"Adaptivity (Channel Access Mechanism)" in Wilink7 (wl1283) module

Hi Soren,

The WiFi Driver/Fw version you are using is really old... The latest is: 7.3.10.0.139
Please update to the latest version and see if it solves the problem.

Regards,
Gigi Joseph.

Hi Gigi,

Thanks for your fast reply. Highly appreciated, and yes we would like to upgrade, but unfortunately we feel we miss the bigger picture of what is the correct package to use (and where to get it from, as well as to which problem an update might fix)?

As of today I know of the following sources, which all seem kind of official, but it's unfortunately unclear which one would be the proper(newest one to use (backport) for kernel 3.4:

TI OMAP 4 Android release (old, but what we today are using):
- git://git.omapzoom.org/device/ti/proprietary-open
- git://git.omapzoom.org/platform/hardware/ti/wlan

Various Firmware gits:
- https://github.com/TI-ECS/ti-utils/tree/master/hw/firmware
- https://github.com/TI-OpenLink/ti-utils/tree/master/hw/firmware
- git://github.com/TI-OpenLink/firmwares/tree/master/ti-connectivity
- git://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git

Various Driver/Kernel gits:
- git://git.kernel.org/pub/scm/linux/kernel/git/luca/wl12xx.git
- git://github.com/TI-OpenLink/wl12xx.git
- https://git.ti.com/wilink8-wlan/wl18xx   (maybe?)

FW 137 seem to be available at https://github.com/TI-OpenLink/ti-utils/tree/master/hw/firmware, but I haven't seen any references to 139 anywhere? Can you please share link, as well as link to corresponding driver (as well as any information you might have on how to easiest "backport" the new material to our 3.4 kernel supplied with the Android 4AJ2.5p2 release.) Which role does the compat project play in this (if any)? Unfortunately a bit unclear as well?

Hoping above request is clear, and that it's an easy task for you to let me know where to pick the different pieces, so that we ensure ending up with a working system when they are put together :-)

Best regards and thanks again in advance - Highly appreciated
  Søren

A few months ago we upgraded the driver to use 7.3.10.0.136 which should work but we have had the same result works at 2.4 Ghz but not at 5.0 Ghz. If you port the 7.3.10.0.139 please post the results.

Hi Gigi and Richard,

@Richard: Thanks for for your reply although it of cause wasn't optimistic news :-(:-) Anyway glad to know that others are facing same kind of trouble and sharing their findings - Highly appreciated :-). Did you resolve/work around the problem in anyway, or are you as well still trying to find a solution?

@Gigi: We have now ported to what we believe being newest formware and driver. Stuff is taken from:
- FW     : git://github.com/TI-OpenLink/ti-utils.git (tag:  ol_R5.SP7.01)
- Driver : git://github.com/TI-OpenLink/wl12xx.git   (tag:  ol_R5.SP7.01)

FW version report 7.3.10.0.139, and it seem to have fixed some other hick-ups we have had, but I don't know yet if it solved the CCA problem (yet), but based on Richards note above that FW 136 didn't solve it for them I unfortunately have small believe that FW 139 would solve it for us, unless you can tell me that problem specifically have been addressed between FW 136 and FW 139? Change log in ti-utils git doesn't seem to state anything like this (unfortunately). Can you please provide me an update on this?

With respect to wpa_supplicant we looked at git://github.com/TI-OpenLink/hostap.git, but here is seem that the latest stuff (ol_r8.a8.10) is much more in line with the Android 4AJ.2.5P2 release than that's tagged as ol_R5.SP7.01? So this project seem to have been updated recently in the Android releases (external/wpa_supplcant_8_ti)?. I believe this being unrelated to above driver files as well as the faced problem, but would anyway like to mention it here to be sure we have all aspects covered...

Best regards and thanks again in advance
  Søren

Soren

No we have not had any success resolving this issue or finding a workaround. We have received various software revisions from TI changing the threshold but no luck so far.  In fairness to TI they have been working with us an been providing pretty good support. Unfortunately the European deadlines are coming and everyone is anxious here.  The good news we have been working this issue for a while and the issue is currently being worked on. I am hoping we are close to a resolution on this.

Hi Richard,

Great and thanks for your very fast following up. We are happy to be on same train and help out where we can to get this working as expected. I as well just posted a reference to this thread in http://e2e.ti.com/support/wireless_connectivity/f/307/p/371006/1308526.aspx, as it seem that feature is confirmed working in wl18xx? Anything you have discussed with TI, as I guess backporting then should be possible :-)

Likewise please don't hesitate to let me know if there are any settings you would like me to try on the 139 FW, or you have any good hints you can share for me/us to try ot help out debugging this?

Best regards and thanks again
  Søren

We have tried numerous things but our main tester is updating some results from last week. I asked him to join this conversation and he will be posting later on.  We found with the 2.4 Ghz  that the interference signal made a difference but so far no luck with the 5 Ghz.. He will summarize some of  those results later when he posts on this thread.

Hi Soren,

I have been working with Richard on this and I wanted to talk about the nature of the interference signal used in your testing. Our test house uses a FM signal with a baseband rate between 100kHz and 1 MHz. After struggling for weeks in the 2.4 band (thinking it was a driver or fw issue), we lucked out when the tech tried a 600kHz baseband rate. The radio reacted appropriately and we are now struggling in the 5GHz band. Did you experience similar issues?

In emails with TI, they seemed very surprised that the nature of the signal would have such an effect and suggested that a simple CW 1MHz off the carrier ought to do the trick. We tried that in the lab without success.

One other question that came up in recent days for us. When you do the test, what mode are you putting the radio in? For instance, we set up a file transfer between our SOM (wl1283 is part of it) and a box and then pipe the interference in.

Thanks for your help, it is a bit of a relief to see other groups dealing with similar issues,

Cheers,

Brandon

Hi Søren,

Your observation is correct. The 5GHz band in 12xx for both the versions you're talking about had a higher CCA threshold configuration then the 2.4GHz. It is the same HW mechanism (which works perfectly well) but FW configuration was set differently for a number of reasons such as air competivness in the presence of WLAN and non WLAN weak interferers. 

Regards,

Oren Shani

Hi Oren,

Thanks for your fast feedback. Highly appreciated. Anything we can change/modify in anyway? As right now we are failing ETSI EN 301 893 V1.7.2 (http://www.etsi.org/deliver/etsi_en/301800_301899/301893/01.07.02_20/en_301893v010702a.pdf) chapter 4.8 test, as we don't fulfill the -73+23-TX_Power rule, as the CCA first get activate around -60dBm, which only would allow us an output power of 10dBm, as -73+23-10 = -60dBm...

In short, we in order to pass (as I see it) the above test therefore need to either:

1. limit maximum output power to 10dBm (not preferred I think) or
2. find a way to configure the CCA level to a lower level (i.e. -65dBm, which would allow 15dBm output)

Can any of 1 or 2 be done/configured by us (end user), or does any of those changes need a FW change from you (TI)? Hoping you can share some more insight on this and that "problem" is clear?

Best regards and thanks again. You help and support is highly appreciated
   Søren

Hi Brandon,

For the 2.4GHz we had it working straight out the box using a like 22MHz wide flat top interferer (check http://www.ietf.org/mail-archive/web/6tsch/current/pdfd3d1acPkgu.pdf page 46 step 3), which show that interferer signal should be wider than occupied channel bandwidth (typically around 16-20MHz).

We therefore used same kind of interferer signal for the 5GHz, although I can't easily find where this this signal type  recommendation is decribed in the 5GHz spec (at least not right now :-))...

Wrt the data testing we do, we run an iperf test between wl1283 and an AP connected over Ethernet to a PC, and expect to see iperf transmission stopping when interferer signal is introduced on link as described in test...

Hoping above described? Otherwise don't hesitate to let me know and I will be happy to try to clarify further. And yes I'm as well kind of relieved that we aren't the only ones with this problem as well, so let's now just hope we can get it fixed :-)

Best regards
   Søren

Hi again,

Problem is completely clear and can be solved easily.

Unfortunatly changing the CCA threshold will require a FW fix.

I will be able to provide you this fixed version.

Please address my personal email :orenshani@ti.com

Regards,

Oren Shani

Hi Oren,

Thanks for your feedback. I just send you a private email as requested.

Assuming you as well take it forward with Brandon and Richard as well?

Best regards and thanks again
   Søren

Hi Soren,

We are in contact with Oren as well and we are moving forward. (Thanks Oren!!)

Good luck with your project Soren, I will be keeping an eye on the thread.

Sincerely,

Brandon

Hi Søren -

Has the alternative firmware gotten you to where you need to be?  Are you passing the 5 GHz band now?  We're still doing testing and we're wondering how things were going for you to compare to where we are.  Unfortunately we haven't yet passed at 5 GHz.

 - Tim

(P.S. I have worked with Brandon and Richard as well regarding this issue.)

Dear Søren,

Sorry for re-opening this old thread, but we have a similar issue that you had.

We're also failing the CCA compliance test. (detection is too late, at <70dBm)
Your initial request on this thread was to find a way to "configure the CCA level to a lower level".

There is a parameter called 'rx_cca_threshold'. Is there any special setting you had to apply in order to pass the CCA adaptative test?

Best regards,

Thomas