Make educated wireless router/AP upgrade decisions
duckware.com
duckware.com
Devices will be sending their GPS coordinates and serial numbers to the FCC to verify what frequencies they can use.
It's unclear to me why the serial numbers need to be sent.
TWDR is a high-resolution doppler radar that measures winds on airport approach paths to detect wind shear and microbursts. It's a life-critical system. It would be nice to use the bandwidth for WiFi, and the FCC made every opportunity to allow it, but WiFi vendors didn't play nice.
https://arstechnica.com/tech-policy/2019/05/5g-networks-will...
It is also important to note that the interesting frequency is 23.8GHz, but the spectrum that was auctioned off is from 24.25 to 24.45 and 24.75 to 25.25GHz... so not overlapping. The Ars article links to a primary source [3] claiming that the Navy desires a guarantee that signals outside of the allocated range are attenuated by -57dB instead of -20dB.
Typically, designers of RF equipment do not want to transmit power that their receivers cannot receive. Power is expensive to waste. So given the separation between the 5G band and water vapor, the attenuation intrinsic to 23.8GHz signals when travelling through the atmosphere, and a desire to produce efficient electronics (better battery life sells phones)... I'm betting this won't be a big problem. Obviously, the scientists that rely on the data are kind of obligated to say "never do anything with the RF spectrum, it interferes with our research", but it's up to the FCC to balance their concerns with the reality that someone other than weather researchers need to use the RF spectrum.
Anyway, this is all theoretical, whereas DFS is not theoretical. You could watch it actively interfere with weather radar, so the next generation of WiFi is using something else. Obviously if 5G interferes with weather satellites, they'll probably ask the licencees to turn their equipment off until they can mitigate the degradation of another licensed user.
[1] https://news.ucar.edu/132669/ucar-statement-opening-24-ghz-s...
[2] https://www.theweatherprediction.com/habyhints2/523/
[3] https://www.wyden.senate.gov/imo/media/doc/Navy%2024Hz%205G%...
The FCC needs to do this much more, radio spectrum is a massively valuable limited resource and you either demonstrate technological excellence or you get out of the way of those who can. We can't have GHz worth of spectrum untouched because people don't want to upgrade their 1960s hardware and/or technology.
Further, TDWR was developed in the 90's and has undergone upgrades since then. Don't dismiss it as some Apollo-era PDP-11 sitting in a closet somewhere just because you don't know what it is.
https://en.wikipedia.org/wiki/Wi-Fi_positioning_system
Or potentially to help make the spectrum allocation in an area more efficient? Maybe that information could be used to help position those point-to-point microwave networks mentioned in the article.
The source IP (v4 or v6) address for the record update, along with the timestamp of update, is already sufficiently identifying, and would be obtained even offhand in a query process to check a given area for interference issues.
The requirement to use GPS also practically mandates that these routers HAVE a GPS receiver inside.
My biggest worries relate to the cost and probable barrier to projects like OpenWRT supporting these devices.
Its a compromise - it allows licensed users to continue to operate without interference, but also allows the spectrum to be used without a license in areas where doing so wouldnt create interference.
What guarantees do I have, as someone in Canada, that my device won't be reporting this data to the US federal government for dubious-at-best reasons (especially in the context of international use)?
Recently, I also stumbled upon a Chinese site [1] which does similar teardowns, but only in Chinese.
I used it to upgrade from my overheating and low range Linksys to a strange rare but efficient Synology router and it was a very good buy in the end.
I keep telling people who complain to me about slow WiFi speed that it is their device's fault. A lot of mid range Android phones only have 1x1 802.11ac, which is about 433 Mbps PHY at 80 MHz channel.
Some more expensive Android phones and all iPhones support 2x2 MIMO which gets you up to 867 Mbps PHY at 80 MHz channel.
Some phones have a buggy implementation of VHT80 client, so on 80MHz the PHY rate drops to <100 Mbps, so in those cases I recommend 802.11n 40MHz on 5GHz band.
The AP/Router/Modem combo that all the ISPs give in my country is very much capable of fast WiFi. It has 4x4 MIMO at 802.11ac with 80MHz + 2x2 MIMO at 802.11n up to 40 MHz (I always tell people to avoid 40 MHz on 2.4GHz band!).
On a good phone, this gets you around 600 Mbps with 867 Mbps PHY, on a cheaper device, between 250 and 290 Mbps on 433 Mbps PHY. Devices with 4x4 can get 1.7Gbps PHY, but I do not have any.
This does not stop people from complaining about the free modem they get, demand it is put into bridge mode, buy an ASUS router for >400€, and get the same result. Some (smartly) return the router they bought right away, and request the ISP put the device back into routing mode, but some keep the useless and expensive trophy of stupidity...
You have a TR069 interface that operates on a VLAN, you can disable that interface and it does not have the ability to contact the ISP or be contacted (this is not DOCSIS, I'm talking about GPON or xDSL).
In general all ISP combo devices are firewalled off from the WAN side, don't reply to pings and no open ports. If there was a kernel network stack bug, you can't bypass that with your own router, unless it's xDSL in which case you can buy ANY compatible modem and use it with the PPPoE login you can extract from your ISP combo device using instructions on my blog.
If it's GPON you can't do anything, it's a full Huawei system (ONT and OLT) and it needs to authorize with a serial number over OMCI, which you cannot replicate cheaply.
You can use a MikroTik router and a Huawei SFP+ transceiver, but that is too expensive to justify it for moderate-level geeks :)
But the ISP device is really unreliable - my work laptop can do 3x3 and that causes the router to lock up. It also "smartly looks for least used channel" which can't be turned off in the ui, and causes few minute long disconnects every hour or so. Oh and it overheats and restarts when you load it with a lot of packets, like with BitTorrent or video call with five people.
The router is very much worth it.
It's less a geek or teenager and more your i.e. tech-illiterate parents[1] that I tell this to.
[1] not implying all parents are tech-illiterate, it's a cliche and mean stereotype, I mean "tech illiterate person", but mention this cliche to be clearer
A couple of years ago I bought a Netgear D7000v2 VDSL modem/router and it was a piece of junk. It took minutes to start up after power-on, any attempt to use the USB drive sharing functionality would lock up the router requiring a power cycle, one firmware upgrade broke its ability to sync to VDSL (luckily I still had an old firmware binary handy, so I could revert it) and when I finally threw in the towel on it and switched to the $1 bundled modem that came with my new broadband service, suddenly I was synching at 25% higher speed. Never buying another Netgear product again. They used to be good, dammit.
For my own home, I used to try to optimize the hell out of it (regularly switching channels, trying new routers etc.)
Now I "installed" a TP-Link Deco M9 system. I have 250 Mbps coming in, and at the worst part of my house, my iPhone will still get 100 Mbps from that. That is more than enough even for streaming 4K. So I have now arrived at the "why bother tinkering with a good enough system" stage of WiFi, which is a huge relief.
My rationale for buying used is that basically normal consumers typically do not buy Ubiquiti enterprise products - so anyone who does is someone who has experience with technology and will take care of it well.
I haven't had a chance to test this out yet, but I hope it works!
I picked up an Edgerouter X SFP for $40 on FB Marketplace (retails for $99), the guy selling it is a mobility architect. It came in great condition. After I bought it I found an Edgerouter PoE5 ALSO being sold for $40 - this retails at $275. I don't really need it, but I've thought about buying it since it's so cheap anyways. I'm also going to purchase an AP AC Long Range for $50 - this retails at $109.
The Edgerouters do not come with an AP built in, so you have to buy one separately. I don't mind doing this since I think it's easier to mount an access point than it is to mount a router since you have to worry about less wires (just ethernet if the ap can be poe'd).
For about $90 total I'll have an enterprise router and AP that'll work much better than the cheap sagecom router I'm renting from my ISP.
Be careful buying second-hand electronics, particularly networking kit!
You should re-flash the device's firmware. The OpenWRT website should be able to guide you to stock firmware images for most modern SOHO devices.
And in the case of some of Ubiquiti's gear, you have to reflash to just get started unforunately (my UDM Pro couldn't complete setup without a reflash and my Unifi Switch 8 couldn't be discovered by the new UDM until a reflash). That's a real fun software issue from Ubiquiti.
What you are worried about is backdoored firmware.
If this is too much work, just buy new kit in sealed packages. Saving a couple hundred bucks just isn't worth the risk.
https://community.ui.com/questions/Introducing-the-UniFi-Nex...
> It is easy to overlook and miss, but beamforming and diversity are the key reasons why you want a 4×4 MIMO router even though most clients are still only 2×2 MIMO. The extra antennas are actually used and offer value (a stronger signal, which translate to better connect speeds for some users)!
Having been a keen user of the TP-Link Omada series of APs I went to check what their capability was. No wonder they don't mention anywhere on the site - it's only 2x2 (for the EAP225). I was able to confirm this by using their maximum speed and the access speeds table in the article.
I don't know how much I'm actually missing out on, but I'm going to stop automatically recommending these products now.
[1] https://www.tp-link.com/us/business-networking/ceiling-mount...
[0] https://www.tp-link.com/uk/business-networking/ceiling-mount...
But I paid less than 90€ for it and the 4x4 routers all seem to be quite a bit more expensive. Am I really missing out on much by not going 4x4?
While I love WiFi for my mobile devices or occasionally when my laptop isn't at my desk, I can't recommend to people enough to prefer actual network cables for their devices which are in "fixed" locations, for example your normal work area for your laptop, or where your TV is with consoles or something like an AppleTV.
Once you move at least mostly to wires, you'll find that most (if not all) your network problems just go away.
If you're worried about aesthetics, my laptop has power, networking, screens and everything working through a single USB-C port, so my desk is quite pristine. Around TVs and their connected devices it's pretty much always super easy to hide wires.
Wireless is a shared medium with everyone in range who is also on wireless, it just can't scale like individual wires (very inexpensively) at 1Gb/s speed EACH. So put everything you practically can on wires, that way your airwaves remain largely open as they're only used by things like mobile phones or light browsing on laptops.
I understand that if you're renting then it may not be possible to do wires. My hope is that in the same way that most houses started being built to accommodate cable tv / telephone points, in the future they'll be built to accomodate network points by at least TVs.
Offices trying to be all wireless are almost certainly never going to be able to provide high speeds to everyone for the foreseeable future. Employees at their desks should use wired connections and wireless should only be used for mobile phones or people not at their desk for some reason (like being in a meeting).
However, roaming support is entirely missing. Handover between access points is really important in my experience. It would have been nice to have an overview on the current situation in WiFi.
That's 'cell planning'. Roaming is different:
Let's say you're moving from your bedroom to the kitchen. The bedroom AP's signal suddenly drops. Your smartphone needs to find the kitchen AP, re-connect, and re-transmit all the packets that might have been lost.
Mobile networks assist the smartphone: it tells the smartphone that the kitchen AP will soon be in range. It already allocates and configures resources at the target AP. Then the handover is more seamless.
In theory, WiFi can do this too. But last time I checked, it seemed a vendor-specific mess.
WiFi actually can't do this. It's a shame, but fully network-directed roaming is not a thing in 802.11 yet, and I don't know when it's going to bother to show up.
The major protocol extensions you can use are 802.11k, 802.11r, and 802.11v. They have a compounding increase in roaming performance in roughly that level of significance.
802.11k is a pure win. It should always be used.
802.11r is hit or miss. When it works, roaming can take under 10ms. When it doesn't, you disassociate entirely and reconnect. I have an SSID with it enabled but never use it.
802.11v is a pure win, but not very well supported.
I use Cisco hardware specifically for this feature set, but I can't really say I recommend it. I kinda hate the controller software, and it's licensed separately. Cheap hardware on eBay though.
Do those extensions require client support, and if so, is it supported by consumer devices?
iOS supports them all: https://support.apple.com/en-us/HT202628
Mac OS doesn't support any of them (much to my chagrin).
On Windows, it has to be implemented by the driver. I suggest using Intel: https://www.intel.com/content/www/us/en/support/articles/000...
With Android, it's hit or miss. All modern flagships support them AFAIK. Going back in time, it's limited to premium vendors. I'd expect the majority of devices to support 802.11k, although they're quiet about it. You'd have to check logs on your AP (if available) to determine if it's in use.
Definitely not consumer friendly to configure using a java webbased application but I never had to reboot or change anything.
They just work day in and day out and the family is happy.
True, but it's actually the slowest supported/enabled modulation, which can be increased. In a crowded 2.4GHz environment (1Mbps min. modulation), raising it to 12Mbps or more can cut 50+% fluff (SSID, ARP, data broadcasts..) down to <5% fluff.
> 2166 → 1083: Client MIMO > You must use the minimum MIMO common to both devices
MU-MIMO should enable the router to split its numerous MIMO channels to multiple devices with less MIMO channels. I have no idea whether it actually works.
For 802.11ax, not sure there's anything available in the openwrt or mainline kernels yet. You might end up with something with a binary blob and you can't upgrade your kernel.
Yes, they still phone home. I don't believe there's an option in the UI to turn it off -- I just blocked the AP's access to the Internet on my firewall.
I use a TP Link 1750AC cheap router($60) that seems work perfectly fine for the whole house for streaming and surfing.
My internet is about 20mbps down and 6mbps up.
I never felt the need that my bandwidth is too slow somehow. What's the super 1Gbps wifi + ISP for? I never need it, maybe until I'm doing VR on Internet some day.
Transfer large files inside the house over wifi is the only time I felt wifi is not fast enough, but that happens once a year, for huge size files I can do USB transfer anyways.
In short, $60 router suffices for me. I see no needs for wifi6 yet for 99% of the households.
Edited to add: I'm starting to think that wireless communications were–on balance–a terrible mistake.
They have a certain amount of data they want to use in a month. If their speed goes up or down the amount of data they want to use stays about the same--the speed just determines how long they have to wait for downloads.
You might look into Comcast's XFi Advantage plan, which includes unlimited data. The cost varies by location but I think it is around $15/month in most. You also have to be renting their modem, so add another $10-14/month. You can add unlimited data to any plan, but that's $50/month. Doing it via XFi Advantage is about half of that.
This will change once the cable boxes go out and all media comes through the internet pipe. I don't know anyone in my network (mid 30s and below) who subscribes to television service. Even the old people in my family have been setup with iPads or Apple TVs or whatnot so they can watch everything on demand.
It would be awesome to have a simple table somewhere that described what to get based on WAN speed and OpenWRT support. Add some affiliate links and it could be a nice income stream.
FWIW, OpenWRT has a "recommended routers" page [0].
---
"15. How to improve speeds
Use Ethernet whenever possible"
>Greatly complicating a decision is that Wi-Fi 6E is just around the corner (early devices expected late 2020) that will require (yet again) new hardware -- existing Wi-Fi 6 devices will not support Wi-Fi 6E.
We could technically have WiFi 6 devices working with WiFI 6E as long as they were designed as such. I recall one of the ASUS engineers said their router would work, the problem is compliance and passing FCC certification and if Upgrade were allowed. Yes may be practically we need to buy new hardware. ( Profits )
And I could rant about WiFi 6/ 802.11ax for hours. Basically either buy a decent WiFi 5 or wait for 6E in hope they fix their mess.
MacBook Pro - As far as I know All 2016+ MacBook Pros has 802.11ac 3x3 WiFi by default. Giving you a "marketing" speed of 1300Mbps. Older Generation MacBook Pros has 3x3 for 15", 2x2 for MacBook Pro 13". All non-pros MacBook has always had 2x2 WiFi.
The reason is simply because only Broadcom make WiFI Client chipset that support 3x3. May be at the request of Apple. And the notebook market is basically an Intel only game. You get incentives and bundle pricing from Intel CPU for using Intel WiFi. And Intel doesn't make 3x3 WiFi ( Or they did but only in Centrino era... ) . Not in 802.11ac and likely not in 802.11ax.
Since demand for 3x3 802.11ac continues to be a niche, PC vendors fail to market better WiFi speed as added value. And Intel doesn't want to do a custom WiFi controller just for ~5M unit of MacBook Pro annually. And why innovate when there is no competition? Intel has 90%+ of Notebook PC market shares anyway.
In case anyone is wondering, there is an Broadcom 3x3 WiFi 6E client chipset announced last year. Hopefully you will see this in the next MacBook Pro Upgrade.
One reason people may have heard about the rumours of Broadcom selling their client WiFi business late last year. Their WiFi PC market is limited by Intel's domination, Smartphone market is a similar situation with Qualcomm. And Broadcom's Client WiFi business is primarily an Apple business. For nearly 4 years I have been anticipating Apple doing WiFi on their own before doing 4G/5G Modem. Considering they use more WiFi unit than Intel's current annual WiFi shipment. But so far only ultra low power chip W3 is being used in Apple Watch.
Samsung S20 is also an 2x2 Devices. May be worth an update in the WiFi 6 table.
>The AC#### naming ......
Those naming are confusing, but one way to look at it is the number represent the theoretical maximum capacity of the WiFi network. Those were never intended to be speed. But of course marketing is working is way. So hypothetically speaking a higher speed router gets you faster speed when you have multiple client.
Nice! Turns out all of my 802,11ax Rant is included in the its own Section.
>(1) 802.11ax is not even expected to be made an official IEEE standard until "Sep 2020" (source),
It should be noted the date is prediction. And that date has been moving up a quarter for every meeting since 2018. Draft 3.0 was suppose to be the last Draft, then Draft 5.0 and now Draft 6.0......
> 160Mhz
It should be noted none of the current Mobile Devices ( Or basically Non Intel Wifi 6 devices irrespective of their form factor ) supports 160Mhz ( 160Mhz / 80+80 ). Both Samsung and Apple are on 80Mhz only. This will likely change with WiFi 6E.
>Not all wifi clients are DFS capable!
Wow. I was always under the assumption DFS is a router problem. This sucks.
Such a great site that explains so much what is wrong about WiFi and its industry as a whole. And how it seems no cares about the user experience. There are increasing amount of people who just want to use a simple and working 4G/5G connection than their WiFi.
This is such a great resource I wish I could pay the authors some money as gratitude. It also reminds me of the Internet in the 90s where people spend their time, heart and soul into making something. It is not for SEO, or Ads Revenue. Simply wrote out of passion. ( or may be anger )
Thank You.
That was Broadcom as well.
>I guess it's because Broadcom doesn't have their WiFI 6 chipset
They did, as per iPhone. But Broadcom didn't have a WiFi 6 3x3 implementation then, and I guess Apple doesn't want to compromise to WiFi 6 2x2 ( 80Mhz ) implementation on 16" Macbook which would actually be slower ( On paper ) than its previous 3x3 WiFi 5 / 802.11ac.
All these is on the assumption Apple wants to continue with 3x3 WiFi. They could wait for WiFi 6E and call it a day. A 160Mhz on WiFi 6E gives you 2400Mbps on paper.