Does OFDMA Really Work? Part 1

[thiggins]

We resume our quest to find OFDMA's promised benefits, by first trying to find a good benchmark test.

Continue reading on SmallNetBuilder...

 

[Razor512]

Can the performance of OFDMA be improved via firmware updates on the router side, or will it largely remain inconsistent in its results, especially with congestion and latency?

 

[Easy Rhino]

what a mess. how carefully did you retest to see if some of your tests were going backwards?

edit: I realize you were working hard on this of course, I just wonder if there's some subtle testing error, or indeed if manufacturers have tried testing this ever.

 

[sfx2000]

Interesting article - and I tend to agree here... Much like MU-MIMO on 802.11ac Wave 2...

Most gains, IMHO...

• 2.4 GHz - 802.11n
• 5 GHz - 802.11ac Wave 1

Should try 802.11ac Wave 2 across multiple client STA vendors (Realtek, QCA, Broadcom/Intel) for Wave 2 11ac...

Pretty much given up on that unicorn - try as I might, cannot make that work as a MU-MIMO group on QSDK on QCA, and OpenWRT on others for the AP side.

 

[sfx2000]

Can the performance of OFDMA be improved via firmware updates on the router side, or will it largely remain inconsistent in its results, especially with congestion and latency?

Not Likely...

OFDMA 0n the uplink is really hard to do - that's why Wimax (802.16e) failed for 4G compared to LTE.

LTE did preformed SC-OFDMA there.. that's why you have 4G-LTE phones, not Wimax or 3GPP2-UMB handsets,

It's complicated - been there, done that, IEEE 802.11 seems to ask the question over and over again, getting the same results.

 

[thiggins]

Can the performance of OFDMA be improved via firmware updates on the router side, or will it largely remain inconsistent in its results, especially with congestion and latency?

It's possible OFDMA performance can be improved by firmware upgrade. But all the experimentation I've done has shown gains are very dependent on traffic conditions and mix.

So don't get your hopes up.

At this point, the main improvements AX provides are higher 2.4 GHz throughput due to new packet handling protocols and higher link rates. In areas where users can access DFS channels, 160 MHz bandwidth will provide higher 5 GHz throughput. But I'm not a fan of eating up so much spectrum for the sake of bragging rights.

Most devices do not need Gigabit throughput except when doing speed tests and downloading very large files.

 

[thiggins]

what a mess. how carefully did you retest to see if some of your tests were going backwards?

Backwards from what?
I've done a lot of OFDMA testing, over many months. I have found specific cases where OFDMA provides its advertised advantages.

I've also discussed these and other test results with router manufacturers and tried their suggested tests, with mixed results.

As I said, if anyone has other test scenarios to suggest, I'm open to them.

 

[RMerlin]

I suspect OFDMA is more suitable for DOCSIS (DOCSIS 3.1 uses it), where you have to manage multiple simultaneous channels, while Wifi is pretty much one single channel, so the only real benefit would be IMHO from the encoding performance (how much data can a carrier wave carry versus QAM).

So, probably a case of a good technology, but not used in a context where it will make much sense.

 

[iwod]

It's complicated - been there, done that, IEEE 802.11 seems to ask the question over and over again, getting the same results.

Any idea why they keep doing that?

One reason I kind of wanted Qualcomm's MuLTEfire is that it simply works. Not sure if they have plan for 5G version of that.

 

[sfx2000]

Any idea why they keep doing that?

One reason I kind of wanted Qualcomm's MuLTEfire is that it simply works. Not sure if they have plan for 5G version of that.

I haven't see any evidence that MuLTEfire is actually deployed...

 

[Cool]

I just notice the latest firmware for Asus AX88u "Version 3.0.0.4.384.9107" have the option for OFDMA Upload and Download in both 2.4G +5G and with MU-MIMO. I am still thinking if enabling them both would make my wifi any faster.

 

[L&LD]

@Cool, no point theorizing. Just set, reboot (both router and your testing client which shouldn't be a handheld mobile device), wait 10 minutes for the router to fully stabilize, test, and keep good notes. Change the settings and repeat as needed. Your notes reported back here will be appreciated by all.

 

[thiggins]

Enabling OFDMA won't speed up your network, even if you have an AX device. I'll be publishing an article on Monday that backs this up.

 

[KidJoe]

@thiggins on a side note, in the article you mentioned picking up RT-AX58U, RAX15 and RAX45, will you be performing any other testing or writing up any reviews on them at some point, beyond just OFDMA?

 

[thiggins]

@thiggins on a side note, in the article you mentioned picking up RT-AX58U, RAX15 and RAX45, will you be performing any other testing or writing up any reviews on them at some point, beyond just OFDMA?

Yes. I hope within the next few weeks.

 

[waingro]

Awesome work man. Disappointing to know MU MIMO and OFDMA are basically not beneficial. Hopefully, BSS Coloring has real tangible benefits, because I live in an apartment block with 50+ visible networks.

P.S. I really look forward to your 160mhz channel bandwidth test with the Intel AX200.

 

[mjc775]

My Netgear RAX80 appears to support OFDMA when "Enable AX" is checked - see screenshot of advanced wireless settings.

 

[Krisbi]

Should [the article] already online? Currently only google cache version is available.

 

[thiggins]

My Netgear RAX80 appears to support OFDMA when "Enable AX" is checked - see screenshot of advanced wireless settings.

How do you know that it supports OFDMA? That checkmark means only that AX rates are supported, not OFDMA.

I checked with NETGEAR to confirm the RAX80 doesn't currently support OFDMA. A new firmware is being released shortly that will enable it.

 

[thiggins]

Should already online? Currently only google cache version is available.

No. The article was taken down after only being up for around 15 minutes, due to a flaw in the data. I removed your link to the cached version since it has invalid data.

New version will be posted today.