New Home Wifi Solution

[azazel1024]

I would say pro

Relatively inexpensive on a per AP basis
POE, so locating them would be relatively easy (not needing power near them)
seamless hand-off

Negatives

Total system cost might be more expensive once you account for POE, compared to client routers
Range is only okay from what I have seen. There are plenty of general consumer routers with better range, so 2 good consumer routers might cover your house, and you might need 3 Unifies to provide roughly the same coverage
Slower than 11ac gear for 5GHz deffinitely. Also not the fastest compared to good 11n/11ac gear on 2.4GHz either. Not slow...just not top tier

So if you need or really, really want seamless handoff, then they are probably your best cheapest option that will likely just work and they are easy to locate and setup.

They will not be as fast as "general consumer" good 11n gear or most 11ac gear and you might (might) need more of them for the same complete coverage at medium to high signal strengths (you might not too).

Me personally, I don't need seamless handoff. What exists with clients making the choice works well enough for my needs and I'd rather faster speeds, possibly less costs in the end and fewer devices to cover my area.

 

[bmather9]

I'm still undecided so I added a poll to see what you think.

 

[remixedcat]

No Meraki or other brands? I'm trying to get a hold of some other companies for you

 

[bmather9]

No Meraki or other brands? I'm trying to get a hold of some other companies for you

Maybe I didn't look hard enough, but from what I can find, Meraki hardware is minimum of $300 per AP while Unifi's are roughly $60 each if you buy the 3-pack. If I were going to spend that money, I'd be considering the $300 Unifi 802.11ac WAPs. If you find something othewise I'd be glad to consider them.

I know seamless roaming may seem trivial to most people who just deal with it or don't experience problems with it, but I'm baffled that there aren't more consumer based solutions for it out there. Or why it hasn't been baked into the 802.11 spec yet? I can certainly live without it, but I don't really want to. So in the end it comes down to speed (which I probably don't need) vs. seamless roaming (which I probably don't need). Really it's a win/win, but which is better?

 

[connorm]

where are you getting 3 unifi ap's for $60 each? do u mean the deal on amazon for 3 for $200?

 

[bmather9]

where are you getting 3 unifi ap's for $60 each? do u mean the deal on amazon for 3 for $200?

Yea that's what I meant, I guess doing more exact math hat really makes them $67 each.

 

[stevech]

Or why it hasn't been baked into the 802.11 spec yet?

good question. I'll speculate that those that need fast (sub-second) handoff due to user mobility are a few enterprise customers. Like Vocera and their customers with VoIP for health care and other afoot workers. the vast majority don't need it nor its expense.

 

[azazel1024]

good question. I'll speculate that those that need fast (sub-second) handoff due to user mobility are a few enterprise customers. Like Vocera and their customers with VoIP for health care and other afoot workers. the vast majority don't need it nor its expense.

True, but it would be nice if was included in the spec, even if it was just an optional. Multiple AP setups in homes are, at least I think, increasingly common. The true need isn't very strong, but it would "improve" things. Especially as telecoms start pushing their devices to both VOIP LTE as well as pushing towards offloading from their systems entirely to hybrid VOIP WIFI and LTE.

 

[bmather9]

True, but it would be nice if was included in the spec, even if it was just an optional. Multiple AP setups in homes are, at least I think, increasingly common. The true need isn't very strong, but it would "improve" things. Especially as telecoms start pushing their devices to both VOIP LTE as well as pushing towards offloading from their systems entirely to hybrid VOIP WIFI and LTE.

Yes, this was what I was getting at. It seems multi-AP home setups are becoming more common, and this should extend to other small business, hotels, etc. Even public wifi at international airports is sorely lacking seamless roaming in my experience. Just an overlooked detail to some, but quite an annoyance when my skype chat disconnects when walking from terminal to terminal. Its not like this problem just popped up yesterday; I remember using campus wifi in college back in 2004 and being annoyed that I couldn't roam between APs. A decade later and it's been given very little attention?

I've dealt with the normal handoff at home previously and had my galaxy s3 perform very poorly as a sticky client. Eventually drove me to the point to use one central router rather than 2 with better coverage because I was constantly finding myself connected to the router with low signal while sitting right next to the other one with strong signal. I tried special apps that were supposed to help the galaxy s3 connect to the stronger AP, but none of them really worked properly. I've had varied results with other devices like my laptop, and tablet, but in the end I found it much more annyoing to deal with 2 WAPs rather than one.

So all that being said, speed is also very important, which is why I consider possibly dealing with the roaming issues to get 802.11ac speeds, but do I really need those speeds? Hence my dillema of speed vs seamless roaming.

 

[azazel1024]

My limited experience, but Android seems rather problematic with Wifi roaming. Both of my kids newer android tablets like staying with whatever AP they are currently connected on. They WILL switch, but they like to sit there for a minute or three with very low signal before finally deciding to move over.

In comparison, my windows and iOS devices tend to switch in a matter of a few seconds when roaming between APs.

 

[stevech]

My limited experience, but Android seems rather problematic with Wifi roaming. Both of my kids newer android tablets like staying with whatever AP they are currently connected on. They WILL switch, but they like to sit there for a minute or three with very low signal before finally deciding to move over.

Many/most WiFi devices do this. My iPad is not good about changing APs. A WiFi client that frequently tries to find a better access device is ideal, but that costs battery life and may cause glitches if you are streaming - because the client device has to pause on the current channel long enough (0.1s) to detect other APs, or worse, has to scan channels. Ideally a client device would do this piecemeal, e.g., scan channels a few at a time to minimize disruption.
To avoid this client work, a controller based WiFi system has proprietary means for the controller to direct a handoff to the client.

Ideally, consumer advocates would pressure the IEEE 802 committee and manufacturers on that committee to get a viable and battery-practical method into the standard and adopted by WiFi AP/router makers. I think the standard lag because the explosion in smart phones got ahead of the standards.

 

[azazel1024]

You are probably right as to why.

I dunno though, my iPad 2 and my wife's iPad 2 never had/don't have any issues. They transition about as gracefully as our iPad 4s/5 do.

Signal strength drops below ~-70dB and they'll transition over to the other access point (which is usually around -50 to -55dB or so at that point). Its not as soon as I'd necessarily like, but it still seems to work pretty seamlessly (occasional drop/pause in facebook connections, but drops are rare). My windows clients with roaming agressiveness turned to max generally will transition right around -65dB, which is generally around when the other one is about -55dB or so.

Generally the windows devices won't transition till signal strength hits at least -60dB and the other one has to be at least 10dB higher in strength. iOS seems to go more for -70dB and at least 15-20dB stronger. At least in my house, physical distance wise, there is a small seperation from around -60dB and -70dB, same thing with signal strength being 10dB higher and being 20dB higher on the other AP. The android tabs seem to follow iOS in terms of when they are will to shift, but they'll still take a couple of minutes before they finally decide to actually make the transition. Only time they switch quickly is if I move them to a location where the signal from the other access point drops to -80dB or lower, which basically means taking one in to my basement office where the one router is sitting, then it'll be around -82dB and around -35dB respectively and they'll transition in maybe only 10-12s. Or reverse it and take one across the house and out to the garage where the basement router is down around -80-85dB and the familyroom AP is around -45 to 50dB.

 

[stevech]

Nit: the unit of measure of signal power is "dBm" whereas dB is mathematically dimensionless - and is used to express changes/differences like "3dB more".
dBm means power on a log scale (dB) relative to 0dBm being 0.001 watt.

All that aside, a decent WiFi speed should come from a signal more positive than about -70dBm. But at -70, if there's a decline of 10dB because the user reorients the antenna (which is inside the handheld), then -80 is not so good for speed. And "body block" (a watery human body in the path), costs 10dB or so. A few more dB in the 5.8GHz WiFi band.

It's hard to optimize, because you don't want the handheld frequently flopping back and forth between APs (thrashing).

Wireless is hard.

 

[azazel1024]

Certainly. I tend to use dB only because I've never been clear on what InSSIDer is reporting. I assume their measurements are actually dBm...but they ONLY list dB, so maybe they are using a different . That or I have false memories about using it.

I don't know exactly what iOS uses. I do know that most wireless drivers if you set it to maximum roaming agressiveness will start looking for stronger SSIDs at -60dBm of signal strength. They will only transition if the signal strength of a stronger SSID is at least 10dBm stronger signal and I believe they require the signal to be stronger for at least 5 seconds before they will release from the existing SSID and join the new SSID (or AP, since it can be from the same SSID).

So, a body block could cause it to shift APs...but in general, it won't just shift straight back, as unblocking would at most cause the access points to be at parity for signal strength or very close.

I've never actually seen an instance of thrashing, even when agressiveness was set to maximum. I have seen it where it occurs when entering and leaving a room.

However, you can also tweak the settings to make them a hair less agressive. I forget the exact Intel settings, but with Broadcom drivers, setting it to best speed starts looking at -60dBm, balanced is -70dBm and best range is -80dBm. Then agressiveness has 5 settings IIRC with the highest requiring only 10dBm of signal difference to switch and each notch I think is 3-5dBm. So minimum agressiveness requires a 24-30dBm difference and normal is around 16-20dBm of difference.

iOS seems to not look until around -70dBm and will transition if the other AP is around 15 or so dBm stronger.

Seamless handoffs really need to be a thing as well as basestation controlled. It doesn't seem like it would take that much to code it in to the basestation, especially if it is a standard and then you can just set a master and set all of the connected basestations to slaves, so there is only one controller deciding on handoffs. Shouldn't be that difficult to put a management page under the admin console to control it all.

Anyway, since that isn't likely to happen anytime soon, at least in my experience, in most cases, client based handoffs are generally okay, except VOIP, which can be problematic and occasionally video chat, which is less problematic. It pretty much is smooth and seamless otherwise. Cranking roaming agressiveness either to max, or even just a notch above regular, at least under windows, prevents sitcky clients and I don't really see any thrashing, nor reduced device battery life.

 

[stevech]

What's frustrating is WiFi user interfaces with only percentage. Not helpful. Signal strength is a log function (hence, the dB scale), due to the laws of physics related to the inverse square law.

Percent is certainly user friendly. I'd rather see a gauge or graphical bar where it can be on a log scale.

Due to human hearing, audio systems Vu meters and digital equivalents have long been on a non-linear scale too.

 

[azazel1024]

I won't argue with you there.

If/when I ever twiddle the settings I simply pull up InSSIDer to check signal strength. Likely followed by an SMB session to my server to see what if anything it did to my connection.

I've seen radio power settings of 25/50/75/100 as well as 10/25/50/75/100. The % rarely line up with anything resembling reality from what I've seen. I've seen some where the difference between 100% and 25% radio power meant ~4dBm, which isn't far from what it would be and others I've seen it where 25% actually meant about a 10dBm difference. I've also seen 10% power only be a 3dBm difference.

*sigh*

 

[CaptainSTX]

FYI:

The actual formula for calculating the increase in Dbs is:

Decibel = 10 log (power ratio )

As an example increasing the power from 5 watts - 10 watts


10 log (10/5 )

10 log (2) = 10 (.3010)

or approx 3 decibels

 

[azazel1024]

FYI:

The actual formula for calculating the increase in Dbs is:

Decibel = 10 log (power ratio )

As an example increasing the power from 5 watts - 10 watts


10 log (10/5 )

10 log (2) = 10 (.3010)

or approx 3 decibels

Are you refering to 3dB = 2x? Or my mention of things being all over the place with radio power almost always being listed as a percentage? Because on the later, I've found the manufacturers don't actually adhere to those percentage settings...with the actual changes being anything from about 15-50% per step instead of the actual listed amount (based upon InSSIDer reported logorithmic signal strength).

 

[stevech]

Signal strength as a percentage is just in consumer-land. In engineering, it's a log scale, dB, and power is in units of dBm.