Large home, roaming wireless setup

[azazel1024]

That is impressive for the netgear, but keep in mind, any range advertised is OPEN AIR range. A typical exterior wall attenuates between 9-20dBm depending on the construction (aluminum siding is the WORST. Uninsulated with 1/2" wallboard 2x4" construction with 11/32" plywood and vinyl siding is going to be the best at around 9dBm).

According to the inverse square rule, every doubling of range reduces signal strength by 6dBm (1/4th the signal strength). So a simple 9dBm reduction from a wall takes that 600ft down to about 200ft. A more typical wall would be in the 12-14dBm range, which means more like 120-140ft.

In general if you get usable signal more than 50-60ft from your house from an indoor access point, you should generally consider yourself lucky (especially if it has to go through an interior wall THEN the exterior wall and finally to you).

Its why I have an exterior AP...so I can actually connect 100+ft away with full bars (>-60dBm signal strength in most locations in my backyard except some of the far points >100ft away).

 

[Nonprofit]

That is impressive for the netgear

Yeah, I am starting to think I've been spoiled by this Netgear AP. I picked it up randomly at BestBuy... who knew it would be such a rock star?

Just to update this thread. I've decided to donate my UniFi APs to the local animal shelter and will be setting them up there this week. In addition to the shorter range, I was experiencing noticeably shakier performance too. I could be one floor directly above the AP and my iPhone would drop to half signal and start to crawl. Same if I was out on the deck, right on the other side of the wall from the AP. Loading website or doing casual browsing and email checking was noticeably slower. Normally I would chalk this up to the problems of wireless and walls. But I spent the weekend trying different positions for the UniFi APs and then switching back and forth to my old 'netgear setup' and measuring things. The Netgear setup, which is buried BEHIND my television, didn't have these issues. It beat the UniFi APs in every configuration and every position I could put them in.

I do think there is one place where I could put the UniFi AP to solve these problems, but alas, it is on the opposite side of the room, far from my only ethernet hookup. I tried a powerline setup to get it over there, but I've never been impressed with powerline, and this was no exception. It didn't really work.

So I'm donating them to a good cause.

In conclusion. I was very impressed with the UniFi controller and the Zero Handoff worked very well... mostly as advertised. But I was less than impressed with the range and power, for lack of a better word, of the APs themselves... maybe my house is unique, or my expectations were too high. I expected them to surpass - or at least match - my Lynksys/Netgear consumer setup. They didn't.

I was able to have some fun setting the controller up in Microsoft Azure and running it on L3 in a virtual machine there. This is how I plan on controlling them when I roll them out to my two offices, which have open floor plans and where range won't matter as much.

But for my home, for whatever reason, they just didn't perform. As much as I loved Zero Handoff and seamless roaming, the hit to range and performance ended up being too much for me.

Edit: Just to give you an example. As I type this I am about 15 feet away from one of the APs - unobstructed, clear line of sight - and my signal strength is -56dBm. I understand that that is a perfectly usable signal strength, but the Netgear measures -49 in the same situation. So you can imagine that as I get farther away and throw in a wall or two, the difference becomes more noticeable.

 

[clintb]

I give you "Minimum RSSI" for Ubiquiti!

http://community.ubnt.com/t5/UniFi-...niFi-Set-minimum-RSSI-for-clients/ta-p/522637

Disclaimer: I haven't used this, nor do I know if it works with Zero Handoff...just bookmarked for the future.

 

[Nonprofit]

Minimum RSSI won't work with Zero Handoff. But might be a good alternative.

Roaming is no longer my issue, however. At least not with these Unifi APs. Zero Handoff is working well with them. General range and signal strength are the problems I'm having with them. Someone suggested I play around with the TX Power settings - but I have to admit, my understanding of exactly how TX Power settings work, and the impact they might have on range or performance, is limited.

I suppose fiddling with different settings is worth a try before I donate them. I'll report back if I stumble upon anything.

 

[stevech]

That is impressive for the netgear, but keep in mind, any range advertised is OPEN AIR range. A typical exterior wall attenuates between 9-20dBm depending on the construction (aluminum siding is the WORST. Uninsulated with 1/2" wallboard 2x4" construction with 11/32" plywood and vinyl siding is going to be the best at around 9dBm).

According to the inverse square rule, every doubling of range reduces signal strength by 6dBm (1/4th the signal strength). So a simple 9dBm reduction from a wall takes that 600ft down to about 200ft. A more typical wall would be in the 12-14dBm range, which means more like 120-140ft.

In general if you get usable signal more than 50-60ft from your house from an indoor access point, you should generally consider yourself lucky (especially if it has to go through an interior wall THEN the exterior wall and finally to you).

Its why I have an exterior AP...so I can actually connect 100+ft away with full bars (>-60dBm signal strength in most locations in my backyard except some of the far points >100ft away).

A nit: RF attenuation and gain is measured in dB. The term dBm is absolute power.
Analogy: dBm is absolute, like your investments in dollars. dB is like your investments' gains and losses, and can be summed to alter your investments' current balance.

Lots of engineering units in RF propagation work...
dB is a relative term, like -10dB for losses or +6dB for gains
dBm is power, log scale of watts
dBi is for antenna gain relative to an ideal spherical radiation pattern
dBd is also for antenna gain but not normally used in 2.4GHz and up

User devices that simplify dBm to percent or "bars" are helpful, but are impossible to correlate to the non-linear (log) scales above.

So a 10dBm signal might be attenuated by 10dB due to distance and/or an obstacle in the path. This reduces the signal to 0dBm.
0dBm in power is 1 milliwatt. 10dBm is 10mW (log scale). 20dBm is 100mW which is 0.1 watt.
etc.

Example
Typical WiFi transmitted signal is around 30mW = 15dBm. (higher with the lowest bit rates).
Residential structure, indoor usage, typical attenuation due to distance + walls (no floors), is on the order of 70dB for 50-100 ft. with 2 or so drywall walls (2 layers) and antenna gains aof 2dBi (typical).

So we have +15dBm - 70dB = -55dBm received signal strength. Real world, it may differ by 10-20 dB. A receiver can achieve a certain data rate with a viable error rate for a signal strength of x. Say, a burst bit rate of 30Mbps with a 1% frame error rate might need a -60dBm received signal strength. Vendor to vendor, these required signal strengths (receiver sensitivity) don't vary much - as the laws of physics regarding noise prevail for all vendors!

This is a basic RF path link budget.

 

[azazel1024]

Yes, true.

Okay, replace my attenuation with dB and keep my RSSI as dBm.

For OP, you could always try upgrading to a modestly new 11ac router, especially if you are looking for a single router to cover your entire residence and can place it in a fairly central location.

It likely isn't going to boost things siginificantly over your Linksys, however in both SMB testings, as well as mine and several others, even using "legacy" 11n gear with an 11ac router sees pretty significantly boosts to 2.4GHz and 5GHz performance at similar distances as 11n routers.

So it might not really improve range much, but the range you do have should perform a fair amount better. This is true of all 11n gear I've tested with my 11ac router. I see 20-80% better performance on 2.4GHz and similar or better with 5GHz at medium to very long distances over a couple of different 11n router/APs I've tried with my 11n clients.

Obviously if it can't cover your residence and you need multiple, you are back in to "roaming" issues, but a newer 11n/11ac router/AP is likely to let you play with Tx settings, in which case you can try turning them down, which will hopefully help encourage your devices to roam sooner.

If you have multiple floors and one is enough to cover a single floor, you can also go with AP/routers with external antennas and then up the size a notch or two. This will compact coverage more on the floor it is on, so once you step up to the next floor, your device is much more likely to connect to the AP/router on that floor. In addition it is a bonus as it'll increase throughput at medium to long range on the existing floor for the router (just don't expect miracles, but going up 2dBi on atenna gain I see between 5-20% increase in throughput at medium to very long range on 5GHz and a bit less on 2.4GHz by moving from 3 to 5dBi on one router and from 5 to 7dBi on another router. The gain just might not be enough to actually increase range inside really).

 

[azazel1024]

A nit: RF attenuation and gain is measured in dB. The term dBm is absolute power.
Analogy: dBm is absolute, like your investments in dollars. dB is like your investments' gains and losses, and can be summed to alter your investments' current balance.

Lots of engineering units in RF propagation work...
dB is a relative term, like -10dB for losses or +6dB for gains
dBm is power, log scale of watts
dBi is for antenna gain relative to an ideal spherical radiation pattern
dBd is also for antenna gain but not normally used in 2.4GHz and up

User devices that simplify dBm to percent or "bars" are helpful, but are impossible to correlate to the non-linear (log) scales above.

So a 10dBm signal might be attenuated by 10dB due to distance and/or an obstacle in the path. This reduces the signal to 0dBm.
0dBm in power is 1 milliwatt. 10dBm is 10mW (log scale). 20dBm is 100mW which is 0.1 watt.
etc.

Example
Typical WiFi transmitted signal is around 30mW = 15dBm. (higher with the lowest bit rates).
Residential structure, indoor usage, typical attenuation due to distance + walls (no floors), is on the order of 70dB for 50-100 ft. with 2 or so drywall walls (2 layers) and antenna gains aof 2dBi (typical).

So we have +15dBm - 70dB = -55dBm received signal strength. Real world, it may differ by 10-20 dB. A receiver can achieve a certain data rate with a viable error rate for a signal strength of x. Say, a burst bit rate of 30Mbps with a 1% frame error rate might need a -60dBm received signal strength. Vendor to vendor, these required signal strengths (receiver sensitivity) don't vary much - as the laws of physics regarding noise prevail for all vendors!

This is a basic RF path link budget.

As a note on the later, very true. Looking at published numbers at least, the largest variance I have seen, at least in 11n routers, is about 4dBm for similar maximum link rates (out of my butt comparison, "worst" I see is maybe something like -72dBm for 54Mbps link rates on a 300Mbps max 11n router and the "best" I have seen is around -76dBm for 54Mbps link rates on a 300Mbps max 11n router. Published numbers (and fudging, I don't remember the exact signal strengths), real world...haven't tested enough different ones to make a firm comparison, though 11ac routers certainly seem to be able to maintain higher MCS levels at similar signal strengths, by at least 1 if not 2 notches).

Path loss is also great indoors, as you mentioned, but sometimes even a tiny increase in signal strength is enough to get meaningful increases in performance, if not much in the way of range. Solely based on my testing, I've found (as previous post mentioned) just a mild 2dBi increase in antenna gain is generally enough to give a noticable, if minor/modest (depending on your point of view) increase in throughput. 5-20% certainly isn't the world, but its better than zero gain and costs little to replace an antenna or three...and jumping a couple of sizes for something like a 4dBi gain is likely even more noticable gain. 4dB higher signal strength often isn't a lot, your body attenuates more if you put yourself between your wireless client and the basestation, but if you are going to do that anyway, you might as well have higher antenna gain. Probably won't mean getting a usable signal to that home office where you couldn't connect at all, but it might mean your entertainment room where you are streaming to your chromecast, you can get 40Mbps of usable bandwidth instead of 30Mbps.

Bonus points if you can replace the antennas on both ends of the link.

 

[rtfm]

Make sure you read up on the Unifi Zero handoff over at the Ubiquiti Forums. There are some catches to it. One catch is all the AP's must be on the same channel, which is not too bad if you have just a couple of devices using wireless.

Argh... I was thinking you'd have adjacent floors on different channels as I do now. That way they can't interfere.

Looks like UniFi ZH is worth avoiding; not sure that /dev/nulls their candidacy....

 

[stevech]

Handoff in WiFi - the client device has to have special software in order to react to a command from an access point to move to some other access point. This is called a directed handoff in cellular systems.
Cisco had/has this in their Cisco Certified Extensions - going back many years. I think they gave up as it was too messy to have a special app on so many types of clients.

A scheme to do this without a special app on the client is for APs to collaborate. Let's say WiFi client A is associated to AP #1. AP#2 and AP#1 exchange back-channel (LAN) messages and decide (e.g., due to signal strength) that client A should be pushed from AP#1 to AP#2. Alas, no software on the client to react to a command to do this. So AP#1 can forcibly de-associate client A. Now Client A "SHOULD" search for a new/better AP. But that depends on how the client is designed. Let's say that searching caused A to choose AP#2, as was the goal.

This is a flaky way to do handoffs as it relies on the client to "do the right thing". And most would search on the same SSID/channel for a better AP before searching other channels and SSIDs for which the WPA key is known.

This is all due to what clients and APs do about IEEE 802.11 which has no real roaming management. (Roaming is intrinsic to cellular systems' standards). There is an IEEE 802.11s standard to help with roaming, but few implement it. WiFi was conceived as a low-mobility service and largely, that's still true.

 

[azazel1024]

Apple actually has it implemented in OSX and iOS (802.11r and 802.11k). I think in the windows environment it is up to the specific wifi drivers to implement, I don't think it is an OS thing there.

On Android...hahahhahhahahaha. Good luck.

My hope is that IEE at some point mandates 802.11r and k to comply with a future standard that is "speed" related. Example if/when 802.11ax comes along, that 802.11r and k are mandated as part of the feature set for 802.11ax.

So anyone implementing an 802.11ax chipset is also going to have to incoporated all of the BSS stuff for fast transitions. Especially with 5GHz, fast and seamless roaming are rather important in Wifi. Chipset makers just don't care a whole lot.

 

[sfx2000]

Apple actually has it implemented in OSX and iOS (802.11r and 802.11k). I think in the windows environment it is up to the specific wifi drivers to implement, I don't think it is an OS thing there.

On Android...hahahhahhahahaha. Good luck.

My hope is that IEE at some point mandates 802.11r and k to comply with a future standard that is "speed" related. Example if/when 802.11ax comes along, that 802.11r and k are mandated as part of the feature set for 802.11ax.

So anyone implementing an 802.11ax chipset is also going to have to incoporated all of the BSS stuff for fast transitions. Especially with 5GHz, fast and seamless roaming are rather important in Wifi. Chipset makers just don't care a whole lot.

One of the reasons why I like keeping all AP's on the same SSID/credentials, as it's layer 2, not layer 3, as such, the transition is much better as one keeps the same IP layer connection context...

FWIW - Apple Airports in a "roaming" configuration actually do a half-way decent job of handover - it's not AP directed, but devices transition across them pretty seamlessly...

 

[stevech]

The issue isn't IEEE... it's WiFi. The internatinal WiFi industrial alliance is chartered, and receives dues to, test and certify products comply with both IEEE standards and also meet the signal quality (transmitted) and receiver frame error rate vs. signal to noise ratio - that WiFi chose.

IEEE doesn't have, is not chartered to test/certify/police, etc. They just publish standards.

The US FCC isn't concerned with IEEE or WiFi - FCC just enforces their Part 15 compliance testing to show radiated emissions (and conducted to power line) are within FCC regulations esp. at band-edges. A product might transmit a badly distorted waveform and that's not an FCC issue. Some FCC peers in other countries follow suit.

What was supposed to happen is that manufacturers could NOT use the WiFi logo/trademark unless their product met the WiFi's criteria. I think this process has turned into the wild west/chaos.

 

[azazel1024]

One of the reasons why I like keeping all AP's on the same SSID/credentials, as it's layer 2, not layer 3, as such, the transition is much better as one keeps the same IP layer connection context...

FWIW - Apple Airports in a "roaming" configuration actually do a half-way decent job of handover - it's not AP directed, but devices transition across them pretty seamlessly...

Facetime doesn't roam all that well. That said, since changing over my AP to a different one, I do some to be getting some actual facetime roaming ability. Previously anytime I'd move between my AP and my router, my phone would switch which it was connected to and my facetime session would disconnect.

Now with the different AP, I get a lot of slow down in my facetime session for 10-20s (a couple of pauses sometimes, 3-4FPS), but then it speeds up. Since it doesn't show the wifi signal strength bars, I can't see if it is roaming and I haven't done "intensive" tests by monitoring my router/AP's management pages to see when the iPhone's MAC shows up in the connected list.

One of the things that might be the "cause" for this is my iPhone LOVEs connecting on 5GHz (why not, it is 150Mbps on 5GHz and only 65Mbps on 2.4GHz). So previously with my "old" AP, it could be that wandering between the two dropped the 5GHz signal strength "too low", between that and the second "disconnect" as it is changing base stations might have been enough to disconnect the session. Now with the new AP, the 5GHz bandwidth might be strong enough between locations that facetime doesn't disconnect.

Not sure really, but I do know previously 9+ out of 10 times I'd lose a facetime session if I or my wife wandered from the basement upstairs or vice versa. A couple of weeks ago she was gone with the kids for the entire week and several times we facetimed and I wandered upstairs down or vice versa and the worst that happened is the facetime session got choppy for a few seconds before it resumed just fine.

That said, all other wifi related tasks I've tried worked just fine with the old AP (file transfers, streaming audio and video, etc. Though I have noticed that airplay doesn't like it if I wander back and forth between the router and AP. It will transition seamlessly up to 2 times in a short period. It won't do it 3 times in less than about 5 minutes. It'll drop the audio and I have to restart it. I can do it a dozen times over the course of a couple of hours no worries, but "too many" times in just a few minutes and the audio streaming stops (I found this out while doing laundry with my phone in my pocket streaming audio to my speakers)).

 

[sfx2000]

Agree with you on Facetime, it's not very tolerant of interruptions - Skype handles it much better in my experience...

On a slightly different front, I also use CounterPath's Bria client as a SIP Phone, and it performs quite well in the same environment...

 

[azazel1024]

Yeah, I have zero problems with Skype and roaming ever. Just that much more annoying that Apple engineers haven't "fixed" this at some point (since I can't imagine that the situation never comes up or that they are not aware of it).

 

[coxhaus]

We just cleaned up the yard and I mowed and edged. So my wife decided to have a drink and facetime with best friend in Houston setting outside on our picnic table. The weather is perfect. I have the new WAP321 on 5GHz in the back yard window pointing toward the picnic table and ready to test. I put it in last night. It seems to work well, but what was interesting was my wife decided she needed a second drink so she brought her iPad in with her as she was talking. The facetime program only had a small moment where it said reconnecting as it switched to another WAP321 and her conversation continued. No calls were dropped.
Oh, it also did the exact same thing on the way back outside. I guess it is not totally seamless but it is the best I have seen at my house. I think there was about 2 seconds of delay.
What happens is you see reconnecting in the top left hand corner but about as fast as you see it, it disappears as the iPad has switched to a different WAP321.

 

[azazel1024]

Nice. I'll have to test again soon. I know in iOS8.3 they released a whole shed load of kernal and network stack changes/fixes. That and I think in 8.1 and 8.2 they did a number as well. It could be that Apple has actually made some changes under the hood of facetime/network stack that has "fixed" the issue of transitioning access points. I haven't facetimed (and wander between APs) in at least a month, maybe 2-3.

 

[sfx2000]

Nice. I'll have to test again soon. I know in iOS8.3 they released a whole shed load of kernal and network stack changes/fixes. That and I think in 8.1 and 8.2 they did a number as well. It could be that Apple has actually made some changes under the hood of facetime/network stack that has "fixed" the issue of transitioning access points. I haven't facetimed (and wander between APs) in at least a month, maybe 2-3.

iOS 8.3 has been pretty solid compared to earlier 8.x releases, probably not as good as iOS 7, but still better than it's been for some time..

I think a lot of the apple issues had to deal with the awdl implementation (AirDrop and Continuity/Handoff) along with the new discovery daemons - 10.10 on the Mac had similar issues, and 10.10.3, along with 8.3 on the iDevices, have made quite a bit of improvement in WiFi and connection stability.

 

[coxhaus]

I guess IOS 8.3 could be responsible for facetime not dropping the call but I sure like these Cisco WAP321 wireless waps on 5 GHz. I have good coverage and about the same speed everywhere in my house and outside in my setting area. 40MHz may not be 80 MHz but I think just having one wireless device you will end up with uneven coverage in your house. I did when I had one wireless device.

 

[remixedcat]

Xirrus still is god tier wifi. The roaming is solid and smooth. So is Cisco Meraki.