Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I've been doing some network troubleshooting, and one of my laptops is only capable of transferring across the local network at about 2 Mbyte/s (15 Mbit/s). The tests I ran were a file transfer of about 200 MB from a wired computer to a wireless G network card on the laptop. The laptop was positioned about 1 meter away from the router, and changing the position or direction of the router had no noticeable differences.

Reading Why is my 802.11g wireless LAN so slow?, it appears that 2 Mbyte/s is apparently a decent speed for a G network. Why? When the theoretical maximum is 54 Mbit/s, I would expect a transfer speed of at least half of that, up to 60%.

I've ran the same tests using the N interface on my router, and I achieve speeds of approximately 60 Mbit/s. I'd expect the answer regarding the G network extrapolates to N also, correct?

share|improve this question
up vote 3 down vote accepted

If you are on a perfectly clean channel, with a signal strength (RSSI) of between -20dBm and -60dBm, and you have a well-optimized TCP/IP stack and application, and high-quality 802.11g chipsets, you should be able to see as much as 25 megabits/sec (30 if both ends support frame bursting).

Note that 1 meter away may be too close. At such close range, it's possible to have a signal strength above -20dBm, which could be "too hot" of a signal and overloads the receiver. High-quality chipsets might be able to handle signals as hot as 0dBm and still receive at maximum data rates, but I've seen plenty of lesser-quality chipsets that lost their top data rates at -20dBm. 2-3m away is a better choice for top data rates.

Here in almost 2012, finding high quality G gear is pretty hard, because 802.11g is from almost a decade ago. Anyone still making G-only chipsets now or in the last 3-4 years was likely only doing it to be as cheap and small and low-power as possible (for the smartphone/tablet/netbook markets, among others), which is kind of the opposite of high quality.

The companies making high-quality 802.11 chipsets in late 2011 and early 2012 are making 3x3:3, HT40 (450 megabits/sec) 802.11n gear. And even then, they spend most of their time making sure their N rates are optimal, and less time worrying about optimizing their backwards compatibility with a/b/g.

Having a well-optimized TCP stack and an app that always keeps the TCP pipe full is good too. I recommend IPerf as a simple performance tool that knows how to use TCP effectively. If you get much better performance with IPerf than you did with the app you were running, then the app you were running is probably non-optimal. See what TCP window IPerf reports your machines are using, and make sure it meets or exceeds the "'bandwidth * delay' product" for your network (you likely need something like 20KiB or larger).

share|improve this answer

Several possible reasons - some major ones listed below.

Firstly, interference - whether from other networks on the same or similar channel or from other stations on your network. If the base station receives an interfering signal then it is likely that the whole packet will have to be retransmitted. In these circumstances an exponential "back off" is used, meaning that the time between retransmissions grows exponentially as a way of trying to reduce interference. To minimise this it is always worth looking at the channels being used by nearby networks to get the maximum distance in channel numbers (adjacent channels are likely to interfere almost as much as overlapping channels).

Of course wired networks also have this problem and the standard back off procedure used on wired networks comes from early wireless networks. But, in general, wired networks only have to worry about stations on the same network.

Secondly, radio waves reflect off surfaces and so the same signal can arrive at the same destination via two different routes at two different times - again causing interference.

Thirdly, application performance. TCP based protocols will use windows to regulate data flow in the network. When the window is full - because all data has yet to be processed at the client side - then further data will be blocked. Not a problem with wireless per se, but would add to the sense that your wireless network is slow.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.