Dropping Internet Connection on WiFi

[pege63]

I live in a apartment complex and wath i did was i talked to all of my neighbors and build up a plan for better WIFI in our homes, and this is wath we did and everybody is happy

 

[pege63]

 

[sfx2000]

Howdy,
Your missing what is being said.

You should be using channel 1 or 6 or 11 for the 2.4Ghz band
Yes, use the 20Mhz wide only for better stability

Anytime you are splitting channels, you can get interference from both sides. What you want to do is use something like insider on a laptop, or Wifi Analyzer on your smart phone to only see what main channel (1,6,11) has the least amount of wifi usage and signal strength .

Remember, splitting channels is not good.

Concur - 1, 6, 11 with narrows channels is best for 2.4GHz - and camping in between not only impacts own WLAN, it impacts everyone nearby...

 

[FastLaneTech]

Thank you for the feedback. I have monitored the logs in the Comcast modem. I have the modem in bridge mode. Comcast residential modem IP is 10.0.0.1. To be on the safe side, I am paying a rental fee and I can change my modem whenever I want. I took it down to the Comcast store and they swapped it out for a better one. I found the issue with Hulu in the bedroom was a known issue. I just had to uninstall and reinstall the Hulu app on the FireStick. It's good now. The only thing I'm noticing is one laptop (the wife's) is having random network drops and my NVIDIA Shield Tablet will drop network connection too. Let me clarify, they both stay connected to the WiFi SSID and show excellent signal, but they look like they have packet loss. When the tablet starts dropping out, I'll pull up my app (Ping Test App) and test ping between tablet and router, tablet and modem, and tablet and www.google.com. I see a 33% packet loss from tablet to Google. No loss between tablet and router or modem. I'm thinking it could be a Comcast issue... but, to be fair, I did test the hard wire connection with www.pingtest.net and www.speedtest.net and speeds are fine, but you can see the "speedometer" stop for a sec or two as it is testing. Pingtest gives it a "B" rating as it cannot test my packet loss due to an error saying I don't have the latest Java installed. I do have the latest Java installed, bot 32 and 64 bit. The Comcast logs don't show any data being dropped or errors. I am able to ping test within the modem itself and it shows its fine. Seems like it's stable 75% of the time. I called Comcast for ideas. They did ask if I could see a cable running on my fence line across the neighborhood... I did. They said they had to rerun an underground feed and it could be due to the temporary line they are using as they have not made the physical connection to the newly run and buried line. They said they'd send someone out to finish the job as it shows an open and incomplete ticket. Maybe everything will be fine afterwards. The residential customer is in my neighborhood. The business customer I will have to go recheck all the connections and I will post any more symptoms or issues. Thank you all for the feedback.

 

[sfx2000]

-ASUS RT-N66R at residential location using 2.4G and 5G wifi band on channel 9 using current Merlin firmware over Comcast internet 100MB/10MB (Speedtest results are 13ms ping/126MB down/12MB up). Hardwire devices over Netgear gigabit network switch (unmanaged) 16 port. 22 devices connected in total.
-ASUS RT-AC66R at business location using 2.4G (5G disabled) wifi band on channel 6 using current Merlin firmware over Comcast internet 100MB/50MB (Speedtest results are 8ms ping/110MB down/51MB up). Hardwire devices over Netgear gigabit network switch (unmanaged) 24 port. 28 devices connected in total.

Lots of folks have been discussing WiFi concerns - the other thing is the number of clients on the router side of the house - and this might be part of your problem - wired or wireless...

Those networks are on the edge, IMHO.. many SOHO gateways can handle that many clients, but if pushed hard, they'll fall over - it's a memory limitation with the amount of RAM maintain NAT tables and the like..

 

[sfx2000]

Howdy,
Your missing what is being said.

You should be using channel 1 or 6 or 11 for the 2.4Ghz band
Yes, use the 20Mhz wide only for better stability

Anytime you are splitting channels, you can get interference from both sides. What you want to do is use something like insider on a laptop, or Wifi Analyzer on your smart phone to only see what main channel (1,6,11) has the least amount of wifi usage and signal strength .

Remember, splitting channels is not good.

Quick question - knowing the situation below - this is a screen scrape from WiFi Explorer on a Macbook Air 2014 - which is known to have hot antennas - which channel would you pick?

(hint, it's not obvious - Kismet/WiSpy investigation and general knowledge of 802.11n)

 

[RMerlin]

Quick question - knowing the situation below - this is a screen scrape from WiFi Explorer on a Macbook Air 2014 - which is known to have hot antennas - which channel would you pick?

(hint, it's not obvious - Kismet/WiSpy investigation and general knowledge of 802.11n)

View attachment 4555

I'd try channel 1 first, as all the overlapping channels are centered on channel 1. 6 and 11 would get interference from those two routers that sit on channel 10.

And that would still require testing, as if any of those other routers on channel 1 do a lot of transfers, that would be more problematic than channels 6 or 11 if the routers on either of them are quieter.

So as usual - you cannot pick the best channel just by looking at those charts. You have to test it, over a certain period of time.

 

[pege63]

Co-Channel interference is NOT a big problem until there are too much Wi‑Fi devices on the same channel so run the client that only supports 2.4GHz on one channel and those clients that support 5GHz on a different channel just for this. Then everything depends more on how close to a neighbor living, what kind of walls you have between the apartments and In the apartment, how many other wireless devices is around you all this make an effect on the WIFI signal. Everything is about try and error and communication between ppl. i can recommend this site for more infomation about this issue http://www.metageek.com/training/resources/why-channels-1-6-11.html and i am talking here in generaly not to a specific person

 

[FastLaneTech]

You bring up a good point sfx2000. Good question for a major network brainiac! lol... Merlin: Is there a practical limit of connected devices (wired/wireless) devices to these residential grade routers? I know theoretically they should support 255 devices (per IP address routes available) but is there a point where these routers should be replaced with a business class router?

 

[sitavata]

I also started having frequent wireless dropouts (connection stays established, but loses internet connectivity) after upgrading to 378.55 from .50
Downgraded to 378.54_2 and it seems to have fixed it.

 

[FastLaneTech]

Interesting sitavata.... Something Merlin can look at maybe? I might try the downgrade too. Which router are you using?

 

[sitavata]

Interesting sitavata.... Something Merlin can look at maybe? I might try the downgrade too. Which router are you using?

Sorry, wanted to add it in my thoughts - AC68U

 

[FastLaneTech]

Maybe test your theory by putting back on the latest firmware and see if it does it again? I could have been a flaky install of the firmware? Just my thoughts. Don't be sorry! Your input is welcome as it helps us all out to figure out the issues so we can all have better stability!

 

[lepa71]

hi all
I installed Merlin's .55 fw about 5 day ago to test the difference between R7000 and 68R. I'm experiencing some strange things. May be it's related to Comcast or may be not.

1. My public IP is completely different like R7000 will report 24.118.x.x and 68R will shows as 174.x.x.x something. Any ideas?
2. I'm getting some wireless clients disconnects while using 68R and I can't to find the pattern. I already tried to reset and it didn't help. Should I try the official fw from Asus or may be John's fork?

btw I'm not using comcast IPv6.

Thanks

 

[sfx2000]

I'd try channel 1 first, as all the overlapping channels are centered on channel 1. 6 and 11 would get interference from those two routers that sit on channel 10.

And that would still require testing, as if any of those other routers on channel 1 do a lot of transfers, that would be more problematic than channels 6 or 11 if the routers on either of them are quieter.

So as usual - you cannot pick the best channel just by looking at those charts. You have to test it, over a certain period of time.

It was a weird use case/example, and counter-intuitive at that - initially I put my travel AP on channel 1, but what I found was very poor performance - same with 6...

Breaking out the VMWare image, fired up Kismet and starting doing some scans - Ch 1 and 6 were very heavily loaded - not just from the hotel network (hhonors/attwifi), but also from across the street - and looking at the wireshark traces, the network access vectors were very high (again, suggesting high number of collisions and backoff's) - the hotel AP's are Cisco 11g models (as per the OUI and absence of HT stanzas) with QBSS reporting enabled - that showed that the channel utilization was upwards of 80 percent on both CH1 and Ch6.

Ch11, again, counter-intuitively, ended up being the best choice overall - even with the two AP's on Ch 10, and the Wide 6+10 AP (mpeshop). While attwifi on ch11 is pretty high, it was relatively unloaded, channel utilization there was about 7 percent, and this was confirmed by a relatively low NAV value - so the AP went there.

Perhaps an unusual case, but consider that this is a planned network in a hotel, one could see similar thing in Condo/Apartment dwellings, where there is little to no plan behind it.

Tools like inSSIDer and AcrylicWiFi show part of picture, but this one they couldn't really show the whole picture, and the tool itself was a bit misleading.

FWIW - QBSS, if present, is a great tool to determine overall channel usage, as this is collected by the AP and reported back to the client stations - ASUS AC1900 class routers all report this, at least the Broadcom based radios do (not too sure about the QTN radio in the AC87U)

 

[kvic]

FWIW - QBSS, if present, is a great tool to determine overall channel usage, as this is collected by the AP and reported back to the client stations - ASUS AC1900 class routers all report this, at least the Broadcom based radios do (not too sure about the QTN radio in the AC87U)

sfx, if channel set to "auto", will it pick the best channel in your hotel case? If not, why does it fail to do so? Just curious..

 

[sfx2000]

You bring up a good point sfx2000. Good question for a major network brainiac! lol... Merlin: Is there a practical limit of connected devices (wired/wireless) devices to these residential grade routers? I know theoretically they should support 255 devices (per IP address routes available) but is there a point where these routers should be replaced with a business class router?

I would say no more than 50, and that wired and wireless - the limit there is how many sessions can be maintain in the NAT tables - on the wireless side, I try to keep it below 25 clients per AP/Radio, as things will start getting very congested - I have seen a Router/AP take over 60 clients (Linksys WRT1900acv1) and survive during a family social event (most clients were mobiles/tablets), but later in the event it started behaving badly by not responding to DHCP requests, so a reboot was in order...

AP's generally do one of two things when maxed out - either denying new clients from attaching, or shedding existing clients that are associated (and they experience a dropout) - shedding is worse, because the client will immediately try to reattach, and this can turn into a storm of sorts and run off the end of the pier...

I've seen this with as little as 16 clients - this was on a common CableModem/Router/AP combo device from a well regarded vendor (Arris/Moto) - it would shed the load, not deny new clients, and it would get bad enough that one would have to power cycle the device.

 

[sfx2000]

sfx, if channel set to "auto", will it pick the best channel in your hotel case? If not, why does it fail to do so? Just curious..

That really depends on implementation - I'm not a big fan of auto, as I like to have a bit of control - that being said, I'm not sure what it would have done there in that use case...

 

[kvic]

That really depends on implementation - I'm not a big fan of auto, as I like to have a bit of control - that being said, I'm not sure what it would have done there in that use case...

I agree there shall be a choice of auto..or not to. But for majority of users opt for auto, they put their false hope on a spec/feature maybe you participated in drafting.

The common perception seems that human can always prevail in picking the right channel by a few steps a person go through (or ppl usually claim by trial and error/need to test in a specific environment..etc etc). If it's a few steps or tens or hundreds of steps.. they're good for putting into some sort of algorithm. Is it really that hard or just the industry is getting the inertia not to move. And maybe waiting for...Apple to think out of box and stunt the rest?

I know I maybe over simplifying the underlying difficulties..

 

[sfx2000]

I agree there shall be a choice of auto..or not to. But for majority of users opt for auto, they put their false hope on a spec/feature maybe you participated in drafting.

The common perception seems that human can always prevail in picking the right channel by a few steps a person go through (or ppl usually claim by trial and error/need to test in a specific environment..etc etc). If it's a few steps or tens or hundreds of steps.. they're good for putting into some sort of algorithm. Is it really that hard or just the industry is getting the inertia not to move. And maybe waiting for...Apple to think out of box and stunt the rest?

I know I maybe over simplifying the underlying difficulties..

Auto-channel selection isn't defined by any Standards spec, so it is implementation - I've seen some consumer grade AP's do a decent job of it, but many don't. Not going to name names, as it would not be appropriate...

One of the things with Auto selection is many AP's will do this one time only, on boot, but as we all know, over time, the environment can be dynamic - the ones that do a decent job leverage into Neighbor/Channel reports from clients that support it, along with OBSS scans (OBSS is a requirement for 11n and 11ac).

Going into Enterprise/Carrier grade - even there, it's a bit of a black box, and some do it better than others, but generally this is done at the WLC rather than the AP itself - and most admins probably would prefer not to let the AP's auto-tune themselves.

Apple does do some interesting things, on both a AP level (Airport Extreme/Express) and on a client level - and they've done a fair amount of work with Cisco in both standards oriented behavior along with Apple/Cisco proprietary stuff.