WiFi 7 has been certified.

[RMerlin]

Perhaps someday MLO will deliver the benefits you cite. But not in the first generation for sure. I would hate to be the engineers developing the AP schedulers. MLO adds another set of decisions to make about how and when to send a packet.

I don't know about latency (MLO is supposed to be able to use the lowest latency link in a transmission by using the first one with an available slot), but in terms of throughput I have seen an Intel demo of Wifi 7 pushing things beyond 3 Gbps through the use of MLO.

 

[Tech9]

We won't see 11Gbps on single band/radio for sure. I'm interested in range @320MHz wide on 6GHz. Usable SNR may be limited to the same room only.

 

[L&LD]

It won't matter if I have WiFi 7 client devices or not, I will still test (Gen 2+) or later WiFi 7 routers when they become available and (hopefully) RMerlin supported.

Why? Because each router upgrade I've ever done has always been a performance increase over the hardware I ran before, even when upgrading from N to AC, from AC to AC, and particularly when upgrading from AC to AX classes. Even if the 'specs' say there should be no difference.

Newer/different/better RF designs (including antennae and antennae orientations), newer firmware that more closely matches the future current conditions of proper WiFi behavior, newer SDKs (or newer subversions of those SDKs), along with all the other unknown closed-source/black box upgrades/changes that new products inevitably have.

Not to mention my (decade-old wish) for real router hardware like an Intel-level CPU with greater than 8GB of RAM, with the firmware running on an SSD (with RAM and SSD upgradable/replaceable too). Even without these last improvements, 1GB RAM was huge when it was introduced for the latency improvements of the network. As was bigger internal storage (bigger nand storage = faster nand storage, if done right).

These variables eventually always seem to come together favorably in one or two products that easily outshine any previously available old tech. And the benefits are still seen to some degree even with older clients.

Tech isn't like 'classic cars'. I don't care for visual design, even if I can appreciate it when it's done right. I don't care for the materials used, a cardboard box will do (i.e. see Ikea 'furniture'). I care mostly for functionality. Along with reliability, dependability, and longevity. Proper chipsets, excellent RF, enough resources (CPU, RAM, Storage) to be able to get the router out of its way. And give me and my customers a network experience that is superior to what was possible before.

Yes, it is easy to always upgrade to a better network. Simply test new hardware in your own environment with your own client devices and your specific workflow.

If the new hardware proves superior, buy it. If it doesn't, return it and wait to test the next possible candidate.

The above process allows me, at any given time, to have the best (at 'bang-for-the-buck') network available while being able to donate, sell, or reuse my existing network hardware while it is still worth doing so (i.e. it is not obsolete, yet).

Sitting in your easy chair and looking at specs to determine if a (specific) new product is 'worth it', or if it will give you a superior network over your current setup, is like a caveman reading about physics and trying to justify in applying that 'knowledge', purely in his mind.

You need to roll up your sleeves and get your hands dirty, your credit card charged, your testing methods repeatable (with your own client devices), and your notetaking skills sharp (use Excel).

In addition to how you prepare for the testing too (i.e. flashing the firmware you want to use, performing a full reset, testing at a base level, and adding only necessary features, scripts, and options, if needed).

Yes, you need to have some passion to do the above. But the network you have will be the best one. For you.

And waiting for 'future WiFi clients', not required.

 

[thiggins]

I don't know about latency (MLO is supposed to be able to use the lowest latency link in a transmission by using the first one with an available slot), but in terms of throughput I have seen an Intel demo of Wifi 7 pushing things beyond 3 Gbps through the use of MLO.

I believe you. But I've seen enough demos of new Wi-Fi technology to know that they use special firmware loads and optimal conditions. YMMV.

As to whether MLO uses the lowest latency link, it depends on how the manufacturer will choose to adjust the knobs. There are three basic choices, i.e. optimize latency, optimize bandwidth, optimize reliability. Note that the effective latency of a packet doesn't depend on only the link characteristics, but also how long the packet is queued waiting for the scheduler to decide how and when to kick it loose. Makes my head hurt...

 

[RMerlin]

As to whether MLO uses the lowest latency link, it depends on how the manufacturer will choose to adjust the knobs. There are three basic choices, i.e. optimize latency, optimize bandwidth, optimize reliability. Note that the effective latency of a packet doesn't depend on only the link characteristics, but also how long the packet is queued waiting for the scheduler to decide how and when to kick it loose. Makes my head hurt..

MLO packet scheduling does seem to be quite complex if they have to take into account all of these variables. I assume they may be leveraging the information available through extensions such as 802.11v?

It remains to be seen how much of this will be implemented by the average manufacturers who implement MLO. It wouldn't surprise me to see partial implementations on either the STA or the AP endpoints.

 

[coxhaus]

My iPhone connects at 1200Mbps link speed, or about 800Mbps possible throughput. What work on your phone requires faster speeds? Just curious.

My iPhones also connect at 1200Mbps right now so there is no pressing problem to buy an iPhone 15. I also have APs in our most used rooms so there is no wall degradation in signal strength. I can tell differences in DNS caching using my Dell I9 laptop using wireless. My Wi-Fi works well. Roaming could be better as I miss 1 to 2 words on Wi-Fi phone calls roaming. I hope the new spec will fix that. I had great AC roaming but it is not as fast as AX.

 

[thiggins]

It remains to be seen how much of this will be implemented by the average manufacturers who implement MLO. It wouldn't surprise me to see partial implementations on either the STA or the AP endpoints.

That's how it has been with MU-MIMO and OFDMA. So, yeah, expect, uh, "partial" implementations.

 

[sfx2000]

This is something @sfx2000 is commenting quite often and no one pays attention. He knows Wi-Fi quite well too.

@Tech9 - you're a bit too kind maybe...

Anyways - very happy to see 11be released - there's been a few Router/AP's released based on the drafts, but now with a formal release and WiFi Alliance also jumping in with their certification program updates...

It's a huge amount of work for both IEEE and WiFi-Alliance to get to this point.

That being said - I suspect that things won't be that much different - there's about 20 percent of 11be that is really useful, and this is what most might see...

Using the old format - N300/AC1900 is good enough for most folks - going to WiFi6 isn't going to show that much more unless one is running specific benchmarks from 1 meter away in an isolated RF environment.

There are a few things that really drive WiFi performance - the noise floor (the other traffic on the same channel), and proximity to the AP. The other is the client capabilities, which are much more important than the AP itself..

AX and now BE add a lot of new things, but as I mention, about 20 percent of those are actually useful...

The two biggest improvements for WiFi - OFDM and MIMO - hands down, this is what makes 2.4 and 5GHz WiFi really useful - WiFi4 for 2.4GHz, and WiFi5 Wave 1 for 5GHz.

Yes, we've see some improvements based on the Router SoC's (they are much faster), and better firmware inside the wireless network interface chips themselve, but if I get 10-15 meters away from the AP, and we start to see physics happen - shannon limits are what they are...

If one has a WiFi6 class Router/AP, you're in a good place - the second generation silicon is pretty good, the firmware is well sorted - so it's really the choice of what flavor of OEM/ODM software you like to use - they're all pretty close.

We're seeing very capable WIFi6 devices now down into the sub-$200 range - and some down into the $100 level - this is good, as it's a sign of maturity of tech as it stands going into the first quarter of 2024...

Looking at WiFi7 - there's a lot of new features, but let's sort out the high level stuff...

• 320 MHz channels (5, 6Ghz) - well, yes, it might, but only in the 6GHz band to be reliable, and the range is halved with the doubling of the channel BW - I suspect 80MHz in both 5 and 6GHz will be most used
• MultiLink Operation (MLO) - This just adds a whole lot of complexity for the client side - most clients at 2*2:2 for a single band - MLO either requires one more dual band radio for 2-stream, or one might be able to do 2 single streams for each band
• 4K QAM - QAM4096 - looks nice on paper, but QAM256 in 2.4 is almost impossible to hit due to noise limits, and same goes for QAM1024 or higher in 5, 6GHz - shannon limits apply here...
• 512 Compressed Block Ack - this is a big improvement, but only in a greenfield WiFi7 deployment - legacy rules come into play with mixed mode operation
• MultiRU to Single STA - OFDMA modes only - so legacy kind of breaks this

Just my $0.02 opinion...

 

[sfx2000]

That's how it has been with MU-MIMO and OFDMA. So, yeah, expect, uh, "partial" implementations.

I think it's incremental improvement - while my post above is a bit dismissive, at the same time, these "partial" improvements do help - but not at the level that the marketing hype might suggest.

The biggest improvement in the consumer AP space has not been the radios, but on the core chipsets with throughput enhancements and better interface support, which to some degree has been driven by the availability of higher rate WAN bandwidths..

 

[Tech9]

Yes, we've see some improvements based on the Router SoC's

I'll copy here my reply to user @Itsjustmd15 in another thread.

"This is because home routers have weak good enough for the application and release time CPU and the newer routers come with a better one. I have an x86 hardware firewall with 4-core Intel CPU and 4GB RAM and my existing AC clients work exactly the same way with AC and AX class access points. In both cases I get the maximum from AC."

I often read "my new router improved my old clients" claim and this is less Wi-Fi technology related and more removing the bottlenecks around the old SoC. The cheaper the AIO router the more prominent the difference. The new routers just reveal what really the "old" technology was capable of. My system has no processing power restriction and my APs are high-end class. And this is exactly what happens - the clients are limited by their own capabilities:

The other is the client capabilities, which are much more important than the AP itself.

 

[sfx2000]

MLO packet scheduling does seem to be quite complex if they have to take into account all of these variables. I assume they may be leveraging the information available through extensions such as 802.11v?

It remains to be seen how much of this will be implemented by the average manufacturers who implement MLO. It wouldn't surprise me to see partial implementations on either the STA or the AP endpoints.

I tend to agree - as packet scheduling in MLO needs to go across radios, which means going io the SoC itself for consumer routers...

And MLO will require feedback across all radios and clients in the BSS...

Can't imagine at the moment how this would apply to mesh systesms - I think they would just opt-out there...

 

[coxhaus]

I'll copy here my reply to user @Itsjustmd15 in another thread.

"This is because home routers have weak good enough for the application and release time CPU and the newer routers come with a better one. I have an x86 hardware firewall with 4-core Intel CPU and 4GB RAM and my existing AC clients work exactly the same way with AC and AX class access points. In both cases I get the maximum from AC."

I often read "my new router improved my old clients" claim and this is less Wi-Fi technology related and more removing the bottlenecks around the old SoC. The cheaper the AIO router the more prominent the difference. The new routers just reveal what really the "old" technology was capable of. My system has no processing power restriction and my APs are high-end class. And this is exactly what happens - the clients are limited by their own capabilities:

I see a difference in AX over AC using Windows 11. AX is faster for me. It is just snappier. Windows 10 was not as fast as Windows 11 for me even using the latest drivers under Windows 10.
I am running Pfsense with an Intel 3.4GHz CPU and a Cisco L3 switch using 3 Cisco 150ax APs compared to my Cisco WAP581 AC APs. The Cisco WAP581 pull POE+ power around 25 watts vs 11 watts on the Cisco 150ax APs. So, the WAP581 would go about 4 to 5 feet farther using more power. But for the same power draw you can run more Cisco 150ax APs which makes up for it. Plus, the Cisco 150ax are $102 each whereas the Cisco WAP581 were about $350 each.
So, I don't where you are coming from thinking AC is better than AX.

 

[sfx2000]

I see a difference in AX over AC using Windows 11. AX is faster for me. It is just snappier.

...

So, I don't where you are coming from thinking AC is better than AX.

It's complicated, but your network has client dependencies - you saw Win11 is subjectively snappier - well put some numbers behind that with a Win10/Win11 compare...

The "feel" is nice, but objective results are what matters...

 

[L&LD]

@sfx2000, while I appreciate the fact that things like the Shannon limit exist, it isn't an absolute cap on performance or future improvements, discoveries, or implementations in the (near) future. It is merely a ratio. Increase sensitivity and selectivity, even at the same noise floor and you've blasted past the 'original' Shannon limit of the previous iteration.

Explained: The Shannon limit

A 1948 paper by Claude Shannon SM ’37, PhD ’40 created the field of information theory — and set its research agenda for the next 50 years.

news.mit.edu

Explained: Gallager codes

In 1993, scientists achieved the maximum rate for data transmission — only to find they’d been scooped 30 years earlier by an MIT grad student.

news.mit.edu

 

[Itsjustmd15]

We won't see 11Gbps on single band/radio for sure. I'm interested in range @320MHz wide on 6GHz. Usable SNR may be limited to the same room only.

I've got my BE96Us set on 320mhz 6ghz. My house is approximately 4500sq ft and has 2 separate upstairs areas as one is a man cave above the garage. They are spread fairly evenly and a decent distance apart through several walls and have flooring and/or roofing between them. There's also decorative brick inside the house as well as a brick fireplace. My router is in the main upstairs area as that's where the fiber connection comes into the home. I've got 2 nodes downstairs and one in the man cave. 2 of my nodes correct directly to the router and have dbms of -56 and -61. Another one daisy chains to the -61 dbm node and has a dbm of -66. All of this is per ASUS's firmware. This is roughly what I was getting when they were connected by 5ghz 160mhz but the speeds are much better and I get no interference from the airport I live by. This is connected to each other though.

When I connect my iPhone 15 Pro Max, Z Fold 3, or Pixel 6 Pro to 6ghz, when in the same room I get full bars. I know they're not using the full 320mhz of course, but it's plenty speedy and often faster than 5ghz when in the same room based on speedtests I've done. Go outside of the room though, and they basically drop to half bars (I know this is not scientific, I haven't tested to see the dbm reported). It's still usable but at a certain point it just makes more sense to be on 5ghz. Sometimes I'll be in an in between spot from the nodes and it'll seem a little slow in comparison to my 5ghz. Because of this I typically stick to 5ghz on any of my devices that can do both. The speed difference between the two in the same room makes no difference for the things I do on my phone. It's just noticeable on a speed test.

Honestly, I only used 6ghz when there was a bug with 5ghz where it wouldn't allow the nodes to be used and everything 5ghz was connecting directly to the router instead of roaming to nodes. I was just too far from the router at times for 5ghz to be usable, and 2.4 was painful to use. It was a pain in the butt until I figured out a workaround to fix it since I couldn't rollback to firmware that worked (ASUS kept forcing the update). Guess we are beta testers as you've said. That's my 0.02.

 

[sfx2000]

@sfx2000, while I appreciate the fact that things like the Shannon limit exist, it isn't an absolute cap on performance or future improvements, discoveries, or implementations in the (near) future. It is merely a ratio. Increase sensitivity and selectivity, even at the same noise floor and you've blasted past the 'original' Shannon limit of the previous iteration.

Shannon hasn't changed - we do get closer to the limits over time...

Recall that my 40 year career has been in wireless across many flavors as a design and product development engineer..

 

[Tech9]

This is roughly what I was getting when they were connected by 5ghz 160mhz

Go outside of the room though, and they basically drop to half bars

This is expected. The routers have higher power output, more streams and almost guaranteed better antennas than your phones or other clients. This connection between the routers tells little about practical usability of the 6GHz band. No client can do the same. On the older RT-AC86U the link between two routers was >2Gbps on AC with all 4 streams and TurboQAM at 80MHz wide channel. No AC client could ever come close to this speed and range for the reason above. The same with AX, the same with BE.

 

[sfx2000]

On the older RT-AC86U the link between two routers was >2Gbps on AC with all 4 streams and TurboQAM at 80MHz wide channel. No AC client could ever come close to this speed and range for the reason above. The same with AX, the same with BE.

Perhaps that's why Roaming and Cross band was always kind of broken with Clever Connect and Dumb Roaming

 

[L&LD]

Shannon hasn't changed - we do get closer to the limits over time...

Recall that my 40 year career has been in wireless across many flavors as a design and product development engineer..

@sfx2000, I greatly value and appreciate your experience and superior expertise here.

I am not even remotely at your level.

But yesterday's absolute limits are regularly surpassed with each progress steps tech makes. While that doesn't change the Shannon limit, it does change the absolute possible with enough progress.

 

[Itsjustmd15]

This is expected. The routers have higher power output, more streams and almost guaranteed better antennas than your phones or other clients. This connection between the routers tells little about practical usability of the 6GHz band. No client can do the same. On the older RT-AC86U the link between two routers was >2Gbps on AC with all 4 streams and TurboQAM at 80MHz wide channel. No AC client could ever come close to this speed and range for the reason above. The same with AX, the same with BE.

You said you were interested about 320mhz so I just answered. Did you not want anyone else to comment or something? I didn't say it was unexpected. Just giving my experience. I would think it's obvious to anyone on this forum that a router would do better than a phone lol. In my opinion, what 6ghz is good for is an interference free high speed backhaul.