# CSI in gr-ieee80211 **URL:** \ **Category:** GNU Radio\ **Created:** [March 26, 2014, 3:06pm UTC](https://www.ruby-forum.com/t/csi-in-gr-ieee80211/240076 "2014-03-26T15:06:59Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![musicdenotation](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/ea666f/32.png) [@musicdenotation](https://www.ruby-forum.com/u/musicdenotation)\ **Post date:** [March 26, 2014, 3:06pm UTC](https://www.ruby-forum.com/t/csi-in-gr-ieee80211/240076/1 "2014-03-26T15:06:59Z")
Dear Bloessl and gnuradio list, I am currently working on gr-ieee80211. Now I want to use the long preamble to estimate the CSI. However, after I apply fft to the long preamble, I found the power for each channel is not the same. I took an average over lots of packets. The power near the middle sub-carriers are always very lower than on the edges. I would like to ask if it is because the Power spectral density in the transmitter or some other reasons. Any feedback would be appreciated. Best regards Alex ---
**Author:** ![alex](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/db5fbb/32.png) [@alex](https://www.ruby-forum.com/u/alex)\ **Post date:** [March 26, 2014, 3:11pm UTC](https://www.ruby-forum.com/t/csi-in-gr-ieee80211/240076/2 "2014-03-26T15:11:06Z")
On 26 Mar 2014, at 15:05, alex [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > I am currently working on gr-ieee80211. Now I want to use the long preamble to > estimate the CSI. However, after I apply fft to the long preamble, I found the > power for each channel is not the same. I took an average over lots of packets. > The power near the middle sub-carriers are always very lower than on the edges. I > would like to ask if it is because the Power spectral density in the transmitter > or some other reasons. If your sample rate is 20MHz you will see an uncompensated filter, but in that case the power of the subcarriers in the center should be higher. ---
**Author:** ![alex](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/db5fbb/32.png) [@alex](https://www.ruby-forum.com/u/alex)\ **Post date:** [March 29, 2014, 2:33pm UTC](https://www.ruby-forum.com/t/csi-in-gr-ieee80211/240076/3 "2014-03-29T14:33:52Z")
On 28.03.14 19:40, Surligas Manos wrote: > Which device did you used for receiving samples? I have noticed a same > phenomenon with my device, caused by a coarse implementation of a high > pass filter. The filter in the Maxim2829 in the XCVR2450 that we used can be configured to 0, 100Hz and 30kHz and the default is 100Hz I believe? If you didn’t change anything there this cannot explain your problem. Any filter than goes above 150kHz would be a problem for the centerish carriers in 802.11 OFDM (~300kHz carrier spacing). what happens if you TX a sinusoid and stepwise tune the RX to slightly off frequencies (take note of the _actual_ RX LO frequency when doing the sweep - the difference to that one and the TX sinusoid matters). Does the RX power also start to decrease as soon as you set the TX-RX offset smaller than 150kHz? -paul PS: as Sebastian said, the outer subcarriers being lower can be explained by odd decimation factors like 5 - which causes the digital-downconversion filter chain to change to a single CIC stage I believe. ---
**Author:** ![alex](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/db5fbb/32.png) [@alex](https://www.ruby-forum.com/u/alex)\ **Post date:** [March 28, 2014, 7:41pm UTC](https://www.ruby-forum.com/t/csi-in-gr-ieee80211/240076/4 "2014-03-28T19:41:53Z")
Which device did you used for receiving samples? I have noticed a same phenomenon with my device, caused by a coarse implementation of a high pass filter. On 03/26/2014 04:08 PM, Bastian B. wrote: > On 26 Mar 2014, at 15:05, alex [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > > > I am currently working on gr-ieee80211. Now I want to use the long preamble to > > estimate the CSI. However, after I apply fft to the long preamble, I found the > > power for each channel is not the same. I took an average over lots of packets. > > The power near the middle sub-carriers are always very lower than on the edges. I > > would like to ask if it is because the Power spectral density in the transmitter > > or some other reasons. > > If your sample rate is 20MHz you will see an uncompensated filter, but in that > case the power of the subcarriers in the center should be higher. > > * * * > > Discuss-gnuradio mailing list > [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) > [Discuss-gnuradio Info Page](https://lists.gnu.org/mailman/listinfo/discuss-gnuradio) – /\* Code is the Law! \*/ ---
**Author:** ![alex](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/db5fbb/32.png) [@alex](https://www.ruby-forum.com/u/alex)\ **Post date:** [March 31, 2014, 3:09pm UTC](https://www.ruby-forum.com/t/csi-in-gr-ieee80211/240076/5 "2014-03-31T15:09:46Z")
I use a card from Per Vices. I do not have the option to change the cuttoff frequency of the filter without flashing the firmware. As a workaround, I tuned my device to a different frequency (+10 MHz) and using an NCO I managed to bypass the power loss at the center sub-carriers. On 03/29/2014 03:11 PM, Paul Fuxjaeger wrote: > what happens if you TX a sinusoid and stepwise tune the RX to slightly > believe. > – > /\* Code is the Law! \*/