[CYW43907] Join delay issue

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
JoLE_3168931
Level 3
Level 3
First like received Welcome!

Hello,

we are currently evaluating the CYW43907 Wi-Fi module for a VoIP like application, where a seamless roaming is expected.

And during the tests of our own roaming manager* we noticed a full second of packets loss during connection to the access point (using wwd_wifi_join_specific() to a specific AP).

After some packets sniffing we found out the problem.

See the captured packets:

steps.png

While the WLAN module can manage an impressive 10 ms association time, there is a huge delay after the AP probe response, just before the association process, causing the packets loss.

Actually, the authentication frame transmit seems to be delayed by a CTS frame sent nearly one second after the probe response (while there is no RTS frame involved).

This CTS frame is sent by the CYW43907 module (checked with the signal power).

And, what is strange is the receiver address of this CTS frame:

cts.png

while the two last digits of the MAC address are those of the module (5A:94), the OUI is now Broadcom (E0:3E:44) and not Murata (44:91:60).

Seems like a bug to us, we never saw this with others WLAN modules.

If this 1 second delay can be avoided, a VoIP roaming without any packet loss could be achieved.

Thank you for your help.

Hardware/software used:

CYW943907AEVAL1F kit

AP Unifi AC Pro (5Ghz band)

Wiced Studio 6.2

Wireshark 2.4.4 with WLAN sniffer

*: we don't use the automatic roaming feature of the chip because it leads to too much packets loss.

0 Likes
3 Replies
PriyaM_16
Moderator
Moderator
Moderator
250 replies posted 100 replies posted 50 replies posted

Hello,

Can you please try with the latest firmware from wiced 6.2 version?

I have captured the logs and I am seeing the OUI as Murata:

pastedImage_0.png

0 Likes

Hello,

On your capture, you can't see this CTS frame, because you filtered with the module's MAC address.

As stated in the previous message, we are already using WICED 6.2 (we will try 6.4).

The only CTS frame with Broadcom OUI is the first one, just before the Authentication frame.

The subsequent CTS frames contain the Murata OUI.

By the way, what bothers us is not this CTS frame, it's the duration of the join process, more precisely the delay between the probe response from the AP and the authentication frame from the module (almost 1 second), causing massive loss of audio frames during roaming.

This delay can easily be checked with the powersave app (POWERSAVE_STANDALONE_TEST_* set to zero), using the join_specific command (you have to set the bssid and channel parameters).

Hello,

We have tested WICED version 6.4 this morning, same behavior.

0 Likes