channel) are coming in at -85 dBm or
worse while mine is around -17 dBM when testing here in my office.
In Intel Advanced Statistics->Statistics signal strength is -18dBm, the
percent transmit errors stat is 0, and nothing else seems to be reflecting
any wireless problems.

FWIW, here is a summary of the Speakeasy download speed test results.
The first is baseline (just through my Comcast cable modem), the next two
include the Wireless Router in the path:

Notebook<-- CM -->Speakeasy (16Mbps)
Notebook<--switched-->WR<-- CM -->Speakeasy (16 Mbps)
Notebook<--wireless-->WR<-- CM -->Speakeasy (7 Mbps)

Next, I used iperf to exercise all paths through the router and eliminated
my cable modem and the Internet from the picture. In these tests I used
my notebook and desktop computers, with both being directly connected
to the specified ports on the router:

Notebook<--wireless-->WR<--uplink-->Desktop (19.6 Mbps)
Notebook<--wireless-->WR<--switched-->Desktop (25 Mbps)
Notebook<--switched-->WR<--uplink-->Desktop (40 Mbps)
Notebook<--switched-->WR<--switched-->Desktop (93.5 Mbps)

Assuming 25Mbps over TCP over a non-accelerated G only wireless
connection is good, I'm inclined to think that my wireless connection is
good and RF interference isn't really a problem. I've played with every
Intel 3945 and WRT54GL setting I could think of and that didn't have
any noticeable impact on performance in any of the above scenarios.
I don't know what is going on, but one thing jumps out and that is the
reduced throughput when traffic is moving through the uplink port.