The WiFi throughput is worse than the expect result on nRF7002

[Camelia1204]

Hi,

I want to run the throughput test under coex condition with Separate antenna. These tests were run with WLAN connected to an AP in 2.4 GHz band.

I follow the test steps as sr_coex sample. ( developer.nordicsemi.com/.../README.html)

But the test result is worse than the result list at developer.nordicsemi.com/.../README.html

I use the code of v2.4.2 version, and I modify My sr_coex/prj.conf is as follows.

CONFIG_SYSTEM_WORKQUEUE_STACK_SIZE=4096

CONFIG_NET_PKT_RX_COUNT=32

CONFIG_NET_PKT_TX_COUNT=32

CONFIG_NET_BUF_RX_COUNT=32

CONFIG_NET_BUF_TX_COUNT=64

CONFIG_STA_SAMPLE_SSID=xxxx

CONFIG_STA_SAMPLE_PASSWORD=xxxxxx

CONFIG_NET_CONFIG_PEER_IPV4_ADDR=xxxx

The result lists (Separate antenna, WiFi run @2.4G):

Wi-Fi throughput only:

$ iperf -s -i 1 -u
------------------------------------------------------------
Server listening on UDP port 5001
Receiving 1470 byte datagrams
UDP buffer size: 208 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.50.135 port 5001 connected with 192.168.50.254 port 51461
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 3] 0.0- 1.0 sec 1006 KBytes 8.24 Mbits/sec 472842.967 ms 0/ 1006 (0%)
[ 3] 1.0- 2.0 sec 972 KBytes 7.96 Mbits/sec 376650.747 ms 0/ 972 (0%)
[ 3] 2.0- 3.0 sec 980 KBytes 8.03 Mbits/sec 290953.846 ms 0/ 980 (0%)
[ 3] 3.0- 4.0 sec 951 KBytes 7.79 Mbits/sec 310350.777 ms 15/ 966 (1.6%)
[ 3] 4.0- 5.0 sec 967 KBytes 7.92 Mbits/sec 282155.134 ms 0/ 967 (0%)
[ 3] 5.0- 6.0 sec 979 KBytes 8.02 Mbits/sec 321029.346 ms 1/ 980 (0.1%)
[ 3] 6.0- 7.0 sec 967 KBytes 7.92 Mbits/sec 401760.932 ms 0/ 967 (0%)
[ 3] 7.0- 8.0 sec 981 KBytes 8.04 Mbits/sec 290953.856 ms 0/ 981 (0%)
[ 3] 8.0- 9.0 sec 992 KBytes 8.13 Mbits/sec 290953.799 ms 0/ 992 (0%)
[ 3] 9.0-10.0 sec 974 KBytes 7.98 Mbits/sec 365260.330 ms 0/ 974 (0%)
[ 3] 10.0-11.0 sec 975 KBytes 7.99 Mbits/sec 310350.921 ms 0/ 975 (0%)
[ 3] 11.0-12.0 sec 950 KBytes 7.78 Mbits/sec 401760.891 ms 0/ 950 (0%)
[ 3] 12.0-13.0 sec 958 KBytes 7.85 Mbits/sec 331040.962 ms 0/ 958 (0%)
[ 3] 13.0-14.0 sec 965 KBytes 7.91 Mbits/sec 1.206 ms 0/ 965 (0%)
[ 3] 14.0-15.0 sec 965 KBytes 7.91 Mbits/sec 0.969 ms 0/ 965 (0%)
[ 3] 15.0-16.0 sec 966 KBytes 7.91 Mbits/sec 0.670 ms 0/ 966 (0%)
[ 3] 16.0-17.0 sec 967 KBytes 7.92 Mbits/sec 310350.798 ms 0/ 967 (0%)
[ 3] 17.0-18.0 sec 976 KBytes 8.00 Mbits/sec 376650.629 ms 1/ 977 (0.1%)
[ 3] 18.0-19.0 sec 960 KBytes 7.86 Mbits/sec 310350.905 ms 0/ 960 (0%)
[ 3] 0.0-20.0 sec 19.0 MBytes 7.97 Mbits/sec 443290.770 ms 17/19463 (0.087%)

Coex of BLE and Wi-Fi throughput both present:

BLE:

Connected as peripheral
Conn. interval is 80 units
LE PHY updated: TX PHY LE 2M, RX PHY LE 2M

====================================================================================
[local] received 916245 bytes (894 KB) in 1851 GATT writes at 365964 bps
Disconnected (reason 0x13)

iperf:

[ 4] local 192.168.50.135 port 5001 connected with 192.168.50.254 port 58935
[ 4] 0.0- 1.0 sec 137 KBytes 1.12 Mbits/sec 20.528 ms 812/ 949 (86%)
[ 4] 1.0- 2.0 sec 766 KBytes 6.28 Mbits/sec 7.157 ms 659/ 1425 (46%)
[ 4] 2.0- 3.0 sec 701 KBytes 5.74 Mbits/sec 4.998 ms 512/ 1213 (42%)
[ 4] 3.0- 4.0 sec 777 KBytes 6.37 Mbits/sec 2.504 ms 430/ 1207 (36%)
[ 4] 4.0- 5.0 sec 538 KBytes 4.41 Mbits/sec 1.287 ms 675/ 1213 (56%)
[ 4] 5.0- 6.0 sec 614 KBytes 5.03 Mbits/sec 2.160 ms 588/ 1202 (49%)
[ 4] 6.0- 7.0 sec 575 KBytes 4.71 Mbits/sec 1.378 ms 645/ 1220 (53%)
[ 4] 7.0- 8.0 sec 472 KBytes 3.87 Mbits/sec 3.358 ms 711/ 1183 (60%)
[ 4] 8.0- 9.0 sec 517 KBytes 4.24 Mbits/sec 4.145 ms 585/ 1102 (53%)
[ 4] 9.0-10.0 sec 741 KBytes 6.07 Mbits/sec 1.522 ms 604/ 1345 (45%)
[ 4] 10.0-11.0 sec 736 KBytes 6.03 Mbits/sec 2.514 ms 461/ 1197 (39%)
[ 4] 11.0-12.0 sec 644 KBytes 5.28 Mbits/sec 12.449 ms 575/ 1219 (47%)
[ 4] 12.0-13.0 sec 786 KBytes 6.44 Mbits/sec 1.212 ms 435/ 1221 (36%)
[ 4] 13.0-14.0 sec 662 KBytes 5.42 Mbits/sec 6.753 ms 508/ 1170 (43%)
[ 4] 14.0-15.0 sec 494 KBytes 4.05 Mbits/sec 3.829 ms 730/ 1224 (60%)
[ 4] 15.0-16.0 sec 430 KBytes 3.52 Mbits/sec 2.547 ms 815/ 1245 (65%)
[ 4] 16.0-17.0 sec 406 KBytes 3.33 Mbits/sec 2.107 ms 822/ 1228 (67%)
[ 4] 17.0-18.0 sec 503 KBytes 4.12 Mbits/sec 3.448 ms 587/ 1090 (54%)
[ 4] 18.0-19.0 sec 664 KBytes 5.44 Mbits/sec 52.655 ms 696/ 1360 (51%)
[ 4] 19.0-20.0 sec 559 KBytes 4.58 Mbits/sec 4.586 ms 636/ 1195 (53%)
[ 4] 0.0-20.0 sec 11.4 MBytes 4.79 Mbits/sec 4.368 ms 12486/24209 (52%)

My Question:

Is there some setting can update the performance of Separate antenna? 

Besides, the html not mention the test result of shared antenna. The below is my test result of the shared antenna scene.

Is it normal?

Shared Antenna:

Connected as peripheral
Conn. interval is 80 units
LE PHY updated: TX PHY LE 2M, RX PHY LE 2M

====================================================================================
[local] received 1693395 bytes (1653 KB) in 3421 GATT writes at 686610 bps
Disconnected (reason 0x13)

$ iperf -s -i 1 -u
------------------------------------------------------------
Server listening on UDP port 5001
Receiving 1470 byte datagrams
UDP buffer size: 208 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.50.135 port 5001 connected with 192.168.50.254 port 52985
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 3] 0.0- 1.0 sec 715 KBytes 5.86 Mbits/sec 138.255 ms 175/ 890 (20%)
[ 3] 1.0- 2.0 sec 593 KBytes 4.86 Mbits/sec 4.006 ms 638/ 1231 (52%)
[ 3] 2.0- 3.0 sec 561 KBytes 4.60 Mbits/sec 7.797 ms 576/ 1137 (51%)
[ 3] 3.0- 4.0 sec 451 KBytes 3.69 Mbits/sec 12.480 ms 707/ 1158 (61%)
[ 3] 4.0- 5.0 sec 633 KBytes 5.19 Mbits/sec 6.619 ms 464/ 1097 (42%)
[ 3] 5.0- 6.0 sec 497 KBytes 4.07 Mbits/sec 3.737 ms 667/ 1164 (57%)
[ 3] 6.0- 7.0 sec 558 KBytes 4.57 Mbits/sec 7.264 ms 533/ 1091 (49%)
[ 3] 7.0- 8.0 sec 550 KBytes 4.51 Mbits/sec 1.958 ms 603/ 1153 (52%)
[ 3] 8.0- 9.0 sec 817 KBytes 6.69 Mbits/sec 0.944 ms 212/ 1029 (21%)
[ 3] 9.0-10.0 sec 617 KBytes 5.05 Mbits/sec 2.880 ms 453/ 1070 (42%)
[ 3] 10.0-11.0 sec 629 KBytes 5.15 Mbits/sec 6.983 ms 474/ 1103 (43%)
[ 3] 11.0-12.0 sec 475 KBytes 3.89 Mbits/sec 13.864 ms 676/ 1151 (59%)
[ 3] 12.0-13.0 sec 415 KBytes 3.40 Mbits/sec 5.453 ms 769/ 1184 (65%)
[ 3] 13.0-14.0 sec 705 KBytes 5.78 Mbits/sec 6.957 ms 340/ 1045 (33%)
[ 3] 14.0-15.0 sec 664 KBytes 5.44 Mbits/sec 9.268 ms 300/ 964 (31%)
[ 3] 15.0-16.0 sec 574 KBytes 4.70 Mbits/sec 3.026 ms 712/ 1286 (55%)
[ 3] 16.0-17.0 sec 431 KBytes 3.53 Mbits/sec 636.114 ms 735/ 1166 (63%)
[ 3] 17.0-18.0 sec 526 KBytes 4.31 Mbits/sec 1.221 ms 587/ 1113 (53%)
[ 3] 18.0-19.0 sec 629 KBytes 5.15 Mbits/sec 5.600 ms 483/ 1112 (43%)
[ 3] 19.0-20.0 sec 462 KBytes 3.78 Mbits/sec 11.181 ms 680/ 1142 (60%)
[ 3] 0.0-20.0 sec 11.2 MBytes 4.71 Mbits/sec 8.390 ms 10784/22291 (48%)

[ovrebekk]

Hi 

All your test seem to show a bit low results, whether you have BLE running or not. 

Are you testing with a standard nRF7002-DK, or using different hardware? 

What is the range between the nRF7002 and the AP when you run the test? 

Are you using the default QSPI interface between the host and the nRF7002, or could there be some differences here that could affect the throughput? 

Best regards
Torbjørn

[Camelia1204]

Hi ovrebekk,

I run the sr_coex sample in a shielded box using a router and the nRF7002-DK, whose distance is closely

In the coex scene which all BLE and WiFi being present, I use the nRF5340 DK as the peer ble device and the test step is just as https://developer.nordicsemi.com/nRF_Connect_SDK/doc/latest/nrf/samples/wifi/sr_coex/README.html

I use the raw sr_coex sample just modify the following settings. Is there the default QSPI interface? 

CONFIG_SYSTEM_WORKQUEUE_STACK_SIZE=4096

CONFIG_NET_PKT_RX_COUNT=32

CONFIG_NET_PKT_TX_COUNT=32

CONFIG_NET_BUF_RX_COUNT=32

CONFIG_NET_BUF_TX_COUNT=64

CONFIG_STA_SAMPLE_SSID=xxxx

CONFIG_STA_SAMPLE_PASSWORD=xxxxxx

CONFIG_NET_CONFIG_PEER_IPV4_ADDR=xxxx

[ovrebekk]

Hi 

It is quite common to get a bit lower readings because of differences in the test setup and environment, but in your case the results seem a bit too low to be caused by these differences alone. Especially if you are testing with a standard DK, and you are testing it in a noise free environment. 

I am a bit curious why you made those changes to the configuration, regarding the TX and RX counts. Did you try with default settings also to see what results you get? 

Camelia1204 said:

Is there the default QSPI interface? 

The interface between the host CPU and the nRF7002 can be either QSPI or SPI based, and QSPI provides the best throughput. If you are running a standard nRF7002DK then the QSPI interface will be used by default. 

Best regards
Torbjørn

[Camelia1204]

Hi,

The reason for "making those changes to the configuration, regarding the TX and RX counts" is that I refer to the following Q&A.

 RE: Run throughput test using iperf2 TCP on nrf7002 DK 

The Q&A is about how to improve the throughput of Wi-Fi.

Is there some special POINTS about connecting the router and test PC? 

[ovrebekk]

Hi

And did making those changes have any effect on the performance?

If I am not mistaken Marte suggested those values because of an issue with dropped packets.  

Can you let me know what kind of router you are using for your test? 

I will try to replicate the test myself and see what results I get. There are no more instructions than what is available in the sample documentation, the instructions have to be general since you might use it with any PC and router combination. 

Best regards
Torbjørn

[Camelia1204]

Hi ovrebekk,

The router of my test is "ASUS RT-AX88U" and I place the router, the nrf7002 DK and the nrf5340 in the Shielding box.

Thanks.

[ovrebekk]

Hi 

Sorry for the slow response. I have been stuck in some other projects lately, but I will try to do some testing on my side and get back to you by the end of the week. 

Best regards
Torbjørn

[ovrebekk]

Hi again

I have now run some testing on my own. I don't have a shielded box available, but I have run the tests at home where there should be relatively little interference from other sources. 

My results look very similar to yours. When I test in WiFi only mode I get a throughput very close to 8Mbps. When I test BLE and WiFi concurrently the WiFi throughput is lower obviously, but I think the main question here is why the base throughput in WiFi only mode is not as high as the documentation claims. 

I tested with and without the recommended values from the other case but I can't spot any noticeable difference in throughput. 

I will check with the developers tomorrow if they have some input on this. Possibly there have been some changes in later version of the SDK that have affected the throughput, or there are some other configuration settings that have to be changed somewhere. 

Best regards
Torbjørn

[Camelia1204]

Hi ovrebekk,

Thanks for your reply. And I look forward to your analysis.

[ovrebekk]

Hi again

Digging into this some more I can't really find any obvious reason why your numbers are lower. I have spoken to people who get similar results as you, but also people who get results similar to those in the documentation without making any configuration changes. It might be related to the Wi-Fi router used in the test, how it is configured, which firmware version it is running etc. 

What happens if you try with a higher CONFIG_WIFI_ZPERF_RATE configuration, such as 20000 or 30000? 

Best regards
Torbjørn