Hacker News Reader

Comment by giantg2 | Hacker News Reader

I'm turning into a luddite - why do we really need this? Just slightly faster transfers?

Keep in mind that, unlike cabled Ethernet, WiFi is a shared medium. Faster transfers also means your device uses less airtime for the same transfer speed. This is especially relevant in dense urban areas, where it's not unusual to see tens or even hundreds of WiFi networks sharing the same few channels.

Is it also good for power for small devices? Can power down for longer between sending.

Eh, seems like the wrong solution if that's the real problem. Building in some signal strength/topology intelligence to weaken the signal would be a better approach.

Congestion, mostly. Really it's a problem at conventions or sports games where every human and probably each pet brings one or two WiFi transceivers. Best way to alleviate that is more bandwidth.

if we just build one more frequency band lane...

Induced demand is good. People just complain about it with cars due to what they claim are excessive environmentally detrimental externalities.

Apart from the environmental externalities, the problem with induced demand and cars is that cars take up too much physical space on the road for the number of people they tend to carry, so induced demand in any reasonably populated area (like a city) will always produce excessive congestion, and you can basically never build enough lanes to fix that (especially because intersections are often a bigger problem).

Whereas with more space-efficient transit, like trains, trams (streetcars) and buses, you can often increase capacity on existing lines or build more capacity for reasonable cost and space, so induced demand isn't as much a problem.

That's not bad though, by the preferences of the people driving the cars.

I've done a couple very informal low sample count surveys, and as far as I can tell the result of pricing in the externalities and building what mass transit is realistic (not much, in certain places) would lead to at least a short-term reduction in satisfaction.

> pricing in externalities .. would lead to ... reduction in satisfaction [for those paying]

I agree with everything you said and would add

> That's not bad though

Of course people who aren't paying something would rather not pay it. In the case of externalizing those costs, they're getting an effective subsidy from society at large for something they disproportionately participate in!

I think the complaint about induced demand for roads is that it induces demand in those who have the means and motive to drive. The land those roads are on (and the environment the cars may pollute) is also demanded for other uses like green space or railway or housing.

It's a rivalrous finite resource and the part of society that doesn't benefit from more roads may be the larger part.

To be clear, this is what happens when you pander to car commuters and build out car infrastructure instead of livable cities: https://ggwash.org/view/4315/houstons-cars-vs-rotterdams-bom...

Google some more before/after car infrastructure photos: https://www.google.com/search?q=before+after+car+infrastruct...

So you see it’s NOT just about the environment. It’s about cities being pleasant or dreadful.

People's preferences are heavily influenced by what is available. Ask the average person in a car-centric American city with no (or barely any) public transit, and of course they prefer to drive. Ask that same question of someone who lives in a city with great public transit (e.g., pick any major city in East Asia, or many cities in Europe) and congested roads, and their answer will be very different.

The whole concept of induced demand is really a misunderstanding of the problem.

Congestion indicates transportation capacity is below demand. But once the medium is congested, you can't measure what the demand is; only how long it's congested.

You can reduce congestion by increasing capacity, or decreasing demand. If you have a pandemic and encourage people to stay at home, you'll find that the transportation networks that are still operating have plenty of capacity. If you add another lane to the 405 in LA, you'll find the duration of congestion has a step change reduction, but as population continues to grow, congestion duration will also continue to grow. If you add mass transit capacity, highway congestion duration will likely be reduced too.

When it's a real situation, the question is usually what intervention allows for the most additional transportation capacity given realistic funds availability. Somewhere like the LA area usually takes an all options approach: there is investment in mass transit as well as highways that service mass transit, individual transportation and commercial shipping. Although intentionally reducing population could work too :p

Eh, I think that’s a misstatement of market dynamics. Yes, congestion is stating that demand is exceeding capacity. However…

Congestion is a pushback signal that reduces actual demand, by reducing quality/utility of the actual ‘product’.

Congestion is demand being throttled through the mechanism of reduction in quality/availability, instead of some other mechanism, like increases in pricing (tolls), or reduction in customers (aka population decreases or exclusionary rules) through the natural usage of the system. Due to induced demand. Because there is almost always more ‘low quality’ demand, than ‘high quality’ demand. It’s a type of tragedy of the commons.

In any ‘free’ system, it almost always ends up this way (see crazy long wait times in nationalized health systems!) precisely because the other options are excluded. Just like on free ways.

> Congestion indicates transportation capacity is below demand. But once the medium is congested, you can't measure what the demand is; only how long it's congested.

But congestion is also part of what controls the demand. It's not a "true" demand that you're just not seeing because of congestion. Once the congestion is solved, the demand will increase as that changes the attractiveness of driving.

Since driving then for some people will outcompete bus, cycling, trains etc., they will then shift over to driving even though they had no demand for it earlier. It will then reach a new equilibrium, not far off the old congestion.

Cars are a bad metaphor for radio physics. Car congestion is a bad metaphor for RF channel collision because of a number of things including TDM and the non fatal quality of what we call a collision in radio space.

Induced demand is not inherently good or bad.

6GHz doesn't go through walls so it's pretty difficult for any low-energy user to disrupt others with it. It's better if we can move towards it for this reason.

> 6GHz doesn't go through walls

That will make it useless for a home network. If I need to mount wired access points in every room, I can just use the wired network I already have.

And run those Ethernet cables directly to your phones and tablets.

Oh, it goes through American home walls. Those aren't real after all.

But how are you going to use Ethernet for your phone and smart lights?

Home Wifi is what it is, but this is a big selling point for things like offices. The less Wifi stations that can "hear" each other, the better for all of them.

The way this is attempted with 2.4/5GHz deployments is to put an AP in every room and turn the power way down, but this only helps a little bit.

As far as going through walls, I don't think there's a big difference between 5GHz and 6GHz.

Can be good for high density environments such as student dorms, hotels and hospitals. They can put an AP in each room (or in the wired hotel TV) and not have to worry about crosstalk or turning the signal down to get a high quality connection.

Or if they build one AP wiring point on the ceiling per room for new homes / small offices or rely on the other bands for backhaul in a mesh network.

"Best way to alleviate that is more bandwidth."

I disagree. The best way to fix that is smaller segmentation for less overlap/congestion. Eg use more APs, make the coverage area for each smaller, intelligently adjust power on the devices so they aren't blasting farther than they need to.

We're already doing that, improving with every version. And 5-6GHz is inherently more range limited to start with.

Is there a reason to only use the best way, and not multiple ways working together? There's even a synergy that more intelligent power adjustment and beamforming is most effective when it gets new frequencies free of legacy equipment.

Only sort of. The 5G is running higher power so they can sell whole home routers that make it through the walls. New tech vs old is true, but it seems unlikely that we can just keep adding new bands when it's untenable to tell consumers their devices won't work anymore when there are so many.

> 5G is running higher power

In the US, 2.4GHz and 5GHz have the same max power, 4 watts EIRP. Except for the part of 5GHz that is only 1 watt. Were 2.4GHz routers not using their full power, or are you talking about somewhere outside the US, or what?

6GHz lets some devices in some bands in some locations run at 4 watts. The rest, the "low power indoor" access points, are limited to 1/16 of a watt EIRP per 20MHz of channel width. 350Mhz of spectrum is only available in this mode. To me that sounds pretty good for reducing congestion.

I just used a WiFi analyzer and in my 70 unit apartment building see 31 APs on 2.4Ghz fighting over 3 channels. 40 APs fighting over 5GHz. and exactly ZERO devices using 6GHz which is VASTLY larger than the other two bands.

The 2.4GHz band is tiny and also is just pure noise at this point and the 5Ghz band can be pretty congested in urban areas. 6GHz gives WiFi a huge amount of extra bandwidth. So much that you could use only WiFi to network a dense office of PCs pretty effectively if you used sophisticated smart WiFi access points that effectively allocated the 6GHz channels.

That could also be due to the high attenuation at 6ghz. Unless your neighbor has a device transmitting on it, it’s way less likely you’d see it.

I mean, that sounds like it'd be even better for high-density environments like apartments.

It is - but much worse for doing a census of networks within x feet of where you are standing.

The census is the number of networks close enough to potentially interfere with my 6GHz AP.

It's only a matter of time when 6GHz networks become wide enough that there's going to be similar congestion there as currently on 5GHz. I remember when available 5GHz spectrum was oh so much wider and oh so empty as well; now you can have 160MHz channels there :)

As others had said: the higher the frequency, the less it penetrates walls, so congestion will be less of a problem.

Then why not build solutions to congestion if that's the problem? If you had all APs adjust their power based on the other networks in range, then you could dynamically reduce power output to reduce congestion. If your AP knows you're in a congested area, then it could adjust its behavior accordingly - reduce power, select the least used channel, etc.

> If you had all APs adjust their power based on the other networks in range, then you could dynamically reduce power output to reduce congestion.

Reducing power also reduces the range. It would be very annoying if my neighbor putting a new AP next to the shared wall meant my AP reduced its range to only half of the room, instead of just sharing the bandwidth while keeping the same range like it currently does.

Why on earth would it do that? It would need a lot of congestion, not 1-2 APs.

there are some cool uses cases in the enterprise space: https://youtu.be/7p_AHPIVo_0?si=cgPQWPSwpN-yML-5

main benefit for home use is that 6 ghz is uncongested, which is very very nice if living in a dense neighborhood.

and of course lightning fast wireless file transfers are possible on the LAN now, which could be beneficial to certain people.

6ghz also has a lot of loss in typical building materials, which is relatively good in that there is a lot less inter building noise. It does require more repeaters though for any given network.

I think on net it works out - surely the average number of access points per network is less than 1.1

To anyone reading: please don't use repeaters. If you can, hard-wire all your wifi APs.

What's the reason not to?

Would you prefer hardwiring to "mesh" as well? Obviously hardwire will always have better performance but the convenience and value should also be a consideration.

Not who you are replying to, but more radio transmission and typically less reliability (you’re either using 2x the channels, or multiplexing on the same channel).

In my experience, it can work but is often not nearly as good as you’d expect.

Hardwired at the APs are way more reliable, and faster.

Is there a downside you have in mind?

personally I was looking at replacing laptop wifi cards with BE200s and a new wifi 7 router to get at least 2Gbps for local network file shares sans cables. People also seem to get 4Gbps with MLO enabled. but the 6GHz routers are very expensive right now

The Unifi U7 Pro is $189 with 2x2 WiFi 7 6Ghz. Seemed reasonably priced to me.

thanks, could definitely consider doing an access point instead of full router replacement

seems like this one doesn't currently have MLO but might come in a future firmware update. Though then limited by single 2.5Gbps ethernet port

i believe the “Pro Max” variant of this access point has MLO in the early access branch. I have not tested it out myself yet.

hopefully they roll it out sometime this year!

Your going to be hard pressed to saturate that “single” 2.5g eth port.

Wifi 7 + MLO = 4Gbps >> 2.5Gbps

Those numbers aren't close to being realistic.

I tested out with U7 Pro Max (MLO enabled on EA firmware) and QC WiFi7 card and it topped out at about 800-900Mbps within the same room. Impressive for WiFi still, but reading the WiFi speed numbers on the box and expecting them to actually be achievable wasn't realistic for a decade now.

Idk why you say so. It doesn't take much searching to find receipts for people showing 3-4Gbps. But "realistic" is gonna be very hardware/firmware dependent, proximity to router, etc.

Examples

2 yrs ago hitting 3.9Gbps https://www.reddit.com/r/HomeNetworking/comments/13ipl7n/tp_...

https://www.reddit.com/r/linux4noobs/comments/1hal8bm/intel_... "I have a BE200 in use on KDE neon. Kernel is compiled with the usual wifi7 modules. I hit 4Gbit/s on an MLO network."

https://www.reddit.com/r/telus/comments/1b3xrlf/wifi_7_speed... "WiFi 7 can reach WiFi speeds up to 4 Gbps on my phone"

To saturate 2.5Gbps in a single direction with WiFi 6/7 modulations with 2x2 MIMO (most phones, tables, laptops have 2 antennas), you need a (total) channel width of 200MHz, and that's for devices in the same room as the AP. If you want to saturate it with devices in the next room, you need closer to 240-300MHz, for example: using say you use MLO with 160MHz on 6GHz, 80MHz on 5GHz and 20MHz on 2.4GHz), totaling 260MHz.

Beware that their (Unifi U7) MLO implementation is still in Alpha and Early Access. 2.5GE at the Ubiquiti Wifi-7 APs is a bad joke.

TP Omada at least has 10GE support.

The UniFi E7 also has a 10GE port.

Seeing as the majority of Wi-Fi devices will get absolutely nowhere near the theoretical maximum though, even 2.5GE is unlikely to be saturated.

Tangential: BE200s remain Intel only.

There's a Mediatek and Qualcomm card for Wifi-7 available.

I'm running the Mediatek MT7925E on a AMD with OpenSUSE Thumbleweed distribution without problems.

Why not? There isn't really a negative externality to it (other than not being able to use the band for something else .. which it isn't being used for at the moment).

Reply on news.ycombinator.com