Getting Wireless N Networking speed!

[bkaler]

Ok, I've been pretty upset with my G router's performance (typically 12Mbps) and thought to ask for a N router this X-mas. So I researched the topic and decided to go with the Linksys WRT310N for reasons of future compatibility with DDWRT and the widespread adoption of linksys routers will hopefully provide some solutions to the common problems that will no doubt occur)

Problem is, at the VERY best I've only been able to get 20Mbps out of the router! Often times it drops much lower, to 5Mbps. I've tried multiple N adapters, even going as far to purchase a 75$ fancy-dancy LINKSYS branded adapter that uses the dual band bonding tech. I thought with a LINKSYS router and a LINKSYS adapter, I would be able to connect at least a couple times faster than before, but ZERO improvement. IN fact, somewhat worse because I'm using the USB WUSB600N adapter that has no external attenna.

So I'm very frustrated in linksys now. I'm thinking of returning the 75$ adapter, and I really can't return the router because it was a gift and I THOUGHT I was making an informed decision.

I found the 5 ways to fix slow N performance article, and have followed every direction there (really thought I had it when I changed over to WPA2/AES but alas, no difference)

So just in case anyone has suggestions, here's the vital information:

Most of the PCs are running Vista, but I only need to get 1 PC with Wireless N speed right now, that PC is running Vista. I do have a Windows XP PC on hand however. The router and the target PC are less than 20 feet from each other. I am using WIDE band (but I've tried auto/20mhz) currently and the advanced wireless settings are default (frame burst on, 100/1/2346/2347) even 30-40 Mbps at this point would make me happy...

I've used 2 different Rosewill N adapters that were favorably reviewed on newegg (they use only 1 band) and the $$$ dual band linksys adapter that does jack. I've checked the product page for the adapter and I can't confirm that it uses draft 2.0 technology, but if this will hurt me when using adapters I'm unsure.

I do use some powerline devices on my network but they're all limited around 10 Mbps, but all my IPERF tests were run from a Vista PC directly connected to the router, thereby bypassing the powerline devices.

I have not tried DD-WRT but due to it's less than perfect nature in many builds and on newer routers, I'd prefer to avoid it, but if someone has a similar story that DD-WRT fixed, I'd give it a try.

 

[bkaler]

I've done some testing and found that I could increase to 30 Mbps by using single band on channel 9 (strangely even though there are virtually no close AP using channels 1 and 11 they are lowest performing - channel 6 being the default for so many routers it's the worst. There are about 10 AP near me (but only about 2-3 with a strong enough signal to actually use) so I'm not sure how that effects things too, only 1 of which is an N router and it's one of the weakest signals.

I'm still hoping for better however, if anyone has ideas that would help (changing the position of the router, covering my head in tin foil, etc)

 

[thiggins]

How are you measuring wireless throughput?

There is no logical reason why channel 9 should work better. If the strongest nearby networks are on 6, then using channel 1 or 11 without channel bonding should work.

What utility are you using to detect the nearby networks?

 

[bkaler]

I'm using Iperf to measure. I would figure 1 or 11 would be the strongest but it's not the case, I drop about 5 Mbps on those channels (and gradually improve on the nearby channels, 9 seemed to be the best but channel 4/8/9 seemed the best

I've used a few, but mostly I've been using the utility with the rosewill adapters, I think it's RALINK. But I'm pretty familiar from the wireless points nearby from previous tweaking using netstumbler/airodump.

 

[thiggins]

I was mainly trying to determine if there were other APs that you weren't seeing, but sounds like you know the environment.

Try running FTP as a sanity check on the throughput #s.

 

[bkaler]

well, with a large fedora ISO, it transfers between 4/4.5 MBps (32-36 Mbps) making it better than my old G (topped at 12) but not by a great margin at all.

Thanks for the help Tim, is there anything else I can do? I just need a good transfer rate between these two points (about 15 feet away)

 

[thiggins]

36 Mbps vs. 12 Mbps is a 3X improvement, about typical for real-world draft 11n.

Also see this thread for iperf/jperf settings.