WARP Project Forums - Wireless Open-Access Research Platform

You are not logged in.

#1 2015-Oct-29 15:33:44

junyi
Member
Registered: 2015-Aug-25
Posts: 8

Timeout problem with reference design.

I'm using the latest 802.11 Reference Design. I connect the WARP board to an AP I established by my laptop and try to measure the throughput. The result shows that it seems that the throughput is always half of the possible rate (for example, when the MCS is set to 54Mbps, I got a throughput around 24Mbps. The throughput is measured by UDP Iperf). I'm sure it's not because of interference. I connect my smartphone to the same AP and get a throughput around 48Mbps.

Further debug shows that in the transmission, every packet timeout once. Specifically, when the packet is first transmitted, it enters the case "WLAN_MAC_STATUS_TX_A_RESULT_TIMEOUT" (line 1082 in wlan_mac_dcf.c). After that, the packet is retransmitted and this time the transmission succeeds. As a result, every packet is actually transmitted twice.

Do you have any idea how to fix this problem?

Offline

 

#2 2015-Oct-29 16:05:16

chunter
Administrator
From: Mango Communications
Registered: 2006-Aug-24
Posts: 1212

Re: Timeout problem with reference design.

junyi wrote:

I'm using the latest 802.11 Reference Design. I connect the WARP board to an AP I established by my laptop and try to measure the throughput. The result shows that it seems that the throughput is always half of the possible rate (for example, when the MCS is set to 54Mbps, I got a throughput around 24Mbps. The throughput is measured by UDP Iperf). I'm sure it's not because of interference. I connect my smartphone to the same AP and get a throughput around 48Mbps.

The maximum achievable MAC throughput for the 54 Mbps is only around 30 Mbps. This accounts for all the MAC overhead of 802.11 and is supported by theory. So, the throughput of 24 Mbps sounds reasonable to me in an actual over-the-air experiment with interference and other signal degradation. Regarding the the 48 Mbps throughput you observe, I suspect your smartphone, when talking to the non-WARP AP, is choosing 802.11n rates higher than 54 Mbps. This can be verified with another Wi-Fi device in monitor mode using Wireshark.

junyi wrote:

Further debug shows that in the transmission, every packet timeout once. Specifically, when the packet is first transmitted, it enters the case "WLAN_MAC_STATUS_TX_A_RESULT_TIMEOUT" (line 1082 in wlan_mac_dcf.c). After that, the packet is retransmitted and this time the transmission succeeds. As a result, every packet is actually transmitted twice.

Do you have any idea how to fix this problem?

Can you clarify how you are performing this debugging? Is this using the SDK debugger with breakpoints? If so, you'd expect that halting the execution could lead to various timeouts as the MAC support core expects software to keep up.

Offline

 

#3 2015-Oct-29 16:41:52

junyi
Member
Registered: 2015-Aug-25
Posts: 8

Re: Timeout problem with reference design.

I double check the debug. Yes, I made a mistake and it's not the problem of timeout.

Sorry for the disturbance. Thank you for the help!

Offline

 

Board footer