Why does the connection speed slow down?

I can tell you my experience : I had the same problem, and after a couple of days of scratching my head, I discovered a school close to me has a wireless lan broadcasting on the same channel 1

I suppose they have installed an outdoor antenna, to reach all the rooms. Altough I'm not able to spot it.

My suggestion is to change from the default channel(1) to another one.

This is normal response to interference and reflections. How much speed juggling it does is dependent on the algorithm used. The way it works is that the client tries to associate with the access point at the highest speed possible. This is advertised in the management frames so there's no trial and error involved like in dialup modems. It usually succeeds making the initial connection at the highest speed because there's very little traffic moving and the probability of a successful association is rather high.

That usually remains the situation until you start moving data. At

54Mbits/sec association speed, anything furthur than about 8ft range (using typical antennas) is going to create data errors. The data arrives corrupted by noise. The access point detect the arrival of corrupted packets and lowers the speed in order to reduce the error rate to reasonable level. Depending on range, it will reduce the speed until it gets a decent error rate (about 1 error in 10^5 bits).

However, that how it should work without any interference present. With interference, there's a problem. The speed can go down to the very slowest speed (1Mbit/sec) and there will still be errors. In fact, since the same size packets take longer to transmit at the slower speeds, the probability of a packet getting trashed by interference is even higher with slower speeds. Having the speed plunge to 1Mbit/sec is a sure sign of local interference.

A good question is how does this system recover from an interference hit? In other words, what criteria or algorithm determines when it should increase in speed again. I had illusions that some sophistocated statistical algorithm was employed. Nope. One AP product used a simple timer. After 10 seconds of no traffic, it would reset the speed back to the maximum speed and just start over. I suspect most are like this. Cheap, crude, functional, but not very elegant or optimal.

Its a known problem with a Intel 2200BG wireless setup (bet it is) and is waiting for Intel to publish a reasonable driver. Uninstall proset and use the WZC setup with XP with Version 8 drivers (install only the card drivers) and thing will get a lot better. Then scream loudly at who sold you the laptop, who hopefully will scream at Intel. If you cant find V8's email me

DS

Wow, I read that thread and that is exactly my problem. It is amazing that it has been unfixed for so long. I have a brand new HP DV4040us with the 2200. The drivers and fixes listed in that thread do not seem to work for most people. I guess I will just have to wait for Intel to release drivers that do fix the problem.

I just read through all 233 articles in that blog. Methinks almost everyone is ignoring a common issue. The few that bothered to post details on their hardware indicated that they were using Linksys WRT54G access points. Much of the rest is consists of trial an error testing of various drivers with very inconsistent results. Are you prehaps using a Linksys router?

What everyone seemed to be ignoring is that it's the access point, and not the Intel Centrino client, that sets the speed. Centrino simply follows the speed that the access point commands. If the forum users had tried to set the speed of the access point to some reasonable speed (i.e. about 12 or 24Mbit/sec), instead of "auto", they would probably have found that their connection was stable. It would be an interesting test.

My guess(tm) is that Intel's driver isn't returning some management request in time for the access point to respond. This is not the first time this has happened with the Philips Centrino chipset. When the Netgear MR814v1 was new in early 2004, when the Centrino chipset started to appear in OEM laptops, the two devices would connect and drop repeatedly. I never did find out exactly which company screwed up, but Netgear was forced to loosen their timing specs to accomidate Centrino. (Biggest company always wins). I'm guessing this is more of the same.

I don't have a fix. Try the various Intel drivers and see what happens. If that doesn't work, try setting the access point to a fixed but reasonable speed and see if that helps.

No Linksys here. I have a D-Link, like some of the people in that blog, (and in a couple of other forums I found), with the same problem. Most do have Linksys, but that may be because that is such a common brand. I think there were a couple of other brands of routers also. The common thread was the Intel 2200 and the other Intel one. I don't know if there was anyone who posted there that had the problem fixed by downloading drivers. A few thought they had fixed it, only to find out later it was not fixed. The only ones who fixed the problem seemed to have changed to something other than the 2200. That seemed to fix it in 100% of the cases, regardless of the brand of router. I am hesitant to start trying a bunch of different drivers as I am not a tech person and it seems the likelihood of that fixing things is slim.

I guess I will just live with it for now and hope that a certain fix is put out soon by Intel. Amazing that it has persisted this long. I called a few acquaintances with this same wireless setup and they all confirmed they had the same thing happening. I try to use an ethernet connection when downloading updates and programs so the wireless card does not affect that. Browsing sometimes slows to the point of dropping the connection, but almost immediately reconnects at a higher speed.

Jeff Liebermann wrote: