CSI collection not working correctly

[cheneso89]

I tried to collect CSI with ASUS RT-AC86U but this is the plot that I got, that's the CSI amplitude (80 Mhz) for each receiving antenna. I couldn't find what's the problem while I'm collecting it.

[zeroby0]

Pilot and Null subcarriers have arbitrarily high csi values and are over shadowing you actual csi data.

Indices for different bandwidths are available here: https://github.com/nexmonster/nexcsi/blob/main/nexcsi/__init__.py#L5

Or you can use the https://github.com/nexmonster/nexcsi and remove the pilots and nulls as decribed in https://github.com/nexmonster/nexcsi#null-and-pilot-subcarriers

[cheneso89]

Also removing pilot/null subcarriers it still doesn't work, higher values correspond to more subcarriers values than the number of null/pilot subcarriers. In the zoomed plot I already removed them.

[zeroby0]

Er, it doesn't seem the right nulls and pilot values have been removed.

For an 80 MHz channel, these are the indices [0, 1, 2, 3, 4, 5, 25, 53, 89, 117, 127, 128, 129, 139, 167, 203, 231, 251, 252, 253, 254, 255] that must be made 0. Maybe I'm not looking right, but it seems indices 0->5 are non zero in the graphs.

Or maybe your CSI is just like that. Can you post a pcap with 10-20 packets?

[cheneso89]

the indices seem correct. This is one of my pcap files

lab_trial_2.pcap.zip

[zeroby0]

import numpy as np
import matplotlib.pyplot as plt

from nexcsi import decoder


device = "rtac86u" # nexus5, nexus6p, rtac86u
samples = decoder(device).read_pcap('lab_trial_2.pcap')

# Accessing CSI as type complex64
csi = decoder(device).unpack(samples['csi'], zero_nulls=True, zero_pilots=True)


f, (ax1, ax2) = plt.subplots(2, 1)
ax1.plot(np.abs(csi[0])) # Amplitude
ax2.plot(np.angle(csi[0])) # Phase

plt.show()

[cheneso89]

I already saw this code.. I'm working on Matlab though, with the csireader.m code

[zeroby0]

Ah. Then maybe modify it to set these indices to 0. Python array index starts from 0 while matlabs starts from 1, so maybe that's the confusion?

I don't know matlab much, however, so I can't help you with that.

[cheneso89]

No no I already shifted indices by 1, it doesn't work. I'm going to work with python. Thank you anyway

[jlinktu]

Your question is not related to this issue. Please, don't choose random issues and post your questions there!
Also, tcpdump is not a tool by us. You can find a very detailed description of its usage with a quick online search, e.g.: the tcpdump man page.

[lionlanlan]

Your question is not related to this issue. Please, don't choose random issues and post your questions there! Also, tcpdump is not a tool by us. You can find a very detailed description of its usage with a quick online search, e.g.: the tcpdump man page.

I am very sorry for my impoliteness.I will delete my comment.Sorry for this again.