1 June 2022 - In this blog post we present the results obtained in measuring the performance of a 12-stream WiFi 6 router, the Netgear RAX120 with 8x8 on 5GHz and 4x4 on 2.4GHz, market price around 350€. We then compared it with other low to medium cost WiFi6 routers, the Huawei WiFi AX3 (model WS7100) for around 50€ and the Xiaomi Mi AIoT Router AX3600 for around 100€. We also compared it with the Netgear 7800 X4S which, although it is a WiFi5 (Wave 2) router, is a good previous generation device.
Is a WiFi6 8x8 much better than the WiF6 4x4?
These tests have been carried out in the laboratory in Madrid and using our Candela Technologies instrumentation.
Aplicaziones has extensive experience in this type of work, with projects developed for all types of national and international clients.
More information about our solutions for network quality measurement can be found here.
In our previous blog post we explained a bit about the instrumentation mock-up we use for this type of WiFi network testing.
Basically, we use CT523 devices and the LANforge SW from Candela Technologies, with which we can measure the quality of service (QoS) of WiFi, either using protocols tailored to the client/project or following the recommendations of the TR398 of the Broadband Forum. Here is a summary of these recommendations and the tests we perform on the WiFi equipment:
The number in front of the description of the test corresponds to the chapter in which its details are developed in the TR398 document.
In the picture below, you can see one of the equipment to generate WiFi 6 and WiFi 5 stations. As already indicated in some of our previous articles, for some tests we use an attenuator to simulate, with precision, the distances of the clients from the WiFi router to be verified.
Our objective for these tests has been to verify how much better WiFi6 8x8 equipment can be vs. WiFi6 or WiFi5 4x4 or 2x2 devices.
To simplify the measurements, we have only performed the "Range vs. Rate" tests (6.3.1 of TR398), i.e. traffic as a function of distance and the 32 connected stations test (6.2.1 Maximun Connection test of TR398).
The devices compared were:
- Netgear RAX120: WiFi6 4x4 (2.4GHz) and 8x8 (5GHz) https://www.netgear.com/es/home/wifi/routers/rax120/ (we have denoted it as Netgear 70 for this report)
- Huawei WS1700: WiFi6, 2x2 https://consumer.huawei.com/latin/routers/ax3-dual-core/ (Huawei6)
- Xiaomi AX3600: WiFi6, 2x2 (2.4GHz) and 4x4 (5 GHz) https://www.mi.com/global/mi-aiot-router-ax3600/specs/ (Xiaomi6)
- Netgear R7800 X4S: WiFi5 (wave 2) and 4x4 on 2.4GHz and 5GHz https://www.netgear.com/home/wifi/routers/r7800/ (Netgear5)
Attached are some photos of the test equipment.
Photos of the rest of the equipment to be compared can be found in this other blog post.
As always, the results we show you are the best that have been achieved from multiple iterations and in a non-interfering scenario. The actual results we will see in our homes will be inferior.
Range versus Rate (RvR) tests
The purpose of this test is to verify the traffic that an AP can carry depending on the distance at which the WiFi client is connected. In this way we can evaluate, above all, the quality of the RF part of the APs.
To simulate the behaviour of a user terminal, a 2*2 ax client (WiFi6) with downstream traffic is used. For the Netgear5 (WiFi5) the station is ac 2x2.
In the following graphs, on the X axis (abscissa) we represent the level of attenuation that we make to the signal by our instrumentation and on the Y axis (ordinate) the traffic flow that we get (thoughput).
5GHz RvR
The graph for the 5GHz band tests is as follows
These tests have been done in direct vision, so to interpret them you would have to add an additional -35dB which is the signal that is normally measured at the output of the AP. Thus, for attenuations of -30dB it is as if we were at a location of about -65dBm.
The Huawei6 (low cost) is the worst performer. The throughput drops rapidly from -20dB attenuation. The Netgear5, although from an earlier generation (WiFi5 Wave2), performs better. At the time it was a high-end device.
Then we have the Xioami6 and the Netgear70 that play in another league, they are superior to the other two devices and with them we can notice the difference of a good WiFi6 nowadays.
We can see that in direct vision and with very little attenuation (-10dB), the Xiaomi is able to carry a little more traffic (5%) than the Netgear70. From here and up to almost -38dB, the Netgear70 performs better and carries between 5% and 10% more traffic than the Xiaomi. From an attenuation of -40dB (very distant locations) the two devices behave almost equally. This is for signal levels above -75dBm both devices are similar.
It is clear that the 8x8 implementation of the Netgear RAX120 does not take advantage of the possibilities of the number of additional streams vs. the Xiaomi 4x4 rig. Other factors, such as the rest of the HW (amplifiers, filters, etc.) and/or SW may not be finely tuned in this Netgear RAX120.
It is possible that this small difference is due to a limited implementation by Netgear. We have not tested many 8x8 devices. Something similar happened to us a few years ago with WiFi5. Then we had to wait for Wave2 to take advantage of 4x4. Now we will probably have to wait for WiFi6 Release2 APs to see the improvements in 8x8 equipment. Then WiFi7 with 16x16 will follow, but taking advantage of that will also take time.
2.4Ghz RvR
Now testing in the 2.4GHz band.
The Netgear70 (RAX120) is supposed to be a 4x4, but in this band it is not able to outperform a 50€ 2x2 WiFi6 device (Huawei). This is not the first WiFi6 device we have encountered that neglects its performance in the 2.4GHz band. The fact that it is a band with little use does not justify it, as WiFi6 also allows new modulations and in this case they are hardly used. Even an older Netgear WiFi5 device performs better from an attenuation of -20dB.
Multiple STA Performance Tests
These tests consist of generating traffic with multiple stations (the TR398 specifies 32 2*2 clients), which leads to "stressing" mainly the processing capacity of the AP. Since we want to generate maximum throughput, these tests are only performed in the 5Ghz band with ac clients (more realistic test based on the terminals that users have on their premises).
It can be seen that the Netgear70's performance decreases depending on the number of stations connected at the same time. Its performance is only better than that of the Huawei6.
Moreover, it does not adequately distribute traffic between all stations:
There are some stations which, in the case of the Netgear70, hardly carry any traffic at all.
Final conclusions
To study the performance of WiFi6 8x8 we tested a high-end device (residential market), the Netgear RAX 120, and compared it with the performance we obtained with other WiFi6/WiFi5 devices. The goal was to see, above all, if we had better coverage/throughput at high signal attenuations. We wanted to see if an 8x8 would allow us to eliminate the need for repeaters in poor signal locations in the house. The results, however, have not been as expected.
This new Netgear WiF6 8x8 device provides much better traffic rates on a per-signal basis (at 5GHz) than previous generation WiFi5 devices, but has not significantly outperformed a mid-cost WiFi6 device (Xioami6). While it is possible to get slightly better coverage in the 5GHz band, the high price tag makes it perhaps more advisable to use this economic difference to buy a WiFi6 Mesh solution (even if it is 2x2) to ensure good performance in remote locations.
Furthermore, the performance of this Netgear RAX120 in the 2.4GHz band and in tests (5GHz) with multiple simultaneous stations is quite irregular.
It is clear that developing good 8x8 equipment is not easy and/or can be very expensive, this may have led Netgear to make some compromises in the design. In the end we have been left with the desire to see what these new WiFi6 possibilities have to offer. In the future we will test other WiFi6 8x8 devices and see if the designs are more finely tuned.
The analysis presented in this report includes a couple of basic tests of the TR-398 that allow us to compare equipment. Perhaps a more detailed analysis with additional tests would allow us to detect better performance of the Netgear RAX120 for other scenarios (e.g. immunity to interference, stability with associations, ...). In any case, these measurements already indicate that there is still a lot to be done to take advantage of the new possibilities of WiFi networks6 by equipment manufacturers.
Finally, it should be pointed out that these measurements have been taken with equipment purchased on the market by our company and with the latest SW from the manufacturers. It is not possible to extrapolate them to other 8x8 equipment on the market (both residential and business). Other manufacturers may have made more successful 8x8 implementations. This is why we recommend to all our customers that, if they can, they should test the equipment they are going to buy and not be guided by the manufacturers' catalogues.
More information about our solutions for WiFis quality measurement can be found here.