Hacker News Reader

Comment by Reason077 | Hacker News Reader

> "If everyone coordinated and reduced their WiFi power to just what was necessary to cover their unit, it would result in less interference and higher net data rates for everyone. But no one is incentivized to do take action unilaterally."

This is something that needs to be fixed in future WiFi specifications and enforced via device certification. You're lucky if a typical WiFi base station operator understands how to set channels, let alone adjust TX power levels.

People expect to be able to plug in a base station and have it "just work", so coordinating with other nearby stations to optimize spectrum usage, minimize TX power, etc is something that needs to be automatic.

View on HN

The current situation is particularly disappointing when you realize that a lot of WiFi devices have been capable of adjusting TX power on a per-packet basis (per frame, really) at least as far back as 802.11n hardware. So your router should never be shouting at your laptop that's in the same room, even if it would use full transmit power to reach a device in the attic/basement/yard.

>at least as far back as 802.11n hardware.

This made me realize how long it's been since I actually bought a router. The last time I was shopping around, getting 802.11n significantly narrowed the options.

I'm impressed. The only good router I have left from that era has several Ethernet ports burned out, and the rest have too little storage and RAM to accommodate a recent OS with features like IPv6 support. And those are purchases from an era where 802.11n support was a given but dual-band 802.11n support was still reserved for more high-end products.

... and to make matters even worse the solutions being marketed for fixing a lot of these problems are... MORE ACCESS POINTS!

Every time a friend tells me about getting a mesh system to cover dead zones in their place I cringe at the thought of more 2.4 GHz blasting everywhere. 99% of the time the problem can be solved by better placing their current access point, whether that means moving their ISP provided router/AP, or installing a standalone AP and disabling whatever they're currently using for WiFi.

I don't really know the answer to fixing the issues but it seems to me it's not technical, it's better information about placing the devices. End users just want it to work but don't have the time to learn how to best do that. ISP installers are not incentivized to care either, install as cheap and quickly as possible and get out.

Mesh definitely isn't the answer though. It's the anti-answer.

But it’s part of the answer isn’t it? As I understand it the answer in general is more, smaller access points each with less tx power. My wish would be that those mesh extensions wouldn’t then rebroadcast the signal at max 2.4ghz power but then you’d be pilloried by ignorant YouTube reviewers who rank solely on “how many bars at max distance”

In most cases I feel your best position is in the ceiling at the center point of the house. I did that myself but at significant cost - fishing cat5 thru walls and ceilings and repositioning the incoming wires to the basement. 99.9% of users will never take that approach.

That could be true. However, it's better to use wired backhaul than more WiFi for the backhaul. Commercial locations do many access points with smaller broadcast area but use copper to bridge everything together.

You're right about the location aspect though. 99% of people will not be bothered. How do we change that? The ISPs could be incentivized to do higher quality install work perhaps? I'm not sure what the answer is.

The eruption in popularity of pro-sumer installs using gear like Ubiquiti gives me some hope that slowly people will see the way. I've installed UAPs at 5 homes now over the last 6-7 years. Zero support calls.

Wired is an option if you own a house. For everyone else it's a no-go for ethernet unless you're ok routing it outside the walls. There's ethernet over power lines, but whether that works depends on your electric installation.

Back in my last apartment we did powerline ethernet and it worked ok so long as the pcs weren't on the same electrical circuit as the ethernet adapters. I think the huge power draw from the gaming pcs was causing interference with the adapter

Ah, interference is an issue too. I was actually thinking of wiring that makes half of the house completely separate. (Seen it in practice, not sure how that works on the wiring level)

That split comes from US residential electricity being 240V split-phase AC. Adjacent circuit breakers are on different phases that each provide 120V, and double-wide circuit breakers allow appliances like dryers and ovens to use the full 240V difference between phases. If your house (like mine) puts all the lighting circuits on one phase and all the 120V outlets on the other phase, then you'll probably have relatively little trouble with powerline networking going beyond a single circuit breaker. But if you're trying to get powerline networking to work between outlets on different phases, you'll have more trouble (but I've seen it suggested that it might work better while the dryer is running and providing more coupling between phases).

Side note: the transformer on our street once had one of its phases fail. All the outlets in our house dropped to some useless voltage like ~30V AC, but the lights worked fine. Next door, the lights went out but the outlets worked fine. The light in the clothes dryer came on at half brightness. This was very puzzling, not the least because we were all used to issues with that transformer knocking out power entirely when it was hit by another teenage driver. Most of us didn't understand until then how a transformer could fail half-way.

After a windstorm my outlets weren't working but the lights were. Took me a while to figure out what was going on, but one of the hot wires had come loose at the weatherhead. The other hot and neutral were still connected. PG&E came out to fix it within an hour.

Maybe try MoCa as the backhaul, if you already have cable tv lines run?

I gave some thought years ago about “leasing” a ubnt setup to wean people off renting modems from their isp. Couldn’t make the math work tho.

I have ubiquiti equipment and wired back haul too. But even among my tech nerd buddies it’s hard to convince people that its worth spending the time and money to run cable and multiple aps.

For their part isp provided equipment has improved dramatically. They must make a killing on the rental fees. Plus they have a vested interest in reducing WiFi service calls as well.

> But it’s part of the answer isn’t it?

Before: One flaky wifi connection between AP and client.

After: Two wifi connections, both of which must work. Using a combination of operating conditions, AP and client that have proved unreliable in the past.

Now with no visible status for the middle link, and a second DHCP server on the repeater, handing out conflicting addresses when the middle link is down.

The typical home setup guide tells you to disable DHCP on your repeater and delegate to the original box. I was told to do that when I set up something similar due to living in a U-shaped apartment with no nice central position for the router.

Naturally DHCP is handled by the original network when the original network can be seen, yes!

But if the original network is unavailable - perhaps the SSID has changed for example - you need to be able to get to the repeater's configuration page somehow. And the user can't google their problems or download any configuration tools while their connection is down, so you'd better make it simple. So many devices like this will be transparent under normal operation, but if there's a network problem will start their own DHCP server.

And depending on how they detect “down” they may also start the DHCP server under normal conditions.

But mesh means a few plugged in small devices. Your soluton means calling yours ISP to have the fiber come to a different place in your house and/or buying another router, having to pull cables (=drilling, cables,...) which is not allowed in some cases (when renting), pulling power cords (if there are no outlets at the optimal position), and having a blinking box in the middle of your dining room.

And mesh... it "just works"

> 99% of the time the problem can be solved by better placing their current access point, whether that means moving their ISP provided router/AP, or installing a standalone AP and disabling whatever they're currently using for WiFi.

This argument is specious and unrealistic. 99% of the time you cannot move the current access point. Everyone tries that first, and it didn't worked. You're stuck with the fiber AP in a corner of your home, and you're not going to place your WiFi AP as your dinner table's center piece. And reinstalling APs is far more expensive than adding a repeater.

I mean, are you aware that a WiFi repeater nowadays goes for less than 20$? You order it, get it immediately delivered to your home, plug it in, press a button. Bam. Problem solved. No recabling, no technician, no appointment, no wasted afternoon. nothing. And you don't have to go through the WiFi seance session whenever you move your laptop. Just plug in the repeater and you're done.

I agree. Cabling homes is not easy as it seems when you're settled in. Also not every home is built the same way so, you may need to drill a lot of concrete walls.

Repeaters are no use with thick walls (I have quite a few). Powerline + access point combo takes a lot of place and is not cheap. Rewiring home is not always possible. Also, not all APs have power settings so you can't optimize power the way you want (also, thick walls). So it's a hard problem.

Throwing "just re-wire it, eh?" into the middle is not helping.

> People expect to be able to plug in a base station and have it "just work", so coordinating with other nearby stations to optimize spectrum usage, minimize TX power, etc is something that needs to be automatic.

... isn't it ? my internet box is from something like 2014 and is able to choose the less used wifi band automatically.

Can it choose to transmit quietly to avoid swamping nearby devices, or does it run the antenna at maximum transmit power?

It won't reduce its power to make nice with the neighbors.

Power usage is based on the worse-case attenuation scenario, the last thing ISP want are call asking why wifi don't work good in the bedroom if the router is on the other side of the apartment / house. Thus transmitting at max allowable power is the best solution.

> Thus transmitting at max allowable power is the best solution.

Transmitting at the max allowable power is the selfish and dumb solution. It isn't the best solution, in a world with multiple overlapping WiFi networks. It isn't even a Nash equilibrium.

Why do you think it isn't a Nash equilibrium? Any actor unilaterally choosing to transmit at lower power is going to experience worse signal quality, even if it's only marginally worse in some cases.

> Any actor unilaterally choosing to transmit at lower power is going to experience worse signal quality, even if it's only marginally worse in some cases.

Yes, your received signal to noise ratio for that transmission will be somewhat decreased, and that might (or might not) hurt data throughput meaningfully. But it also means that your transmissions are causing less interference for your neighbor(s), and as a result your neighbor's equipment will not have to send as many retransmissions that potentially interfere with your own communications. It's a second-order effect, but also one that's proportional to how many neighbors you have in range that are trying to use the same slice of spectrum. So it's entirely possible that by unilaterally shouting less, you cause a significant reduction in wasted airtime that benefits both sides (though probably benefiting your neighbors more than you, depending on where your priorities lie between throughput and latency). It's not always a zero-sum game.

In general (at least for WiFi), it's better to worry about airtime as the scarce resource rather than focusing on bytes per second. For fixed-rate wired networking, transmit time and bytes delivered are exactly proportional, but a variable-rate shared wireless medium decouples the two.

While you are right this is still a Nash equilibrium. Nash equilibriums aren't supposed to be "good", they just are equilibriums (ie any actor is worse off by deviating)

> (ie any actor is worse off by deviating)

I don't think you quite understood what I was saying: that decreasing your transmit power can actually lead to better overall network performance for you (and your neighbors) without requiring any corresponding change in strategy from your neighbors. Therefore it is not always an equilibrium (this may depend on the severity of local congestion).

Only if every ISP cooperate and there is no single one bad actors (eg. individuals) boosting their own transmit power for their own good.

Also, TX power is not all about interference: attenuation, there is very good reason to transmit at max power for best wall penetration / range. Even in a suburban/rural setting, with more and more devices connected remotely, you need the range. It's not in a manufacturer best interest to decrease the range of their routers. Only a small fraction of user will look at setting the TX power / investing in a better antenna / setup mesh network. The first reflex will always be "buy a new router".

So once again, max legal transmit power is the best overall bet.

> Only if every ISP cooperate and there is no single one bad actors (eg. individuals) boosting their own transmit power for their own good.

Could you please try to re-write that sentence in a way that gives me some indication that you actually read and understood my last two comments? Because it seems like you're simply repeating an assertion without addressing any of the points I brought up, and I'm not sure whether my explanations were inadequate or just unwanted.

The points you brought up are irrelevant to the "effective range" argument. Range requires TX power, so maximum legal range is inherently the same as maximum legal TX power.

In fact, range (power) is a sales feature with more power being better.

For frequency use, this already happens with automatic channel selection for neighbours, and DFS for avoiding a regulated frequency as soon as a signature is detected. WiFi 6 presents more efficiency with at a higher layer with BSS coloring to mitigate crowded channels.

But as you said, most consumer wireless access points are still shouting over each other by default and it needs to be in the specification because there's zero incentive for a manufacturer to "worsen" their RF performance.

It's a pointless consumer market IMO, most people don't need a Wireless LAN. Mobile carriers already provide Internet access to thousands of devices with just a handful of radios over several city blocks, but pricing means many homes end up standing up their own emitters for exactly the same purpose. Now we see bickering like this over what should be a far better coordinated EM spectrum.

Worth note, most business/enterprise gear, which typically gets installed professionally, usually defaults to adjusting TX power automatically because more often than not it benefits the RF environment.

people very often need to setup their own WAP because cellular providers sell metered connections on the order of GB/month, whereas comcast (the most prevalent ISP in america) meters at TB/month. in addition, not every device currently has the radio / components necessary to receive cellular data

also there's the limitation that cellular providers often don't allow tethering (or more generally data access outside of the device the sim card is slotted in)

What I was getting at but forgot to include in my comment. Unfortunately it’s still a distant dream of mine that all carriers, fixed or mobile, would simply provide a dumb pipe billed based on max throughput and nothing more.

But if your apartment is large, and you need coverage in your bathroom at the other side of it, you need more power.... this can't be automated, and manufacturers prefer higher range to users calls to support and helping them raise the power manually (especially when compared to an old AP, which had that range).

What about 802.11h? In addition to defining DFS, it also provides for Transmit Power Control (TPC). I was under the impression that it's an FCC requirement for 5Ghz 802.11a Wifi devices.

Here, it appears to be working as expected. I see messages like "Limiting TX power to 27 (30 - 3) dBm as advertised" in the kernel logs of my linux devices.

Is TPC not well supported?

edit: Whoops, I think TPC isn't really that interesting, and I'm confusing it with DTPC which is Cisco-proprietary.

Ummm... All WiFi devices adjust bitrate dynamically...

For a given amount of data to be transmitted, adjusting bitrate up and freeing the channel quicker is equivalent to using a lower power for more time. Either way the disadvantages to other spectrum users are equal.

> For a given amount of data to be transmitted, adjusting bitrate up and freeing the channel quicker is equivalent to using a lower power for more time.

It's not an exact equivalence. It is often possible to reduce transmit power without having to also significantly reduce bitrate. See eg. https://blog.linuxplumbersconf.org/2014/ocw/proposals/2439

Shanon says that it is an exact equivalence, provided transmit power is >> thermal noise (it is)... If in real world devices that isn't the case, that's a shortcoming of the devices or protocol.

Well, obviously in the real world, devices do not have unbounded transmit rates. But you can get pretty damn close to the transmitter. The Shannon limit isn't the only relevant limit, and is no reason to dismiss out of hand an entire class of optimizations that do help in the real world.

(Also, even if your equipment is capable of hitting the Shannon limit, I'm not convinced that transmitting at the highest power level is going to be optimal for multi-party communications with geographically distributed APs and stations. The information channel between my neighbor's AP and his laptop is neither identical to nor completely separate from the information channel between my AP and my laptop.)

> If in real world devices that isn't the case, that's a shortcoming of the devices or protocol

Yes, welcome to the real world. There isn’t any protocol/hardware in use that scales up bandwidth with signal indefinitely.

Reply on news.ycombinator.com