802.11 b/g compatibility?

Archived from groups: alt.internet.wireless (More info?)

"Smartin" <not@here.ok> wrote in message
news:2nvhr053u2n4asbfouekjdlb36ikf0pvet@4ax.com...
> Hi wireless fans... long time listener, first time caller...
>
> We've enjoyed a nice wireless LAN with a Linksys BEFW11S4 (firmware v
> 1.45z, an 802.11b router) providing a gateway to cable internet,
> serving a few laptop PC's:
> (1) Windows XP home, internal Broadcom 802.11g adapter
> (2) Windows XP home, external Linksys WPC11 (802.11b) adapter
> (3) Windows XP pro, internal LanExpress 802.11b adapter
> (4) Windows XP home, internal 802.11g adapter
> ... and a couple wired clients
>
> The router is configured for MAC address filtering, but no WEP.
>
> Everything was groovy for almost a year. Typical throughput was on the
> order of 1200 bps on all clients. Recently, pc (1) started getting

I think you meant 1200 Kbps (1.2Mbps), right? That would be an aggregate
throughput of of 4.8 Mbps, about what you'd expect for 802.11b.

> terrible throughput, say 120 kbps, rather spontaneously. I tried the
> usual suspects: reset the router, reboot PC's, reinstall drivers on pc
> (1), reset TCP/IP, all to no avail. All the other PC's meanwhile were
> fine. Digging deeper, I did a "repair" install of Windows on (1). No
> joy.

If you haven't changed the configuration on the router or any of the
clients, then something has to have changed in the operating environment -
host hardware problems, host processor overload, some other network
interfering, something new between the client and the router that absorbs
the signal, etc.

Did you check the site survey to see if any new networks are showing up? Did
you try changing the network's center channel? Have you recently bought a
2.4 Ghz cordless phone or some other device that uses 2.4 Ghz (like a video
sender, for example)?

>
> Since PC (1)'s throughput's rate was killed so suddenly I suspected
> either some crazy software was installed, the internal adapter on (1)
> was going south, or there was something screwed with TCP/IP. I could
> not find any suspicious software on (1), so I installed (and plugged
> in) the Linksys external "b" adapter in (1). It worked great: 1200
> bps.

Your instinct is right, that something environmental changed.

>
> Thinking this confirmed my suspicions about the internal adapter on
> (1), I bought a new external Linksys g adapter for (1). To my
> amazement, not only did the throughput on (1) NOT improve with this
> new adapter, installing the same adapter on PC's (2) and (3) also
> brought them down to pitiful throughputs.
>
> So my conclusion is, this Linksys BEFW11S4 (802.11b) is not really
> compatible with some 801.11g adapters, including some Linksys wireless
> adapters.

This conclusion is incompatible with the fact the the Broadcom G adapter
worked fine for a year. The fact that the Linkys B adapter worked but the
Linksys G did not proves nothing. Those two adapters probably differ in more
ways than just B vs. G. They may have been made by different vendors (resold
by Linksys), using different antennas, and may have different receiver
sensitivities. It sounds to me like you have some new source of interference
that pushed the old Broadcom card over the edge, but your old Linksys B card
happens to be able to tolerate it.

>
> I've read some other posts, mostly old, that poo-poo the idea of
> incompatibility between b and g equipment. Any ideas here? I will
> probably acquire a "g" WAP in the near future -- in part to test my
> theory, and in part to satiate PC (1) who has terrible throughput at
> the moment...

Let's define compatibility. According to the standards, it means that a G
access point supports a mix of B and G clients, by design. It does *not*
mean there is no performance penalty. There is always a penalty. If the B
clients are very quiet, the G clients can operate at full G bitrates, with
only a little throughput penalty for certain messages used to co-ordinate
between the different client types. If the B clients generate even moderate
traffic, all clients will be dragged down to B bitrates.

If your access point is configured to support a B network, then *all*
clients will connect as B clients, whether they are capable of G or not. A B
access point cannot support or even recognize a G client. The G client
recognizes B, and switches to that mode when associating with the access
point.

If you buy a G router, and host 1 goes back to full throughput, it still
doesn't prove the theory that G is incompatible with B. It probably just
proves that your new router has a higher transmit power, better antennas, or
both.

Archived from groups: alt.internet.wireless (More info?)

On Sat, 11 Dec 2004 05:15:34 GMT, "gary" <pleasenospam@sbcglobal.net>
wrote:

>"Smartin" <not@here.ok> wrote in message
>news:2nvhr053u2n4asbfouekjdlb36ikf0pvet@4ax.com...
>> Hi wireless fans... long time listener, first time caller...
>>
>> We've enjoyed a nice wireless LAN with a Linksys BEFW11S4 (firmware v
>> 1.45z, an 802.11b router) providing a gateway to cable internet,
>> serving a few laptop PC's:
>> (1) Windows XP home, internal Broadcom 802.11g adapter
>> (2) Windows XP home, external Linksys WPC11 (802.11b) adapter
>> (3) Windows XP pro, internal LanExpress 802.11b adapter
>> (4) Windows XP home, internal 802.11g adapter
>> ... and a couple wired clients
>>
>> The router is configured for MAC address filtering, but no WEP.
>>
>> Everything was groovy for almost a year. Typical throughput was on the
>> order of 1200 bps on all clients. Recently, pc (1) started getting
>
>I think you meant 1200 Kbps (1.2Mbps), right? That would be an aggregate
>throughput of of 4.8 Mbps, about what you'd expect for 802.11b.

Well duh me... what's a factor of 1000 between friends?

>> terrible throughput, say 120 kbps, rather spontaneously. I tried the
>> usual suspects: reset the router, reboot PC's, reinstall drivers on pc
>> (1), reset TCP/IP, all to no avail. All the other PC's meanwhile were
>> fine. Digging deeper, I did a "repair" install of Windows on (1). No
>> joy.
>
>If you haven't changed the configuration on the router or any of the
>clients, then something has to have changed in the operating environment -
>host hardware problems, host processor overload, some other network
>interfering, something new between the client and the router that absorbs
>the signal, etc.
>
>Did you check the site survey to see if any new networks are showing up? Did
>you try changing the network's center channel? Have you recently bought a
>2.4 Ghz cordless phone or some other device that uses 2.4 Ghz (like a video
>sender, for example)?

There is no new equipment and no changes on my network. Nothing new as
far as electronics in the house (unless you count Christmas tree
lights? ... but those were installed after the problems began (^: )

Interesting question about new networks... PC(1) does not see any new
networks. However PC(4) routinely sees 2 or 3 foreign networks, and
occasionally prefers to connect to one of them even though my AP is
certainly the closest. Both (1) and (4) have internal G adapters, and
my AP is strictly B so I think I can understand (4)'s preference,
however (1) does not see the same opportunities.

The other thing I neglected to mention about (1) is its curious ping
results. Even when (1)'s throughput was normal, a typical ping (say,
to www.google.com) would show 25-50% dropouts. The successful packets
had times consistent with other machines on this network. Same is true
pinging any machine on the LAN. Pinging the loopback address is always
100% successful. Now that the throughout is poor, pings have not
really changed: still 25-50% dropouts pinging any other host, and
still normal timing on returned packets. This profile is also observed
when (1) uses the external B adapter. Since the ping behavior does not
vary with connection quality or method, it seems doubtful the high
dropout rate and poor throughput share the same cause, but it seems
unusual enough to warrant a mention.

>> Since PC (1)'s throughput's rate was killed so suddenly I suspected
>> either some crazy software was installed, the internal adapter on (1)
>> was going south, or there was something screwed with TCP/IP. I could
>> not find any suspicious software on (1), so I installed (and plugged
>> in) the Linksys external "b" adapter in (1). It worked great: 1200
>> bps.
>
>Your instinct is right, that something environmental changed.
>
>>
>> Thinking this confirmed my suspicions about the internal adapter on
>> (1), I bought a new external Linksys g adapter for (1). To my
>> amazement, not only did the throughput on (1) NOT improve with this
>> new adapter, installing the same adapter on PC's (2) and (3) also
>> brought them down to pitiful throughputs.
>>
>> So my conclusion is, this Linksys BEFW11S4 (802.11b) is not really
>> compatible with some 801.11g adapters, including some Linksys wireless
>> adapters.
>
>This conclusion is incompatible with the fact the the Broadcom G adapter
>worked fine for a year. The fact that the Linkys B adapter worked but the
>Linksys G did not proves nothing. Those two adapters probably differ in more
>ways than just B vs. G. They may have been made by different vendors (resold
>by Linksys), using different antennas, and may have different receiver
>sensitivities. It sounds to me like you have some new source of interference
>that pushed the old Broadcom card over the edge, but your old Linksys B card
>happens to be able to tolerate it.
>
>>
>> I've read some other posts, mostly old, that poo-poo the idea of
>> incompatibility between b and g equipment. Any ideas here? I will
>> probably acquire a "g" WAP in the near future -- in part to test my
>> theory, and in part to satiate PC (1) who has terrible throughput at
>> the moment...
>
>Let's define compatibility. According to the standards, it means that a G
>access point supports a mix of B and G clients, by design. It does *not*
>mean there is no performance penalty. There is always a penalty. If the B
>clients are very quiet, the G clients can operate at full G bitrates, with
>only a little throughput penalty for certain messages used to co-ordinate
>between the different client types. If the B clients generate even moderate
>traffic, all clients will be dragged down to B bitrates.
>
>If your access point is configured to support a B network, then *all*
>clients will connect as B clients, whether they are capable of G or not. A B
>access point cannot support or even recognize a G client. The G client
>recognizes B, and switches to that mode when associating with the access
>point.
>
>If you buy a G router, and host 1 goes back to full throughput, it still
>doesn't prove the theory that G is incompatible with B. It probably just
>proves that your new router has a higher transmit power, better antennas, or
>both.

Thanks for your input.