search

dhalperi / linux-80211n-csitool-supplementary

802.11n CSI Tool based on iwlwifi and Linux-2.6
http://dhalperi.github.com/linux-80211n-csitool/
195 stars 129 forks source link

In the same location at different time to get three very different data？ #52

Closed chenyongxi closed 9 years ago

chenyongxi commented 9 years ago

Collecting data at different times at the same position（for example half an hour interval）.The data I collected very different. One's mean（30 subcarriers） is 22db， the other's mean is 15db.This is normal? Where was I wrong?

dhalperi commented 9 years ago

This is normal.

  1. Different transmit rates (# of streams, transmit antenna selection) affect the link.
  2. Movement in the environment or repositioning any RF-visible materials, such as metal water and glass, including people, will affect the link.
  3. Even, sometimes, rebooting the hardware will result in recalibration which can affect the reported CSI.
chenyongxi commented 9 years ago

Thank you！The second case I have taken into account as far as possible avoid, the measurement at night. I want to know how to avoid the third case.( rebooting the hardware will result in recalibration which can affect the reported CSI.)

chenyongxi commented 9 years ago

Such lager fluctuations is nomal?

dhalperi commented 9 years ago

Yes. Number 1 is the most likely still, however, if you've ruled out any movement or change in the environment.

chenyongxi commented 9 years ago

But measurement is the same ap.

chenyongxi commented 9 years ago

Throughout the experiment with a same device I have used. Are there any ways to avoid the impact of the first case (Different transmit rates (# of streams, transmit antenna selection) affect the link.)?

dhalperi commented 9 years ago

It sounds like you are connecting to a commercial AP? The ap will select rates and or antennas, and that will be out of your control. You can compensate for this by clustering your received csi reports based on the ap configuration. You can observe the transit rate directly; you will have to come up with some way to detect when the ap uses antenna selection or beam forming.

Alternately, use both transmitter and receiver as Intel 5300 devices and use injection mode, then you will precisely control the transmitters use of antennas and rates, etc. On Tue, Jun 9, 2015 at 23:11 chenyongxi notifications@github.com wrote:

Throughout the experiment with a same device I have used. Are there any ways to avoid the impact of the first case (Different transmit rates (# of streams, transmit antenna selection) affect the link.)?

— Reply to this email directly or view it on GitHub https://github.com/dhalperi/linux-80211n-csitool-supplementary/issues/52#issuecomment-110604579 .

chenyongxi commented 9 years ago

I did connecting to a commercial AP. We can not connecting to a commercial AP?

dhalperi commented 9 years ago

You absolutely may connect to a commercial AP – as you have already found, the CSI tool will still work!

However, you have no way of guaranteeing that the commercial AP will not do things like change which antenna it sends data from, for instance. Thus you have no reasonable expectation that the CSI will remain stable over time as the transmitter changes its behavior in unknown ways!

You may find it useful to read my thesis, especially chapter 4: http://homes.cs.washington.edu/~dhalperi/pubs/halperin_thesis.pdf

chenyongxi commented 9 years ago

If I want to avoid receiving CSI such a large jump, how can I do?

chenyongxi commented 9 years ago

Have any commercial AP sends data stable？

dhalperi commented 9 years ago

Not that I know of. The recommended strategy for these types of long-term measurements is to use a CSI Tool transmitter as well, so that you have full control over both sides. Any commercial device is going to have unknown behavior.

On Wed, Jun 10, 2015 at 7:39 PM, chenyongxi notifications@github.com wrote:

Have any commercial AP sends data stable？

— Reply to this email directly or view it on GitHub https://github.com/dhalperi/linux-80211n-csitool-supplementary/issues/52#issuecomment-110979492 .

chenyongxi commented 9 years ago

I don't understand what is you said these types of long-term measurements is to use a CSI Tool transmitter. Can you say that more points？

dhalperi commented 9 years ago

Use an Intel 5300 device running the CSI tool to both transmit and receive. Do not use a commercial AP.

On Thu, Jun 11, 2015 at 8:03 AM, chenyongxi notifications@github.com wrote:

I don't understand what is you said these types of long-term measurements is to use a CSI Tool transmitter. Can you say that more points？

— Reply to this email directly or view it on GitHub https://github.com/dhalperi/linux-80211n-csitool-supplementary/issues/52#issuecomment-111166082 .