WARP Project Forums - Wireless Open-Access Research Platform

Christian wrote:

You definitely need to check/adjust the packet detector thresholds if you are so "far" off. The TX capabilities diminish by getting closer to the specified bounds of the MAX2829 chip.

I have tried 2.490 GHz, and there is also a lot of misdetection. Actually, 2.412 GHz is also at the boundary of MAX2829 chip.

Last edited by adherentx (2014-Nov-21 09:38:47)

welsh wrote:

Please take a look at this thread.  This might be an easier solution than changing the lower level code. 

Are you finding that you are getting a lot of false positives (ie the node is detecting packets when there are none) or are you getting a lot of false negatives (ie the node does not detect a valid packet)?  Have you run these tests over an attenuated cable to understand if the misdetection rate is similar to your OTA tests?

The method in this thread can switch to a WiFi band, but I want WARP nodes to work in non-WiFi band, i.e., outside 2.412-2.484 GHz.
I have seen a lot of packets just lost: the node does not detect a valid packet. I have not run these tests over an attenuated cable yet.

Last edited by adherentx (2014-Nov-22 07:01:23)

This problem seems to have been solved now. I find that the misdetection problem (false negative) is severe only when I place the antennas on the two WARP nodes close and parallel. If I shift one antenna an angel (20 deg, 90 deg, whatever), there will be few misdetection. If I place the two antennas farther, there will be few misdetection either.
Any idea why?

The default packet detection configuration in the 802.11 ref design is set for good performance at low- and mid-SNR receptions. At very high Rx power it is likely the default config will miss more detections.

One possible solution for this is enabling the RSSI-based pkt det subsystem. In wlan_phy_util.c, change the value of the RSSI detection threshold in the call to wlan_phy_rx_pktDet_RSSI_cfg(). The default value is the maximum (1023) effectively disabling the detector.

murphpo wrote:

This problem seems to have been solved now. I find that the misdetection problem (false negative) is severe only when I place the antennas on the two WARP nodes close and parallel. If I shift one antenna an angel (20 deg, 90 deg, whatever), there will be few misdetection. If I place the two antennas farther, there will be few misdetection either.
Any idea why?

The default packet detection configuration in the 802.11 ref design is set for good performance at low- and mid-SNR receptions. At very high Rx power it is likely the default config will miss more detections.

One possible solution for this is enabling the RSSI-based pkt det subsystem. In wlan_phy_util.c, change the value of the RSSI detection threshold in the call to wlan_phy_rx_pktDet_RSSI_cfg(). The default value is the maximum (1023) effectively disabling the detector.

After a long time I come back to this old post now.
I met detection problem when using antennas with higher gains. I have tried adjusting the RSSI detection threshold. The adjustment made the detection better, since I observed that the green LED blinked (before the RSSI adjustment it rarely blinked). However, the problem is still not solved: the WARP nodes can not get associated even after a long time. I guess it should be the problem of the AGC. Perhaps the current AGC configuration only fits low receive power. How should I change the AGC to adapt it to high receive power? I have tried modifying the parameters in wlan_agc_config(), but it does not help.
Previously when I use other antennas with lower gains, it worked well and I collected some raw IQ data. From the raw IQ data I found that the first half of STS was saturated and truncated (before AGC takes effect). The later part, after AGC takes effect, was not saturated. From this observation, I guess when I use the high-gain antennas, the saturation may be more severe, and perhaps the whole packet is saturated.

Last edited by adherentx (2016-Feb-18 07:01:47)